An electronic system may obtain data corresponding to its current surroundings using any number of sensors, and may attempt to make use of the data in order to perform one or more operations. However, making sense of sensor data is a challenge, especially when objects in the system's surroundings change state or position relative to the system. In addition, data from certain sensors may be more relevant than data from other sensors, and determining which sensors to rely on in any instance or iteration to determine state or position changes presents additional challenges. As an example, vehicle or other machine perception systems make predictions about future state or position changes in order to anticipate potential obstacles. In some instances, these systems employ association functions to indicate associations between measurements of certain sensors and those predictions. However, designing such conventional association functions is a manual and difficult process.
Embodiments of the present disclosure relate to determining associations between sensor data and predicted states in multi-sensor systems. Systems and methods are disclosed that employ machine learning models—such as deep neural networks (DNNs)—to predict association costs between sensor measurements and predicted states in recursive frameworks for multi-sensor fusion software of autonomous or semi-autonomous vehicle or machine applications.
In contrast to conventional systems, such as those described above, systems and methods in accordance with the present disclosure do not require manual iterations in determining distance metrics and their corresponding weights for establishing association functions of different sensors. The systems and methods can provide a data-driven approach to determining a suitable set of parameters that may be automatically updated when there are changes in dependencies. The systems and methods can develop more complicated distance metrics than could be generated in previous attempted solutions. In non-limiting examples, neural network models can learn (or be updated to form) complicated functions to perform the task of association in a multi-sensor environment, and the neural network model can generate complex nonlinear functions that find correlations between varied input parameters.
At least one aspect relates to a processor. In various embodiments, the processor can comprise, or can be, one or more circuits to determine a predicted state of an object detected in an environment. In various embodiments, the one or more circuits can generate, using one or more neural network models and based at least on sensor data generated using a plurality of sensors and one or more values of one or more, input parameters corresponding to the predicted state, a score indicative of an association between the predicted state of the object and one or more representations of the object in the sensor data. In various embodiments, the neural network model can be generated using modeling data. In various embodiments, the modeling data can be, or can comprise, (i) a plurality of predicted states of objects and/or (ii) a plurality of sensor data each corresponding to a respective one of the plurality of predicted states and obtained from a respective plurality of sensors. In various embodiments, the one or more circuits can update data corresponding to the environment according to whether the score exceeds an association threshold.
In various embodiments, the one or more circuits can receive, from the plurality of sensors or from another device or system, the sensor data generated using the plurality of sensors. In various embodiments, the one or more circuits can generate the sensor data using output from the plurality of sensors.
In various embodiments, the one or more neural network models can be or can comprise a multi-layer perceptron (MLP) model. In various embodiments, the one or more input parameters comprise at least one of: (i) the predicted state or (ii) an identification of a sensor. The plurality of sensors can be part of a system. In various embodiments, the one or more circuits can generate an instruction to cause a change in the system in response/reaction to the updated data corresponding to the environment. In various embodiments, the data is updated to indicate a change in a position of the object in the environment relative to the system. In various embodiments, the data is updated to indicate a change in a trajectory of the object in the environment relative to the system. In various embodiments, the data is updated to indicate one or more changes in a position and/or a trajectory of the object.
In various embodiments, the one or more circuits can normalize output from the respective plurality of sensors to obtain the sensor data. In various embodiments, the one or more neural network models are generated, at least, using modeling data comprising (i) a plurality of predicted states of objects, and (ii) a plurality of sensor data, each corresponding to a respective one of the plurality of predicted state and obtained from a respective plurality of sensors. The respective plurality of sensors can be or can comprise sensors of a plurality of sensor types. In various embodiments, a first portion of the modeling data can correspond to positive samples and a second portion of the modeling data can correspond to negative samples.
In various embodiments, the data corresponds to one or more object tracks corresponding to one or more objects in the environment. In various embodiments, the data is updated to include an updated object track corresponding to the object based at least on the score. In various embodiments, the one or more neural network models can be generated, at least, by generating the modeling data, and updating one or more parameters of the one or more neural network models using the modeling data. In various embodiments, the modeling data can be, or can comprise, (i) the plurality of predicted states of objects, and (ii) the plurality of sensor data. In various embodiments, the neural network model can be generated by updating the neural network model using the modeling data. In various embodiments, during training, the one or more neural network models can be updated to receive input which can be, or can comprise, (a) a first predicted state of a first object and (b) an identification of a first sensor. In various embodiments, the neural network model can be updated to provide an output score indicative of an association between one or more first object representations corresponding to sensor data from the first sensor and the first predicted state.
At least one aspect relates to a method. In various embodiments, the method can comprise receiving, from a plurality of sensors, sensor data corresponding to an environment. In various embodiments, the method can comprise generating, using the sensor data, a predicted state of an object corresponding to a time. The method can include determining a predicted state of an object corresponding to a time. In various embodiments, the method can comprise generating, based at least on one or more neural network models processing multi-sensor sensor data and one or more values of one or more input parameters corresponding to the predicted state, a score indicative of an association between the predicted state of the object and a detected state of the object at the time. In various embodiments, the neural network model can be generated using modeling data. In various embodiments, the modeling data can be, or can comprise, (i) a plurality of predicted states of objects, and (ii) a plurality of sensor data each corresponding to a respective one of the plurality of predicted states and obtained from a respective plurality of sensors. In various embodiments, the method can comprise updating an object track corresponding to the object based at least on the score.
In various embodiments, the one or more neural network models can be or can comprise a multi-layer perceptron (MLP) model. In various embodiments, the MLP model can be generated using data that is, or that comprises, (i) a plurality of predicted states of objects, and (ii) a plurality of sensor data each corresponding to a respective one of the plurality of predicted state s and obtained from a respective plurality of sensors. In various embodiments, the one or more input parameters can be, or can comprise, at least one of (i) the predicted state and (ii) an identification of a sensor. In various embodiments, the multi-sensor data is generated using a plurality of sensors of a system. In various embodiments, the method further comprises generating an instruction to cause a change in the system based at least on the updated object track. In various embodiments, the multi-sensor data is generated using a plurality of sensors of a system. In various embodiments, updating the object track indicates a change in a position of the object in the environment relative to the system. In various embodiments, updating the object track indicates a change in a trajectory of the object in the environment relative to the system. In various embodiments, the data is updated to indicate one or more changes in a position and/or a trajectory of the object.
In various embodiments, the method can comprise generating the one or more neural network models. In various embodiments, generating the one or more neural network models can be, or can comprise, generating modeling data, and updating the one or more neural network models using the modeling data to receive input and provide an output score. In various embodiments, the input can be, or can comprise, (a) a first predicted state of a first object and (b) an identification of a first sensor. In various embodiments, the output score can be indicative of an association between a first detected state of the object and the first predicted state. In various embodiments, the modeling data can be, or can comprise, (i) the plurality of predicted states of objects, and (ii) the plurality of sensor data.
At least another aspect relates to a processor. The processor can be or can comprise one more circuits to perform one or more operations based at least on an object track corresponding to an object at a time. The object track can be determined using one or more association scores computed using one or more neural network models based at least on the one or more neural network models processing multi-sensor sensor data and data representative of one or more values of one or more parameters corresponding to a predicted state of the object at the time. In various embodiments, the modeling data can be or can comprise (i) a plurality of predicted states of objects, and (ii) a plurality of sensor data. In various embodiments, each sensor data of the plurality of sensor data can correspond to a respective one of the plurality of predicted states. In various embodiments, each sensor data of the plurality of sensor data can be obtained from a respective plurality of sensors. In various embodiments, the one or more circuits can update, using the modeling data, a neural network model to receive input comprising (a) a first predicted state of an object and (b) an identification of a first sensor. In various embodiments, the one or more circuits can update the neural network model to generate a score indicative of an association between sensor data of the first sensor and the first predicted state.
In various embodiments, at each iteration of the one or more neural network models, the one or more neural network models output an association score between each detected object in the multi-sensor sensor data and the object. The one or more circuits can receive, from a plurality of sensors, sensor data corresponding to an environment. In various embodiments, the one or more circuits can generate, using the sensor data, a predicted state for an object detected in the environment. In various embodiments, the one or more circuits can provide, to the neural network model, (a) input parameters corresponding to the predicted state and (b) the sensor data. In various embodiments, the input parameters and the sensor data can be provided to the neural network model to obtain a score indicative of an association between the predicted state and the sensor data. In various embodiments, the one or more circuits can update data on the environment according to whether the score exceeds an association threshold.
At least another aspect relates to a processor. The processor can be, or can comprise, one more circuits to implement a neural network model. In various embodiments, the neural network model can be obtained by generating modeling data that is or that comprises (i) a plurality of predicted states of objects, and (ii) a plurality of sensor data, each sensor data of the plurality of sensor data corresponding to a respective one of the plurality of predicted states and being obtained from a respective plurality of sensors. In various embodiments, the neural network model can be obtained by updating, using the modeling data, the neural network model to receive inputs that are or that comprise (a) a first predicted state of an object and (b) an identification of a first sensor, and to generate a score indicative of an association between sensor data of the first sensor and the first predicted state.
In various embodiments, the one or more circuits are part of a system. In various embodiments, the one or more circuits can receive, from a plurality of sensors, sensor data corresponding to an environment of the system. In various embodiments, the one or more circuits can generate, using the sensor data, a predicted state for an object detected in the environment. In various embodiments, the one or more circuits can provide, to the neural network model, (a) input parameters corresponding to the predicted state, and (b) the sensor data, to obtain a score indicative of an association between the predicted state and the sensor data. In various embodiments, the one or more circuits can update data on the environment according to whether the score exceeds an association threshold. In various embodiments, the one or more circuits can generate an instruction to cause a change in the system in reaction/response to the updated data on the environment. In various embodiments, the neural network model can be or can comprise a multi-layer perceptron (MLP) model.
At least another aspect relates to a method. The method can comprise implementing a neural network model. In various embodiments, the neural network model can be obtained by generating modeling data that is or that comprises (i) a plurality of predicted states of objects, and (ii) a plurality of sensor data, each sensor data of the plurality of sensor data corresponding to a respective one of the plurality of predicted states and being obtained from a respective plurality of sensors. In various embodiments, the neural network model is obtained by updating, using the modeling data, the neural network model to receive inputs comprising (a) a first predicted state of an object and (b) an identification of a first sensor, and to generate a score indicative of an association between sensor data of the first sensor and the first predicted state.
In various embodiments, the method can comprise receiving, from a plurality of sensors, sensor data corresponding to an environment of a system. In various embodiments, the method can comprise generating, using the sensor data, a predicted state for an object detected in the environment. In various embodiments, the method can comprise providing, to the neural network model, (a) input parameters corresponding to the predicted state, and (b) the sensor data, to obtain a score indicative of an association between the predicted state and the sensor data. In various embodiments, the method can comprise updating data on the environment according to whether the score exceeds an association threshold. In various embodiments, the method can comprise generating an instruction to cause a change in the system in reaction/response to the updated data on the environment. In various embodiments, the neural network model can be or can comprise a multi-layer perceptron (MLP) model.
In various embodiments, the processors, systems, and/or methods described herein can be implemented by, or can be included in, at least one of a control system for an autonomous or semi-autonomous machine; a perception system for an autonomous or semi-autonomous machine; a system for performing simulation operations; a system for performing digital twin operations; a system for performing light transport simulation; a system for performing collaborative content creation for 3D assets; a system for performing deep learning operations; a system implemented using an edge device; a system for generating or presenting at least one of virtual reality, augmented reality, or mixed reality content; a system implemented using a robot; a system for performing conversational AI operations; a system for generating synthetic data; a system incorporating one or more virtual machines (VMs); a system implemented at least partially in a data center; or a system implemented at least partially using cloud computing resources.
The present systems and methods for determining associations between sensor data and predicted states in autonomous and semi-autonomous systems and applications are described in detail below with reference to the attached drawing figures, wherein:
Systems and methods are disclosed related to determining associations between sensor data and predicted states in multi-sensor systems. Although the present disclosure may be described with respect to an example autonomous or semi-autonomous vehicle or machine 100 (alternatively referred to herein as “vehicle 100” or “ego-machine 100,” an example of which is described with respect to at least one of
Autonomous or semi-autonomous navigation perception systems, for example, can use multiple sensors to understand the environment of the system. These systems can combine information from multiple sensors (via “fusion” of a variety of sensor data) to build a robust system that can identify objects in the system's surroundings. Such objects may be obstacles that may get too close to the system (or even collide with the system) depending on what actions (e.g., changes in motion and position) are taken by the system and/or the object. The obstacles detected by different sensors are tracked by such a system by taking in information from the different sensors. In such modules, for each obstacle newly detected by any sensor, a new track may be created to store the information corresponding to that obstacle, from its inception (following the object's initial detection) until the obstacle is no longer “visible” (or otherwise detectable), and/or until the object is deemed to be irrelevant or otherwise not significant enough to be a concern to the navigation system. The track can be used to predict the next state of the corresponding obstacle by leveraging information about the obstacle's history, so that suitable next steps can be determined (e.g., changing trajectory or speed), if any.
For each new time instance in which the perception system identifies or receives an indication of a detected obstacle, the perception system may also generate predicted states for that time instance for all other existing tracks (e.g., tracks corresponding to objects that are still visible and relevant). These predicted states and measurements are then compared against each other, and an association cost is computed for each pair. Association cost can be a metric which measures/indicates how similar the compared predicted state and the measurement are. This metric can be referred to as a “distance” metric, where distance here does not refer to physical separation, but rather similarity (e.g., greater distances are correlated with less similarity, and smaller distances are correlated with more similarity).
Association cost may itself be defined by a weighted combination of multiple different metrics, where certain “weights” are assigned to each computed metric. The weights can indicate the “importance” or relevance of that corresponding metric to the association cost (e.g., how much of an effect the metric has on the association cost), such that the greater the weight of a metric, the more relevant that metric is to the association cost. These weights may be obtained by performing evaluations on datasets and identifying how much influence the different computed metrics have on the association cost, which in turn affects the overall performance of the navigation perception system. The computed association costs for all predicted states and measurement pairs can then be used to “greedily” associate different predicted states to different measurements, and these associations can then be used to update the state of the track for the corresponding obstacle. As used herein, a “greedy” association is one that does not necessarily yield the most optimal association value (which would exhaustively consider all potential solutions), but rather involves a non-exhaustive determination that is deemed to be the best or otherwise suitable at that moment based on available data.
Designing an apt association function which associates only the best predicted states and measurement pairs involves choosing parameters such as the appropriate distance metrics and the weights for these metrics. Determining all these parameters can be extremely challenging and time consuming, as these would involve manually choosing certain metrics and corresponding weights for the metrics, and then determining the effectiveness of the choice by obtaining certain key performance indicator (KPI) values, and then using the KPI values through multiple iterations (making adjustments as needed for each subsequent iteration) until the best combination of distance metrics and their weights is determined. This is not only cumbersome and time consuming, but can be very susceptible to change, as the optimal combination of metrics and corresponding weights could change whenever certain parts of the upstream algorithms are changed or whenever new data is used.
Addressing the above shortcomings, various embodiments of systems and methods disclosed herein can employ a more automated, data-driven approach, instead of relying on manually-designed association functions. The problem of designing an optimal association function may be solved by using neural network models (or other machine learning models) trained using data generated specifically for this task. As a non-limiting example, the approach may employ a Multi-Layer Perceptron (MLP) model trained using data generated for this task. In other examples, the approach may employ other deep neural network (DNN) architectures.
The neural network model (e.g., MLP model or other model) can be used to compute the association confidence for a given pair of predicted states and measurements. The neural network model may be trained to compute an association score (e.g., from zero (0) to one (1), or another defined range) for a pair of predicted state and measurements, with 1 (or value above a threshold, such as 0.5 or greater) indicating that the state-measurement pair are to be associated, and 0 (or value below the threshold, such as less than 0.5) indicating otherwise.
Embodiments of the disclosed approach can provide many advantages. For example, training a neural network model (such as an MLP model) to determine association score (e.g., a score between 0 and 1 indicative of confidence in associating the given predicted state and measurement, with 1 indicating the highest confidence and 0 indicating no confidence for association of the given pair) reduces or eliminates the need for manually-designed association functions. There is no need to choose the best distance metrics and their corresponding weights to design an optimal association function. The neural network model can be fed existing parameters of the predicted state and the measurement, and upon training with the data, the neural network model can determine what weights to assign to each of the input parameters of the predicted state and the measurement. This approach can also reduce or eliminate the need for generating distance metrics, and assigning weights to the distance metrics. In various embodiments, the neural network model can form/develop/design its own distance metrics via training, and can assign weights to the distance metrics in the form of the weights of the artificial “neurons” of the neural network model.
Another significant advantage of various embodiments is that, if there are significant changes in the data or the upstream algorithms, the neural network model may be updated with the new data to obtain an association function suited to the new data. There is no need to manually choose weights again. This can reduce a significant amount of developer time and can also provide a more robust solution. The neural network model can be regularly improved/updated as more data becomes available. A larger neural network model (e.g., one that employs a greater number of neurons, or a larger number of “layers,” each layer including multiple neurons) can be used, if needed, to allow the neural network model to generate more complex association functions (e.g., corresponding to various sensor data from a greater number of sensors).
Another advantage of various embodiments of the systems and methods is that a single neural network model (e.g., a single MLP model), for example, can handle data from all sensors. Rather than employing different association functions for different sensors or different kinds of sensors (e.g., one association function for a camera, and another association function for a radar-based sensor), in the disclosed approach, a neural network model can be designed such that relevant sensor data can be associated/paired with predicted states of the obstacle tracks regardless of which sensors or sensor types provided the sensor data. In certain embodiments, all that may be needed is identifying which sensor the measurement data is from, and using this information, the neural network model can compute the association score for the given pair of sensor data and predicted state. This can reduce or eliminate the need for designing and maintaining multiple association functions and hence involves less development time.
However, in embodiments, more than one neural network model may be used to compute the associations without departing from the scope of the present disclosure—for example, individual models may be used for a single sensor modality, individual models may be used for each sensor, and/or the like.
With reference to
The systems and methods described herein may be used by, without limitation, non-autonomous vehicles or machines, semi-autonomous vehicles or machines (e.g., in one or more adaptive driver assistance systems (ADAS)), autonomous vehicles or machines, piloted and un-piloted robots or robotic platforms, warehouse vehicles, off-road vehicles, vehicles coupled to one or more trailers, flying vessels, boats, shuttles, emergency response vehicles, motorcycles, electric or motorized bicycles, aircraft, construction vehicles, underwater craft, drones, and/or other vehicle types. Further, the systems and methods described herein may be used for a variety of purposes, by way of example and without limitation, for machine control, machine locomotion, machine driving, synthetic data generation, model training, perception, augmented reality, virtual reality, mixed reality, robotics, security and surveillance, simulation and digital twinning, autonomous or semi-autonomous machine applications, deep learning, environment simulation, object or actor simulation and/or digital twinning, data center processing, conversational AI, light transport simulation (e.g., ray-tracing, path tracing, etc.), collaborative content creation for 3D assets, cloud computing and/or any other suitable applications.
Disclosed embodiments may be comprised in a variety of different systems such as automotive systems (e.g., a control system for an autonomous or semi-autonomous machine, a perception system for an autonomous or semi-autonomous machine), systems implemented using a robot, aerial systems, medial systems, boating systems, smart area monitoring systems, systems for performing deep learning operations, systems for performing simulation operations, systems for performing digital twin operations, systems implemented using an edge device, systems incorporating one or more virtual machines (VMs), systems for performing synthetic data generation operations, systems implemented at least partially in a data center, systems for performing conversational AI operations, systems for hosting real-time streaming applications, systems for presenting one or more of virtual reality content, augmented reality content, or mixed reality content, systems for performing light transport simulation, systems for performing collaborative content creation for 3D assets, systems implemented at least partially using cloud computing resources, and/or other types of systems.
The vehicle 100 may include components such as a chassis, a vehicle body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other components of a vehicle. The vehicle 100 may include a propulsion system 150, such as an internal combustion engine, hybrid electric power plant, an all-electric engine, and/or another propulsion system type. The propulsion system 150 may be connected to a drive train of the vehicle 100, which may include a transmission, to enable the propulsion of the vehicle 100. The propulsion system 150 may be controlled in response to receiving signals from the throttle/accelerator 152.
A steering system 154, which may include a steering wheel, may be used to steer the vehicle 100 (e.g., along a desired path or route) when the propulsion system 150 is operating (e.g., when the vehicle is in motion). The steering system 154 may receive signals from a steering actuator 156. The steering wheel may be optional for full automation (Level 5) functionality.
The brake sensor system 146 may be used to operate the vehicle brakes in response to receiving signals from the brake actuators 148 and/or brake sensors.
Controller(s) 136, which may include one or more system on chips (SoCs) 104 (
The controller(s) 136 may provide the signals for controlling one or more components and/or systems of the vehicle 100 in response to sensor data received from one or more sensors (e.g., sensor inputs). The sensor data may be received from, for example and without limitation, global navigation satellite systems (“GNSS”) sensor(s) 158 (e.g., Global Positioning System sensor(s)), RADAR sensor(s) 160, ultrasonic sensor(s) 162, LIDAR sensor(s) 164, inertial measurement unit (IMU) sensor(s) 166 (e.g., accelerometer(s), gyroscope(s), magnetic compass(es), magnetometer(s), etc.), microphone(s) 196, stereo camera(s) 168, wide-view camera(s) 170 (e.g., fisheye cameras), infrared camera(s) 172, surround camera(s) 174 (e.g., 360 degree cameras), long-range and/or mid-range camera(s) 198, speed sensor(s) 144 (e.g., for measuring the speed of the vehicle 100), vibration sensor(s) 142, steering sensor(s) 140, brake sensor(s) (e.g., as part of the brake sensor system 146), and/or other sensor types.
One or more of the controller(s) 136 may receive inputs (e.g., represented by input data) from an instrument cluster 132 of the vehicle 100 and provide outputs (e.g., represented by output data, display data, etc.) via a human-machine interface (HMI) display 134, an audible annunciator, a loudspeaker, and/or via other components of the vehicle 100. The outputs may include information such as vehicle velocity, speed, time, map data (e.g., the High Definition (“HD”) map 122 of
The vehicle 100 further includes a network interface 124 which may use one or more wireless antenna(s) 126 and/or modem(s) to communicate over one or more networks. For example, the network interface 124 may be capable of communication over Long-Term Evolution (“LTE”), Wideband Code Division Multiple Access (“WCDMA”), Universal Mobile Telecommunications System (“UMTS”), Global System for Mobile communication (“GSM”), IMT-CDMA Multi-Carrier (“CDMA2000”), etc. The wireless antenna(s) 126 may also enable communication between objects in the environment (e.g., vehicles, mobile devices, etc.), using local area network(s), such as Bluetooth, Bluetooth Low Energy (“LE”), Z-Wave, ZigBee, etc., and/or low power wide-area network(s) (“LPWANs”), such as LoRaWAN, SigFox, etc.
The camera types for the cameras may include, but are not limited to, digital cameras that may be adapted for use with the components and/or systems of the vehicle 100. The camera(s) may operate at automotive safety integrity level (ASIL) B and/or at another ASIL. The camera types may be capable of any image capture rate, such as 60 frames per second (fps), 120 fps, 240 fps, etc., depending on the embodiment. The cameras may be capable of using rolling shutters, global shutters, another type of shutter, or a combination thereof. In some examples, the color filter array may include a red clear clear clear (RCCC) color filter array, a red clear clear blue (RCCB) color filter array, a red blue green clear (RBGC) color filter array, a Foveon X3 color filter array, a Bayer sensors (RGGB) color filter array, a monochrome sensor color filter array, and/or another type of color filter array. In some embodiments, clear pixel cameras, such as cameras with an RCCC, an RCCB, and/or an RBGC color filter array, may be used in an effort to increase light sensitivity.
In some examples, one or more of the camera(s) may be used to perform advanced driver assistance systems (ADAS) functions (e.g., as part of a redundant or fail-safe design). For example, a Multi-Function Mono Camera may be installed to provide functions including lane departure warning, traffic sign assist and intelligent headlamp control. One or more of the camera(s) (e.g., all of the cameras) may record and provide image data (e.g., video) simultaneously.
One or more of the cameras may be mounted in a mounting assembly, such as a custom designed (three dimensional (“3D”) printed) assembly, in order to cut out stray light and reflections from within the car (e.g., reflections from the dashboard reflected in the windshield mirrors) which may interfere with the camera's image data capture abilities. With reference to wing-mirror mounting assemblies, the wing-mirror assemblies may be custom 3D printed so that the camera mounting plate matches the shape of the wing-mirror. In some examples, the camera(s) may be integrated into the wing-mirror. For side-view cameras, the camera(s) may also be integrated within the four pillars at each corner of the cabin.
Cameras with a field of view that include portions of the environment in front of the vehicle 100 (e.g., front-facing cameras) may be used for surround view, to help identify forward facing paths and obstacles, as well aid in, with the help of one or more controllers 136 and/or control SoCs, providing information critical to generating an occupancy grid and/or determining the preferred vehicle paths. Front-facing cameras may be used to perform many of the same ADAS functions as LIDAR, including emergency braking, pedestrian detection, and collision avoidance. Front-facing cameras may also be used for ADAS functions and systems including Lane Departure Warnings (“LDW”), Autonomous Cruise Control (“ACC”), and/or other functions such as traffic sign recognition.
A variety of cameras may be used in a front-facing configuration, including, for example, a monocular camera platform that includes a complementary metal oxide semiconductor (“CMOS”) color imager. Another example may be a wide-view camera(s) 170 that may be used to perceive objects coming into view from the periphery (e.g., pedestrians, crossing traffic or bicycles). Although only one wide-view camera is illustrated in
Any number of stereo cameras 168 may also be included in a front-facing configuration. In at least one embodiment, one or more of stereo camera(s) 168 may include an integrated control unit comprising a scalable processing unit, which may provide a programmable logic (“FPGA”) and a multi-core micro-processor with an integrated Controller Area Network (“CAN”) or Ethernet interface on a single chip. Such a unit may be used to generate a 3D map of the vehicle's environment, including a distance estimate for all the points in the image. An alternative stereo camera(s) 168 may include a compact stereo vision sensor(s) that may include two camera lenses (one each on the left and right) and an image processing chip that may measure the distance from the vehicle to the target object and use the generated information (e.g., metadata) to activate the autonomous emergency braking and lane departure warning functions. Other types of stereo camera(s) 168 may be used in addition to, or alternatively from, those described herein.
Cameras with a field of view that include portions of the environment to the side of the vehicle 100 (e.g., side-view cameras) may be used for surround view, providing information used to create and update the occupancy grid, as well as to generate side impact collision warnings. For example, surround camera(s) 174 (e.g., four surround cameras 174 as illustrated in
Cameras with a field of view that include portions of the environment to the rear of the vehicle 100 (e.g., rear-view cameras) may be used for park assistance, surround view, rear collision warnings, and creating and updating the occupancy grid. A wide variety of cameras may be used including, but not limited to, cameras that are also suitable as a front-facing camera(s) (e.g., long-range and/or mid-range camera(s) 198, stereo camera(s) 168), infrared camera(s) 172, etc.), as described herein.
Each of the components, features, and systems of the vehicle 100 in
Although the bus 102 is described herein as being a CAN bus, this is not intended to be limiting. For example, in addition to, or alternatively from, the CAN bus, FlexRay and/or Ethernet may be used. Additionally, although a single line is used to represent the bus 102, this is not intended to be limiting. For example, there may be any number of busses 102, which may include one or more CAN busses, one or more FlexRay busses, one or more Ethernet busses, and/or one or more other types of busses using a different protocol. In some examples, two or more busses 102 may be used to perform different functions, and/or may be used for redundancy. For example, a first bus 102 may be used for collision avoidance functionality and a second bus 102 may be used for actuation control. In any example, each bus 102 may communicate with any of the components of the vehicle 100, and two or more busses 102 may communicate with the same components. In some examples, each SoC 104, each controller 136, and/or each computer within the vehicle may have access to the same input data (e.g., inputs from sensors of the vehicle 100), and may be connected to a common bus, such the CAN bus.
The vehicle 100 may include one or more controller(s) 136, such as those described herein with respect to
The vehicle 100 may include a system(s) on a chip (SoC) 104. The SoC 104 may include CPU(s) 106, GPU(s) 108, processor(s) 110, cache(s) 112, accelerator(s) 114, data store(s) 116, and/or other components and features not illustrated. The SoC(s) 104 may be used to control the vehicle 100 in a variety of platforms and systems. For example, the SoC(s) 104 may be combined in a system (e.g., the system of the vehicle 100) with an HD map 122 which may obtain map refreshes and/or updates via a network interface 124 from one or more servers (e.g., server(s) 178 of
The CPU(s) 106 may include a CPU cluster or CPU complex (alternatively referred to herein as a “CCPLEX”). The CPU(s) 106 may include multiple cores and/or L2 caches. For example, in some embodiments, the CPU(s) 106 may include eight cores in a coherent multi-processor configuration. In some embodiments, the CPU(s) 106 may include four dual-core clusters where each cluster has a dedicated L2 cache (e.g., a 2 MB L2 cache). The CPU(s) 106 (e.g., the CCPLEX) may be configured to support simultaneous cluster operation enabling any combination of the clusters of the CPU(s) 106 to be active at any given time.
The CPU(s) 106 may implement power management capabilities that include one or more of the following features: individual hardware blocks may be clock-gated automatically when idle to save dynamic power; each core clock may be gated when the core is not actively executing instructions due to execution of WFI/WFE instructions; each core may be independently power-gated; each core cluster may be independently clock-gated when all cores are clock-gated or power-gated; and/or each core cluster may be independently power-gated when all cores are power-gated. The CPU(s) 106 may further implement an enhanced algorithm for managing power states, where allowed power states and expected wakeup times are specified, and the hardware/microcode determines the best power state to enter for the core, cluster, and CCPLEX. The processing cores may support simplified power state entry sequences in software with the work offloaded to microcode.
The GPU(s) 108 may include an integrated GPU (alternatively referred to herein as an “iGPU”). The GPU(s) 108 may be programmable and may be efficient for parallel workloads. The GPU(s) 108, in some examples, may use an enhanced tensor instruction set. The GPU(s) 108 may include one or more streaming microprocessors, where each streaming microprocessor may include an L1 cache (e.g., an L1 cache with at least 96 KB storage capacity), and two or more of the streaming microprocessors may share an L2 cache (e.g., an L2 cache with a 512 KB storage capacity). In some embodiments, the GPU(s) 108 may include at least eight streaming microprocessors. The GPU(s) 108 may use compute application programming interface(s) (API(s)). In addition, the GPU(s) 108 may use one or more parallel computing platforms and/or programming models (e.g., NVIDIA's CUDA).
The GPU(s) 108 may be power-optimized for best performance in automotive and embedded use cases. For example, the GPU(s) 108 may be fabricated on a Fin field-effect transistor (FinFET). However, this is not intended to be limiting and the GPU(s) 108 may be fabricated using other semiconductor manufacturing processes. Each streaming microprocessor may incorporate a number of mixed-precision processing cores partitioned into multiple blocks. For example, and without limitation, 64 PF32 cores and 32 PF64 cores may be partitioned into four processing blocks. In such an example, each processing block may be allocated 16 FP32 cores, 8 FP64 cores, 16 INT32 cores, two mixed-precision NVIDIA TENSOR COREs for deep learning matrix arithmetic, an L0 instruction cache, a warp scheduler, a dispatch unit, and/or a 64 KB register file. In addition, the streaming microprocessors may include independent parallel integer and floating-point data paths to provide for efficient execution of workloads with a mix of computation and addressing calculations. The streaming microprocessors may include independent thread scheduling capability to enable finer-grain synchronization and cooperation between parallel threads. The streaming microprocessors may include a combined L1 data cache and shared memory unit in order to improve performance while simplifying programming.
The GPU(s) 108 may include a high bandwidth memory (HBM) and/or a 16 GB HBM2 memory subsystem to provide, in some examples, about 900 GB/second peak memory bandwidth. In some examples, in addition to, or alternatively from, the HBM memory, a synchronous graphics random-access memory (SGRAM) may be used, such as a graphics double data rate type five synchronous random-access memory (GDDR5).
The GPU(s) 108 may include unified memory technology including access counters to allow for more accurate migration of memory pages to the processor that accesses them most frequently, thereby improving efficiency for memory ranges shared between processors. In some examples, address translation services (ATS) support may be used to allow the GPU(s) 108 to access the CPU(s) 106 page tables directly. In such examples, when the GPU(s) 108 memory management unit (MMU) experiences a miss, an address translation request may be transmitted to the CPU(s) 106. In response, the CPU(s) 106 may look in its page tables for the virtual-to-physical mapping for the address and transmits the translation back to the GPU(s) 108. As such, unified memory technology may allow a single unified virtual address space for memory of both the CPU(s) 106 and the GPU(s) 108, thereby simplifying the GPU(s) 108 programming and porting of applications to the GPU(s) 108.
In addition, the GPU(s) 108 may include an access counter that may keep track of the frequency of access of the GPU(s) 108 to memory of other processors. The access counter may help ensure that memory pages are moved to the physical memory of the processor that is accessing the pages most frequently.
The SoC(s) 104 may include any number of cache(s) 112, including those described herein. For example, the cache(s) 112 may include an L3 cache that is available to both the CPU(s) 106 and the GPU(s) 108 (e.g., that is connected both the CPU(s) 106 and the GPU(s) 108). The cache(s) 112 may include a write-back cache that may keep track of states of lines, such as by using a cache coherence protocol (e.g., MEI, MESI, MSI, etc.). The L3 cache may include 4 MB or more, depending on the embodiment, although smaller cache sizes may be used.
The SoC(s) 104 may include an arithmetic logic unit(s) (ALU(s)) which may be leveraged in performing processing with respect to any of the variety of tasks or operations of the vehicle 100—such as processing DNNs. In addition, the SoC(s) 104 may include a floating point unit(s) (FPU(s))—or other math coprocessor or numeric coprocessor types—for performing mathematical operations within the system. For example, the SoC(s) 104 may include one or more FPUs integrated as execution units within a CPU(s) 106 and/or GPU(s) 108.
The SoC(s) 104 may include one or more accelerators 114 (e.g., hardware accelerators, software accelerators, or a combination thereof). For example, the SoC(s) 104 may include a hardware acceleration cluster that may include optimized hardware accelerators and/or large on-chip memory. The large on-chip memory (e.g., 4 MB of SRAM), may enable the hardware acceleration cluster to accelerate neural networks and other calculations. The hardware acceleration cluster may be used to complement the GPU(s) 108 and to off-load some of the tasks of the GPU(s) 108 (e.g., to free up more cycles of the GPU(s) 108 for performing other tasks). As an example, the accelerator(s) 114 may be used for targeted workloads (e.g., perception, convolutional neural networks (CNNs), etc.) that are stable enough to be amenable to acceleration. The term “CNN,” as used herein, may include all types of CNNs, including region-based or regional convolutional neural networks (RCNNs) and Fast RCNNs (e.g., as used for object detection).
The accelerator(s) 114 (e.g., the hardware acceleration cluster) may include a deep learning accelerator(s) (DLA). The DLA(s) may include one or more Tensor processing units (TPUs) that may be configured to provide an additional ten trillion operations per second for deep learning applications and inferencing. The TPUs may be accelerators configured to, and optimized for, performing image processing functions (e.g., for CNNs, RCNNs, etc.). The DLA(s) may further be optimized for a specific set of neural network types and floating point operations, as well as inferencing. The design of the DLA(s) may provide more performance per millimeter than a general-purpose GPU, and vastly exceeds the performance of a CPU. The TPU(s) may perform several functions, including a single-instance convolution function, supporting, for example, INT8, INT16, and FP16 data types for both features and weights, as well as post-processor functions.
The DLA(s) may quickly and efficiently execute neural networks, especially CNNs, on processed or unprocessed data for any of a variety of functions, including, for example and without limitation: a CNN for object identification and detection using data from camera sensors; a CNN for distance estimation using data from camera sensors; a CNN for emergency vehicle detection and identification and detection using data from microphones; a CNN for facial recognition and vehicle owner identification using data from camera sensors; and/or a CNN for security and/or safety related events.
The DLA(s) may perform any function of the GPU(s) 108, and by using an inference accelerator, for example, a designer may target either the DLA(s) or the GPU(s) 108 for any function. For example, the designer may focus processing of CNNs and floating point operations on the DLA(s) and leave other functions to the GPU(s) 108 and/or other accelerator(s) 114.
The accelerator(s) 114 (e.g., the hardware acceleration cluster) may include a programmable vision accelerator(s) (PVA), which may alternatively be referred to herein as a computer vision accelerator. The PVA(s) may be designed and configured to accelerate computer vision algorithms for the advanced driver assistance systems (ADAS), autonomous driving, and/or augmented reality (AR) and/or virtual reality (VR) applications. The PVA(s) may provide a balance between performance and flexibility. For example, each PVA(s) may include, for example and without limitation, any number of reduced instruction set computer (RISC) cores, direct memory access (DMA), and/or any number of vector processors.
The RISC cores may interact with image sensors (e.g., the image sensors of any of the cameras described herein), image signal processor(s), and/or the like. Each of the RISC cores may include any amount of memory. The RISC cores may use any of a number of protocols, depending on the embodiment. In some examples, the RISC cores may execute a real-time operating system (RTOS). The RISC cores may be implemented using one or more integrated circuit devices, application specific integrated circuits (ASICs), and/or memory devices. For example, the RISC cores may include an instruction cache and/or a tightly coupled RAM.
The DMA may enable components of the PVA(s) to access the system memory independently of the CPU(s) 106. The DMA may support any number of features used to provide optimization to the PVA including, but not limited to, supporting multi-dimensional addressing and/or circular addressing. In some examples, the DMA may support up to six or more dimensions of addressing, which may include block width, block height, block depth, horizontal block stepping, vertical block stepping, and/or depth stepping.
The vector processors may be programmable processors that may be designed to efficiently and flexibly execute programming for computer vision algorithms and provide signal processing capabilities. In some examples, the PVA may include a PVA core and two vector processing subsystem partitions. The PVA core may include a processor subsystem, DMA engine(s) (e.g., two DMA engines), and/or other peripherals. The vector processing subsystem may operate as the primary processing engine of the PVA, and may include a vector processing unit (VPU), an instruction cache, and/or vector memory (e.g., VMEM). A VPU core may include a digital signal processor such as, for example, a single instruction, multiple data (SIMD), very long instruction word (VLIW) digital signal processor. The combination of the SIMD and VLIW may enhance throughput and speed.
Each of the vector processors may include an instruction cache and may be coupled to dedicated memory. As a result, in some examples, each of the vector processors may be configured to execute independently of the other vector processors. In other examples, the vector processors that are included in a particular PVA may be configured to employ data parallelism. For example, in some embodiments, the plurality of vector processors included in a single PVA may execute the same computer vision algorithm, but on different regions of an image. In other examples, the vector processors included in a particular PVA may simultaneously execute different computer vision algorithms, on the same image, or even execute different algorithms on sequential images or portions of an image. Among other things, any number of PVAs may be included in the hardware acceleration cluster and any number of vector processors may be included in each of the PVAs. In addition, the PVA(s) may include additional error correcting code (ECC) memory, to enhance overall system safety.
The accelerator(s) 114 (e.g., the hardware acceleration cluster) may include a computer vision network on-chip and SRAM, for providing a high-bandwidth, low latency SRAM for the accelerator(s) 114. In some examples, the on-chip memory may include at least 4 MB SRAM, consisting of, for example and without limitation, eight field-configurable memory blocks, that may be accessible by both the PVA and the DLA. Each pair of memory blocks may include an advanced peripheral bus (APB) interface, configuration circuitry, a controller, and a multiplexer. Any type of memory may be used. The PVA and DLA may access the memory via a backbone that provides the PVA and DLA with high-speed access to memory. The backbone may include a computer vision network on-chip that interconnects the PVA and the DLA to the memory (e.g., using the APB).
The computer vision network on-chip may include an interface that determines, before transmission of any control signal/address/data, that both the PVA and the DLA provide ready and valid signals. Such an interface may provide for separate phases and separate channels for transmitting control signals/addresses/data, as well as burst-type communications for continuous data transfer. This type of interface may comply with ISO 26262 or IEC 61508 standards, although other standards and protocols may be used.
In some examples, the SoC(s) 104 may include a real-time ray-tracing hardware accelerator, such as described in U.S. patent application Ser. No. 16/101,232, filed on Aug. 10, 2018. The real-time ray-tracing hardware accelerator may be used to quickly and efficiently determine the positions and extents of objects (e.g., within a world model), to generate real-time visualization simulations, for RADAR signal interpretation, for sound propagation synthesis and/or analysis, for simulation of SONAR systems, for general wave propagation simulation, for comparison to LIDAR data for purposes of localization and/or other functions, and/or for other uses. In some embodiments, one or more tree traversal units (TTUs) may be used for executing one or more ray-tracing related operations.
The accelerator(s) 114 (e.g., the hardware accelerator cluster) have a wide array of uses for autonomous driving. The PVA may be a programmable vision accelerator that may be used for key processing stages in ADAS and autonomous vehicles. The PVA's capabilities are a good match for algorithmic domains needing predictable processing, at low power and low latency. In other words, the PVA performs well on semi-dense or dense regular computation, even on small data sets, which need predictable run-times with low latency and low power. Thus, in the context of platforms for autonomous vehicles, the PVAs are designed to run classic computer vision algorithms, as they are efficient at object detection and operating on integer math.
For example, according to one embodiment of the technology, the PVA is used to perform computer stereo vision. A semi-global matching-based algorithm may be used in some examples, although this is not intended to be limiting. Many applications for Level 3-5 autonomous driving require motion estimation/stereo matching on-the-fly (e.g., structure from motion, pedestrian recognition, lane detection, etc.). The PVA may perform computer stereo vision function on inputs from two monocular cameras.
In some examples, the PVA may be used to perform dense optical flow. According to process raw RADAR data (e.g., using a 4D Fast Fourier Transform) to provide Processed RADAR. In other examples, the PVA is used for time of flight depth processing, by processing raw time of flight data to provide processed time of flight data, for example.
The DLA may be used to run any type of network to enhance control and driving safety, including for example, a neural network that outputs a measure of confidence for each object detection. Such a confidence value may be interpreted as a probability, or as providing a relative “weight” of each detection compared to other detections. This confidence value enables the system to make further decisions regarding which detections should be considered as true positive detections rather than false positive detections. For example, the system may set a threshold value for the confidence and consider only the detections exceeding the threshold value as true positive detections. In an automatic emergency braking (AEB) system, false positive detections would cause the vehicle to automatically perform emergency braking, which is obviously undesirable. Therefore, only the most confident detections should be considered as triggers for AEB. The DLA may run a neural network for regressing the confidence value. The neural network may take as its input at least some subset of parameters, such as bounding box dimensions, ground plane estimate obtained (e.g. from another subsystem), inertial measurement unit (IMU) sensor 166 output that correlates with the vehicle 100 orientation, distance, 3D location estimates of the object obtained from the neural network and/or other sensors (e.g., LIDAR sensor(s) 164 or RADAR sensor(s) 160), among others.
The SoC(s) 104 may include data store(s) 116 (e.g., memory). The data store(s) 116 may be on-chip memory of the SoC(s) 104, which may store neural networks to be executed on the GPU and/or the DLA. In some examples, the data store(s) 116 may be large enough in capacity to store multiple instances of neural networks for redundancy and safety. The data store(s) 112 may comprise L2 or L3 cache(s) 112. Reference to the data store(s) 116 may include reference to the memory associated with the PVA, DLA, and/or other accelerator(s) 114, as described herein.
The SoC(s) 104 may include one or more processor(s) 110 (e.g., embedded processors). The processor(s) 110 may include a boot and power management processor that may be a dedicated processor and subsystem to handle boot power and management functions and related security enforcement. The boot and power management processor may be a part of the SoC(s) 104 boot sequence and may provide runtime power management services. The boot power and management processor may provide clock and voltage programming, assistance in system low power state transitions, management of SoC(s) 104 thermals and temperature sensors, and/or management of the SoC(s) 104 power states. Each temperature sensor may be implemented as a ring-oscillator whose output frequency is proportional to temperature, and the SoC(s) 104 may use the ring-oscillators to detect temperatures of the CPU(s) 106, GPU(s) 108, and/or accelerator(s) 114. If temperatures are determined to exceed a threshold, the boot and power management processor may enter a temperature fault routine and put the SoC(s) 104 into a lower power state and/or put the vehicle 100 into a chauffeur to safe stop mode (e.g., bring the vehicle 100 to a safe stop).
The processor(s) 110 may further include a set of embedded processors that may serve as an audio processing engine. The audio processing engine may be an audio subsystem that enables full hardware support for multi-channel audio over multiple interfaces, and a broad and flexible range of audio I/O interfaces. In some examples, the audio processing engine is a dedicated processor core with a digital signal processor with dedicated RAM.
The processor(s) 110 may further include an always on processor engine that may provide necessary hardware features to support low power sensor management and wake use cases. The always on processor engine may include a processor core, a tightly coupled RAM, supporting peripherals (e.g., timers and interrupt controllers), various I/O controller peripherals, and routing logic.
The processor(s) 110 may further include a safety cluster engine that includes a dedicated processor subsystem to handle safety management for automotive applications. The safety cluster engine may include two or more processor cores, a tightly coupled RAM, support peripherals (e.g., timers, an interrupt controller, etc.), and/or routing logic. In a safety mode, the two or more cores may operate in a lockstep mode and function as a single core with comparison logic to detect any differences between their operations.
The processor(s) 110 may further include a real-time camera engine that may include a dedicated processor subsystem for handling real-time camera management.
The processor(s) 110 may further include a high-dynamic range signal processor that may include an image signal processor that is a hardware engine that is part of the camera processing pipeline.
The processor(s) 110 may include a video image compositor that may be a processing block (e.g., implemented on a microprocessor) that implements video post-processing functions needed by a video playback application to produce the final image for the player window. The video image compositor may perform lens distortion correction on wide-view camera(s) 170, surround camera(s) 174, and/or on in-cabin monitoring camera sensors. In-cabin monitoring camera sensor is preferably monitored by a neural network running on another instance of the Advanced SoC, configured to identify in cabin events and respond accordingly. An in-cabin system may perform lip reading to activate cellular service and place a phone call, dictate emails, change the vehicle's destination, activate or change the vehicle's infotainment system and settings, or provide voice-activated web surfing. Certain functions are available to the driver only when the vehicle is operating in an autonomous mode, and are disabled otherwise.
The video image compositor may include enhanced temporal noise reduction for both spatial and temporal noise reduction. For example, where motion occurs in a video, the noise reduction weights spatial information appropriately, decreasing the weight of information provided by adjacent frames. Where an image or portion of an image does not include motion, the temporal noise reduction performed by the video image compositor may use information from the previous image to reduce noise in the current image.
The video image compositor may also be configured to perform stereo rectification on input stereo lens frames. The video image compositor may further be used for user interface composition when the operating system desktop is in use, and the GPU(s) 108 is not required to continuously render new surfaces. Even when the GPU(s) 108 is powered on and active doing 3D rendering, the video image compositor may be used to offload the GPU(s) 108 to improve performance and responsiveness.
The SoC(s) 104 may further include a mobile industry processor interface (MIPI) camera serial interface for receiving video and input from cameras, a high-speed interface, and/or a video input block that may be used for camera and related pixel input functions. The SoC(s) 104 may further include an input/output controller(s) that may be controlled by software and may be used for receiving I/O signals that are uncommitted to a specific role.
The SoC(s) 104 may further include a broad range of peripheral interfaces to enable communication with peripherals, audio codecs, power management, and/or other devices. The SoC(s) 104 may be used to process data from cameras (e.g., connected over Gigabit Multimedia Serial Link and Ethernet), sensors (e.g., LIDAR sensor(s) 164, RADAR sensor(s) 160, etc. that may be connected over Ethernet), data from bus 102 (e.g., speed of vehicle 100, steering wheel position, etc.), data from GNSS sensor(s) 158 (e.g., connected over Ethernet or CAN bus). The SoC(s) 104 may further include dedicated high-performance mass storage controllers that may include their own DMA engines, and that may be used to free the CPU(s) 106 from routine data management tasks.
The SoC(s) 104 may be an end-to-end platform with a flexible architecture that spans automation levels 3-5, thereby providing a comprehensive functional safety architecture that leverages and makes efficient use of computer vision and ADAS techniques for diversity and redundancy, provides a platform for a flexible, reliable driving software stack, along with deep learning tools. The SoC(s) 104 may be faster, more reliable, and even more energy-efficient and space-efficient than conventional systems. For example, the accelerator(s) 114, when combined with the CPU(s) 106, the GPU(s) 108, and the data store(s) 116, may provide for a fast, efficient platform for level 3-5 autonomous vehicles.
The technology thus provides capabilities and functionality that cannot be achieved by conventional systems. For example, computer vision algorithms may be executed on CPUs, which may be configured using high-level programming language, such as the C programming language, to execute a wide variety of processing algorithms across a wide variety of visual data. However, CPUs are oftentimes unable to meet the performance requirements of many computer vision applications, such as those related to execution time and power consumption, for example. In particular, many CPUs are unable to execute complex object detection algorithms in real-time, which is a requirement of in-vehicle ADAS applications, and a requirement for practical Level 3-5 autonomous vehicles.
In contrast to conventional systems, by providing a CPU complex, GPU complex, and a hardware acceleration cluster, the technology described herein allows for multiple neural networks to be performed simultaneously and/or sequentially, and for the results to be combined together to enable Level 3-5 autonomous driving functionality. For example, a CNN executing on the DLA or dGPU (e.g., the GPU(s) 120) may include a text and word recognition, allowing the supercomputer to read and understand traffic signs, including signs for which the neural network has not been specifically trained. The DLA may further include a neural network that is able to identify, interpret, and provides semantic understanding of the sign, and to pass that semantic understanding to the path planning modules running on the CPU Complex.
As another example, multiple neural networks may be run simultaneously, as is required for Level 3, 4, or 5 driving. For example, a warning sign consisting of “Caution: flashing lights indicate icy conditions,” along with an electric light, may be independently or collectively interpreted by several neural networks. The sign itself may be identified as a traffic sign by a first deployed neural network (e.g., a neural network that has been trained), the text “Flashing lights indicate icy conditions” may be interpreted by a second deployed neural network, which informs the vehicle's path planning software (preferably executing on the CPU Complex) that when flashing lights are detected, icy conditions exist. The flashing light may be identified by operating a third deployed neural network over multiple frames, informing the vehicle's path-planning software of the presence (or absence) of flashing lights. All three neural networks may run simultaneously, such as within the DLA and/or on the GPU(s) 108.
In some examples, a CNN for facial recognition and vehicle owner identification may use data from camera sensors to identify the presence of an authorized driver and/or owner of the vehicle 100. The always on sensor processing engine may be used to unlock the vehicle when the owner approaches the driver door and turn on the lights, and, in security mode, to disable the vehicle when the owner leaves the vehicle. In this way, the SoC(s) 104 provide for security against theft and/or carjacking.
In another example, a CNN for emergency vehicle detection and identification may use data from microphones 196 to detect and identify emergency vehicle sirens. In contrast to conventional systems, that use general classifiers to detect sirens and manually extract features, the SoC(s) 104 use the CNN for classifying environmental and urban sounds, as well as classifying visual data. In a preferred embodiment, the CNN running on the DLA is trained to identify the relative closing speed of the emergency vehicle (e.g., by using the Doppler Effect). The CNN may also be trained to identify emergency vehicles specific to the local area in which the vehicle is operating, as identified by GNSS sensor(s) 158. Thus, for example, when operating in Europe the CNN will seek to detect European sirens, and when in the United States the CNN will seek to identify only North American sirens. Once an emergency vehicle is detected, a control program may be used to execute an emergency vehicle safety routine, slowing the vehicle, pulling over to the side of the road, parking the vehicle, and/or idling the vehicle, with the assistance of ultrasonic sensors 162, until the emergency vehicle(s) passes.
The vehicle may include a CPU(s) 118 (e.g., discrete CPU(s), or dCPU(s)), that may be coupled to the SoC(s) 104 via a high-speed interconnect (e.g., PCIe). The CPU(s) 118 may include an X86 processor, for example. The CPU(s) 118 may be used to perform any of a variety of functions, including arbitrating potentially inconsistent results between ADAS sensors and the SoC(s) 104, and/or monitoring the status and health of the controller(s) 136 and/or infotainment SoC 130, for example.
The vehicle 100 may include a GPU(s) 120 (e.g., discrete GPU(s), or dGPU(s)), that may be coupled to the SoC(s) 104 via a high-speed interconnect (e.g., NVIDIA's NVLINK). The GPU(s) 120 may provide additional artificial intelligence functionality, such as by executing redundant and/or different neural networks, and may be used to train and/or update neural networks based on input (e.g., sensor data) from sensors of the vehicle 100.
The vehicle 100 may further include the network interface 124 which may include one or more wireless antennas 126 (e.g., one or more wireless antennas for different communication protocols, such as a cellular antenna, a Bluetooth antenna, etc.). The network interface 124 may be used to enable wireless connectivity over the Internet with the cloud (e.g., with the server(s) 178 and/or other network devices), with other vehicles, and/or with computing devices (e.g., client devices of passengers). To communicate with other vehicles, a direct link may be established between the two vehicles and/or an indirect link may be established (e.g., across networks and over the Internet). Direct links may be provided using a vehicle-to-vehicle communication link. The vehicle-to-vehicle communication link may provide the vehicle 100 information about vehicles in proximity to the vehicle 100 (e.g., vehicles in front of, on the side of, and/or behind the vehicle 100). This functionality may be part of a cooperative adaptive cruise control functionality of the vehicle 100.
The network interface 124 may include a SoC that provides modulation and demodulation functionality and enables the controller(s) 136 to communicate over wireless networks. The network interface 124 may include a radio frequency front-end for up-conversion from baseband to radio frequency, and down conversion from radio frequency to baseband. The frequency conversions may be performed through well-known processes, and/or may be performed using super-heterodyne processes. In some examples, the radio frequency front end functionality may be provided by a separate chip. The network interface may include wireless functionality for communicating over LTE, WCDMA, UMTS, GSM, CDMA2000, Bluetooth, Bluetooth LE, Wi-Fi, Z-Wave, ZigBee, LoRaWAN, and/or other wireless protocols.
The vehicle 100 may further include data store(s) 128 which may include off-chip (e.g., off the SoC(s) 104) storage. The data store(s) 128 may include one or more storage elements including RAM, SRAM, DRAM, VRAM, Flash, hard disks, and/or other components and/or devices that may store at least one bit of data.
The vehicle 100 may further include GNSS sensor(s) 158. The GNSS sensor(s) 158 (e.g., GPS, assisted GPS sensors, differential GPS (DGPS) sensors, etc.), to assist in mapping, perception, occupancy grid generation, and/or path planning functions. Any number of GNSS sensor(s) 158 may be used, including, for example and without limitation, a GPS using a USB connector with an Ethernet to Serial (RS-232) bridge.
The vehicle 100 may further include RADAR sensor(s) 160. The RADAR sensor(s) 160 may be used by the vehicle 100 for long-range vehicle detection, even in darkness and/or severe weather conditions. RADAR functional safety levels may be ASIL B. The RADAR sensor(s) 160 may use the CAN and/or the bus 102 (e.g., to transmit data generated by the RADAR sensor(s) 160) for control and to access object tracking data, with access to Ethernet to access raw data in some examples. A wide variety of RADAR sensor types may be used. For example, and without limitation, the RADAR sensor(s) 160 may be suitable for front, rear, and side RADAR use. In some example, Pulse Doppler RADAR sensor(s) are used.
The RADAR sensor(s) 160 may include different configurations, such as long range with narrow field of view, short range with wide field of view, short range side coverage, etc. In some examples, long-range RADAR may be used for adaptive cruise control functionality. The long-range RADAR systems may provide a broad field of view realized by two or more independent scans, such as within a 250 m range. The RADAR sensor(s) 160 may help in distinguishing between static and moving objects, and may be used by ADAS systems for emergency brake assist and forward collision warning. Long-range RADAR sensors may include monostatic multimodal RADAR with multiple (e.g., six or more) fixed RADAR antennae and a high-speed CAN and FlexRay interface. In an example with six antennae, the central four antennae may create a focused beam pattern, designed to record the vehicle's 100 surroundings at higher speeds with minimal interference from traffic in adjacent lanes. The other two antennae may expand the field of view, making it possible to quickly detect vehicles entering or leaving the vehicle's 100 lane.
Mid-range RADAR systems may include, as an example, a range of up to 160 m (front) or 80 m (rear), and a field of view of up to 42 degrees (front) or 150 degrees (rear). Short-range RADAR systems may include, without limitation, RADAR sensors designed to be installed at both ends of the rear bumper. When installed at both ends of the rear bumper, such a RADAR sensor systems may create two beams that constantly monitor the blind spot in the rear and next to the vehicle.
Short-range RADAR systems may be used in an ADAS system for blind spot detection and/or lane change assist.
The vehicle 100 may further include ultrasonic sensor(s) 162. The ultrasonic sensor(s) 162, which may be positioned at the front, back, and/or the sides of the vehicle 100, may be used for park assist and/or to create and update an occupancy grid. A wide variety of ultrasonic sensor(s) 162 may be used, and different ultrasonic sensor(s) 162 may be used for different ranges of detection (e.g., 2.5 m, 4 m). The ultrasonic sensor(s) 162 may operate at functional safety levels of ASIL B.
The vehicle 100 may include LIDAR sensor(s) 164. The LIDAR sensor(s) 164 may be used for object and pedestrian detection, emergency braking, collision avoidance, and/or other functions. The LIDAR sensor(s) 164 may be functional safety level ASIL B. In some examples, the vehicle 100 may include multiple LIDAR sensors 164 (e.g., two, four, six, etc.) that may use Ethernet (e.g., to provide data to a Gigabit Ethernet switch).
In some examples, the LIDAR sensor(s) 164 may be capable of providing a list of objects and their distances for a 360-degree field of view. Commercially available LIDAR sensor(s) 164 may have an advertised range of approximately 100 m, with an accuracy of 2 cm-3 cm, and with support for a 100 Mbps Ethernet connection, for example. In some examples, one or more non-protruding LIDAR sensors 164 may be used. In such examples, the LIDAR sensor(s) 164 may be implemented as a small device that may be embedded into the front, rear, sides, and/or corners of the vehicle 100. The LIDAR sensor(s) 164, in such examples, may provide up to a 120-degree horizontal and 35-degree vertical field-of-view, with a 200 m range even for low-reflectivity objects. Front-mounted LIDAR sensor(s) 164 may be configured for a horizontal field of view between 45 degrees and 135 degrees.
In some examples, LIDAR technologies, such as 3D flash LIDAR, may also be used. 3D Flash LIDAR uses a flash of a laser as a transmission source, to illuminate vehicle surroundings up to approximately 200 m. A flash LIDAR unit includes a receptor, which records the laser pulse transit time and the reflected light on each pixel, which in turn corresponds to the range from the vehicle to the objects. Flash LIDAR may allow for highly accurate and distortion-free images of the surroundings to be generated with every laser flash. In some examples, four flash LIDAR sensors may be deployed, one at each side of the vehicle 100. Available 3D flash LIDAR systems include a solid-state 3D staring array LIDAR camera with no moving parts other than a fan (e.g., a non-scanning LIDAR device). The flash LIDAR device may use a 5 nanosecond class I (eye-safe) laser pulse per frame and may capture the reflected laser light in the form of 3D range point clouds and co-registered intensity data. By using flash LIDAR, and because flash LIDAR is a solid-state device with no moving parts, the LIDAR sensor(s) 164 may be less susceptible to motion blur, vibration, and/or shock.
The vehicle may further include IMU sensor(s) 166. The IMU sensor(s) 166 may be located at a center of the rear axle of the vehicle 100, in some examples. The IMU sensor(s) 166 may include, for example and without limitation, an accelerometer(s), a magnetometer(s), a gyroscope(s), a magnetic compass(es), and/or other sensor types. In some examples, such as in six-axis applications, the IMU sensor(s) 166 may include accelerometers and gyroscopes, while in nine-axis applications, the IMU sensor(s) 166 may include accelerometers, gyroscopes, and magnetometers.
In some embodiments, the IMU sensor(s) 166 may be implemented as a miniature, high performance GPS-Aided Inertial Navigation System (GPS/INS) that combines micro-electro-mechanical systems (MEMS) inertial sensors, a high-sensitivity GPS receiver, and advanced Kalman filtering algorithms to provide estimates of position, velocity, and attitude. As such, in some examples, the IMU sensor(s) 166 may enable the vehicle 100 to estimate heading without requiring input from a magnetic sensor by directly observing and correlating the changes in velocity from GPS to the IMU sensor(s) 166. In some examples, the IMU sensor(s) 166 and the GNSS sensor(s) 158 may be combined in a single integrated unit.
The vehicle may include microphone(s) 196 placed in and/or around the vehicle 100. The microphone(s) 196 may be used for emergency vehicle detection and identification, among other things.
The vehicle may further include any number of camera types, including stereo camera(s) 168, wide-view camera(s) 170, infrared camera(s) 172, surround camera(s) 174, long-range and/or mid-range camera(s) 198, and/or other camera types. The cameras may be used to capture image data around an entire periphery of the vehicle 100. The types of cameras used depends on the embodiments and requirements for the vehicle 100, and any combination of camera types may be used to provide the necessary coverage around the vehicle 100. In addition, the number of cameras may differ depending on the embodiment. For example, the vehicle may include six cameras, seven cameras, ten cameras, twelve cameras, and/or another number of cameras. The cameras may support, as an example and without limitation, Gigabit Multimedia Serial Link (GMSL) and/or Gigabit Ethernet. Each of the camera(s) is described with more detail herein with respect to
The vehicle 100 may further include vibration sensor(s) 142. The vibration sensor(s) 142 may measure vibrations of components of the vehicle, such as the axle(s). For example, changes in vibrations may indicate a change in road surfaces. In another example, when two or more vibration sensors 142 are used, the differences between the vibrations may be used to determine friction or slippage of the road surface (e.g., when the difference in vibration is between a power-driven axle and a freely rotating axle).
The vehicle 100 may include an ADAS system 138. The ADAS system 138 may include a SoC, in some examples. The ADAS system 138 may include autonomous/adaptive/automatic cruise control (ACC), cooperative adaptive cruise control (CACC), forward crash warning (FCW), automatic emergency braking (AEB), lane departure warnings (LDW), lane keep assist (LKA), blind spot warning (BSW), rear cross-traffic warning (RCTW), collision warning systems (CWS), lane centering (LC), and/or other features and functionality.
The ACC systems may use RADAR sensor(s) 160, LIDAR sensor(s) 164, and/or a camera(s). The ACC systems may include longitudinal ACC and/or lateral ACC. Longitudinal ACC monitors and controls the distance to the vehicle immediately ahead of the vehicle 100 and automatically adjust the vehicle speed to maintain a safe distance from vehicles ahead. Lateral ACC performs distance keeping, and advises the vehicle 100 to change lanes when necessary. Lateral ACC is related to other ADAS applications such as LCA and CWS.
CACC uses information from other vehicles that may be received via the network interface 124 and/or the wireless antenna(s) 126 from other vehicles via a wireless link, or indirectly, over a network connection (e.g., over the Internet). Direct links may be provided by a vehicle-to-vehicle (V2V) communication link, while indirect links may be infrastructure-to-vehicle (I2V) communication link. In general, the V2V communication concept provides information about the immediately preceding vehicles (e.g., vehicles immediately ahead of and in the same lane as the vehicle 100), while the I2V communication concept provides information about traffic further ahead. CACC systems may include either or both I2V and V2V information sources. Given the information of the vehicles ahead of the vehicle 100, CACC may be more reliable and it has potential to improve traffic flow smoothness and reduce congestion on the road.
FCW systems are designed to alert the driver to a hazard, so that the driver may take corrective action. FCW systems use a front-facing camera and/or RADAR sensor(s) 160, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to driver feedback, such as a display, speaker, and/or vibrating component. FCW systems may provide a warning, such as in the form of a sound, visual warning, vibration and/or a quick brake pulse. AEB systems detect an impending forward collision with another vehicle or other object, and may automatically apply the brakes if the driver does not take corrective action within a specified time or distance parameter. AEB systems may use front-facing camera(s) and/or RADAR sensor(s) 160, coupled to a dedicated processor, DSP, FPGA, and/or ASIC. When the AEB system detects a hazard, it typically first alerts the driver to take corrective action to avoid the collision and, if the driver does not take corrective action, the AEB system may automatically apply the brakes in an effort to prevent, or at least mitigate, the impact of the predicted collision. AEB systems, may include techniques such as dynamic brake support and/or crash imminent braking.
LDW systems provide visual, audible, and/or tactile warnings, such as steering wheel or seat vibrations, to alert the driver when the vehicle 100 crosses lane markings. A LDW system does not activate when the driver indicates an intentional lane departure, by activating a turn signal. LDW systems may use front-side facing cameras, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to driver feedback, such as a display, speaker, and/or vibrating component.
LKA systems are a variation of LDW systems. LKA systems provide steering input or braking to correct the vehicle 100 if the vehicle 100 starts to exit the lane.
BSW systems detects and warn the driver of vehicles in an automobile's blind spot. BSW systems may provide a visual, audible, and/or tactile alert to indicate that merging or changing lanes is unsafe. The system may provide an additional warning when the driver uses a turn signal. BSW systems may use rear-side facing camera(s) and/or RADAR sensor(s) 160, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to driver feedback, such as a display, speaker, and/or vibrating component.
RCTW systems may provide visual, audible, and/or tactile notification when an object is detected outside the rear-camera range when the vehicle 100 is backing up. Some RCTW systems include AEB to ensure that the vehicle brakes are applied to avoid a crash. RCTW systems may use one or more rear-facing RADAR sensor(s) 160, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to driver feedback, such as a display, speaker, and/or vibrating component.
Conventional ADAS systems may be prone to false positive results which may be annoying and distracting to a driver, but typically are not catastrophic, because the ADAS systems alert the driver and allow the driver to decide whether a safety condition truly exists and act accordingly. However, in an autonomous vehicle 100, the vehicle 100 itself must, in the case of conflicting results, decide whether to heed the result from a primary computer or a secondary computer (e.g., a first controller 136 or a second controller 136). For example, in some embodiments, the ADAS system 138 may be a backup and/or secondary computer for providing perception information to a backup computer rationality module. The backup computer rationality monitor may run a redundant diverse software on hardware components to detect faults in perception and dynamic driving tasks. Outputs from the ADAS system 138 may be provided to a supervisory MCU. If outputs from the primary computer and the secondary computer conflict, the supervisory MCU must determine how to reconcile the conflict to ensure safe operation.
In some examples, the primary computer may be configured to provide the supervisory MCU with a confidence score, indicating the primary computer's confidence in the chosen result. If the confidence score exceeds a threshold, the supervisory MCU may follow the primary computer's direction, regardless of whether the secondary computer provides a conflicting or inconsistent result. Where the confidence score does not meet the threshold, and where the primary and secondary computer indicate different results (e.g., the conflict), the supervisory MCU may arbitrate between the computers to determine the appropriate outcome.
The supervisory MCU may be configured to run a neural network(s) that is trained and configured to determine, based on outputs from the primary computer and the secondary computer, conditions under which the secondary computer provides false alarms. Thus, the neural network(s) in the supervisory MCU may learn when the secondary computer's output may be trusted, and when it cannot. For example, when the secondary computer is a RADAR-based FCW system, a neural network(s) in the supervisory MCU may learn when the FCW system is identifying metallic objects that are not, in fact, hazards, such as a drainage grate or manhole cover that triggers an alarm. Similarly, when the secondary computer is a camera-based LDW system, a neural network in the supervisory MCU may learn to override the LDW when bicyclists or pedestrians are present and a lane departure is, in fact, the safest maneuver. In embodiments that include a neural network(s) running on the supervisory MCU, the supervisory MCU may include at least one of a DLA or GPU suitable for running the neural network(s) with associated memory. In preferred embodiments, the supervisory MCU may comprise and/or be included as a component of the SoC(s) 104.
In other examples, ADAS system 138 may include a secondary computer that performs ADAS functionality using traditional rules of computer vision. As such, the secondary computer may use classic computer vision rules (if-then), and the presence of a neural network(s) in the supervisory MCU may improve reliability, safety and performance. For example, the diverse implementation and intentional non-identity makes the overall system more fault-tolerant, especially to faults caused by software (or software-hardware interface) functionality. For example, if there is a software bug or error in the software running on the primary computer, and the non-identical software code running on the secondary computer provides the same overall result, the supervisory MCU may have greater confidence that the overall result is correct, and the bug in software or hardware on primary computer is not causing material error.
In some examples, the output of the ADAS system 138 may be fed into the primary computer's perception block and/or the primary computer's dynamic driving task block. For example, if the ADAS system 138 indicates a forward crash warning due to an object immediately ahead, the perception block may use this information when identifying objects. In other examples, the secondary computer may have its own neural network which is trained and thus reduces the risk of false positives, as described herein.
The vehicle 100 may further include the infotainment SoC 130 (e.g., an in-vehicle infotainment system (IVI)). Although illustrated and described as a SoC, the infotainment system may not be a SoC, and may include two or more discrete components. The infotainment SoC 130 may include a combination of hardware and software that may be used to provide audio (e.g., music, a personal digital assistant, navigational instructions, news, radio, etc.), video (e.g., TV, movies, streaming, etc.), phone (e.g., hands-free calling), network connectivity (e.g., LTE, Wi-Fi, etc.), and/or information services (e.g., navigation systems, rear-parking assistance, a radio data system, vehicle related information such as fuel level, total distance covered, brake fuel level, oil level, door open/close, air filter information, etc.) to the vehicle 100. For example, the infotainment SoC 130 may radios, disk players, navigation systems, video players, USB and Bluetooth connectivity, carputers, in-car entertainment, Wi-Fi, steering wheel audio controls, hands free voice control, a heads-up display (HUD), an HMI display 134, a telematics device, a control panel (e.g., for controlling and/or interacting with various components, features, and/or systems), and/or other components. The infotainment SoC 130 may further be used to provide information (e.g., visual and/or audible) to a user(s) of the vehicle, such as information from the ADAS system 138, autonomous driving information such as planned vehicle maneuvers, trajectories, surrounding environment information (e.g., intersection information, vehicle information, road information, etc.), and/or other information.
The infotainment SoC 130 may include GPU functionality. The infotainment SoC 130 may communicate over the bus 102 (e.g., CAN bus, Ethernet, etc.) with other devices, systems, and/or components of the vehicle 100. In some examples, the infotainment SoC 130 may be coupled to a supervisory MCU such that the GPU of the infotainment system may perform some self-driving functions in the event that the primary controller(s) 136 (e.g., the primary and/or backup computers of the vehicle 100) fail. In such an example, the infotainment SoC 130 may put the vehicle 100 into a chauffeur to safe stop mode, as described herein.
The vehicle 100 may further include an instrument cluster 132 (e.g., a digital dash, an electronic instrument cluster, a digital instrument panel, etc.). The instrument cluster 132 may include a controller and/or supercomputer (e.g., a discrete controller or supercomputer). The instrument cluster 132 may include a set of instrumentation such as a speedometer, fuel level, oil pressure, tachometer, odometer, turn indicators, gearshift position indicator, seat belt warning light(s), parking-brake warning light(s), engine-malfunction light(s), airbag (SRS) system information, lighting controls, safety system controls, navigation information, etc. In some examples, information may be displayed and/or shared among the infotainment SoC 130 and the instrument cluster 132. In other words, the instrument cluster 132 may be included as part of the infotainment SoC 130, or vice versa.
The server(s) 178 may receive, over the network(s) 190 and from the vehicles, image data representative of images showing unexpected or changed road conditions, such as recently commenced road-work. The server(s) 178 may transmit, over the network(s) 190 and to the vehicles, neural networks 192, updated neural networks 192, and/or map information 194, including information regarding traffic and road conditions. The updates to the map information 194 may include updates for the HD map 122, such as information regarding construction sites, potholes, detours, flooding, and/or other obstructions. In some examples, the neural networks 192, the updated neural networks 192, and/or the map information 194 may have resulted from new training and/or experiences represented in data received from any number of vehicles in the environment, and/or based on training performed at a datacenter (e.g., using the server(s) 178 and/or other servers).
The server(s) 178 may be used to train machine learning models (e.g., neural networks) based on training data. The training data may be generated by the vehicles, and/or may be generated in a simulation (e.g., using a game engine). In some examples, the training data is tagged (e.g., where the neural network benefits from supervised learning) and/or undergoes other pre-processing, while in other examples the training data is not tagged and/or pre-processed (e.g., where the neural network does not require supervised learning). Training may be executed according to any one or more classes of machine learning techniques, including, without limitation, classes such as: supervised training, semi-supervised training, unsupervised training, self-learning, reinforcement learning, federated learning, transfer learning, feature learning (including principal component and cluster analyses), multi-linear subspace learning, manifold learning, representation learning (including spare dictionary learning), rule-based machine learning, anomaly detection, and any variants or combinations therefor. Once the machine learning models are trained, the machine learning models may be used by the vehicles (e.g., transmitted to the vehicles over the network(s) 190, and/or the machine learning models may be used by the server(s) 178 to remotely monitor the vehicles.
In some examples, the server(s) 178 may receive data from the vehicles and apply the data to up-to-date real-time neural networks for real-time intelligent inferencing. The server(s) 178 may include deep-learning supercomputers and/or dedicated AI computers powered by GPU(s) 184, such as a DGX and DGX Station machines developed by NVIDIA. However, in some examples, the server(s) 178 may include deep learning infrastructure that use only CPU-powered datacenters.
The deep-learning infrastructure of the server(s) 178 may be capable of fast, real-time inferencing, and may use that capability to evaluate and verify the health of the processors, software, and/or associated hardware in the vehicle 100. For example, the deep-learning infrastructure may receive periodic updates from the vehicle 100, such as a sequence of images and/or objects that the vehicle 100 has located in that sequence of images (e.g., via computer vision and/or other machine learning object classification techniques). The deep-learning infrastructure may run its own neural network to identify the objects and compare them with the objects identified by the vehicle 100 and, if the results do not match and the infrastructure concludes that the AI in the vehicle 100 is malfunctioning, the server(s) 178 may transmit a signal to the vehicle 100 instructing a fail-safe computer of the vehicle 100 to assume control, notify the passengers, and complete a safe parking maneuver.
For inferencing, the server(s) 178 may include the GPU(s) 184 and one or more programmable inference accelerators (e.g., NVIDIA's TensorRT). The combination of GPU-powered servers and inference acceleration may make real-time responsiveness possible. In other examples, such as where performance is less critical, servers powered by CPUs, FPGAs, and other processors may be used for inferencing.
Although the various blocks of
The interconnect system 202 may represent one or more links or busses, such as an address bus, a data bus, a control bus, or a combination thereof. The interconnect system 202 may include one or more bus or link types, such as an industry standard architecture (ISA) bus, an extended industry standard architecture (EISA) bus, a video electronics standards association (VESA) bus, a peripheral component interconnect (PCI) bus, a peripheral component interconnect express (PCIe) bus, and/or another type of bus or link. In some embodiments, there are direct connections between components. As an example, the CPU 206 may be directly connected to the memory 204. Further, the CPU 206 may be directly connected to the GPU 208. Where there is direct, or point-to-point connection between components, the interconnect system 202 may include a PCIe link to carry out the connection. In these examples, a PCI bus need not be included in the computing device 200.
The memory 204 may include any of a variety of computer-readable media. The computer-readable media may be any available media that may be accessed by the computing device 200. The computer-readable media may include both volatile and nonvolatile media, and removable and non-removable media. By way of example, and not limitation, the computer-readable media may comprise computer-storage media and communication media.
The computer-storage media may include both volatile and nonvolatile media and/or removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules, and/or other data types. For example, the memory 204 may store computer-readable instructions (e.g., that represent a program(s) and/or a program element(s), such as an operating system. Computer-storage media may include, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which may be used to store the desired information and which may be accessed by computing device 200. As used herein, computer storage media does not comprise signals per se.
The computer storage media may embody computer-readable instructions, data structures, program modules, and/or other data types in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media. The term “modulated data signal” may refer to a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal. By way of example, and not limitation, the computer storage media may include wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared and other wireless media. Combinations of any of the above should also be included within the scope of computer-readable media.
The CPU(s) 206 may be configured to execute at least some of the computer-readable instructions to control one or more components of the computing device 200 to perform one or more of the methods and/or processes described herein. The CPU(s) 206 may each include one or more cores (e.g., one, two, four, eight, twenty-eight, seventy-two, etc.) that are capable of handling a multitude of software threads simultaneously. The CPU(s) 206 may include any type of processor, and may include different types of processors depending on the type of computing device 200 implemented (e.g., processors with fewer cores for mobile devices and processors with more cores for servers). For example, depending on the type of computing device 200, the processor may be an Advanced RISC Machines (ARM) processor implemented using Reduced Instruction Set Computing (RISC) or an x86 processor implemented using Complex Instruction Set Computing (CISC). The computing device 200 may include one or more CPUs 206 in addition to one or more microprocessors or supplementary co-processors, such as math co-processors.
In addition to or alternatively from the CPU(s) 206, the GPU(s) 208 may be configured to execute at least some of the computer-readable instructions to control one or more components of the computing device 200 to perform one or more of the methods and/or processes described herein. One or more of the GPU(s) 208 may be an integrated GPU (e.g., with one or more of the CPU(s) 206 and/or one or more of the GPU(s) 208 may be a discrete GPU. In embodiments, one or more of the GPU(s) 208 may be a coprocessor of one or more of the CPU(s) 206. The GPU(s) 208 may be used by the computing device 200 to render graphics (e.g., 3D graphics) or perform general purpose computations. For example, the GPU(s) 208 may be used for General-Purpose computing on GPUs (GPGPU). The GPU(s) 208 may include hundreds or thousands of cores that are capable of handling hundreds or thousands of software threads simultaneously. The GPU(s) 208 may generate pixel data for output images in response to rendering commands (e.g., rendering commands from the CPU(s) 206 received via a host interface). The GPU(s) 208 may include graphics memory, such as display memory, for storing pixel data or any other suitable data, such as GPGPU data. The display memory may be included as part of the memory 204. The GPU(s) 208 may include two or more GPUs operating in parallel (e.g., via a link). The link may directly connect the GPUs (e.g., using NVLINK) or may connect the GPUs through a switch (e.g., using NVSwitch). When combined together, each GPU 208 may generate pixel data or GPGPU data for different portions of an output or for different outputs (e.g., a first GPU for a first image and a second GPU for a second image). Each GPU may include its own memory, or may share memory with other GPUs.
In addition to or alternatively from the CPU(s) 206 and/or the GPU(s) 208, the logic unit(s) 220 may be configured to execute at least some of the computer-readable instructions to control one or more components of the computing device 200 to perform one or more of the methods and/or processes described herein. In embodiments, the CPU(s) 206, the GPU(s) 208, and/or the logic unit(s) 220 may discretely or jointly perform any combination of the methods, processes and/or portions thereof. One or more of the logic units 220 may be part of and/or integrated in one or more of the CPU(s) 206 and/or the GPU(s) 208 and/or one or more of the logic units 220 may be discrete components or otherwise external to the CPU(s) 206 and/or the GPU(s) 208. In embodiments, one or more of the logic units 220 may be a coprocessor of one or more of the CPU(s) 206 and/or one or more of the GPU(s) 208.
Examples of the logic unit(s) 220 include one or more processing cores and/or components thereof, such as Data Processing Units (DPUs), Tensor Cores (TCs), Tensor Processing Units (TPUs), Pixel Visual Cores (PVCs), Vision Processing Units (VPUs), Graphics Processing Clusters (GPCs), Texture Processing Clusters (TPCs), Streaming Multiprocessors (SMs), Tree Traversal Units (TTUs), Artificial Intelligence Accelerators (AIAs), Deep Learning Accelerators (DLAs), Arithmetic-Logic Units (ALUs), Application-Specific Integrated Circuits (ASICs), Floating Point Units (FPUs), input/output (I/O) elements, peripheral component interconnect (PCI) or peripheral component interconnect express (PCIe) elements, and/or the like.
The communication interface 210 may include one or more receivers, transmitters, and/or transceivers that enable the computing device 200 to communicate with other computing devices via an electronic communication network, included wired and/or wireless communications. The communication interface 210 may include components and functionality to enable communication over any of a number of different networks, such as wireless networks (e.g., Wi-Fi, Z-Wave, Bluetooth, Bluetooth LE, ZigBee, etc.), wired networks (e.g., communicating over Ethernet or InfiniBand), low-power wide-area networks (e.g., LoRaWAN, SigFox, etc.), and/or the Internet. In one or more embodiments, logic unit(s) 220 and/or communication interface 210 may include one or more data processing units (DPUs) to transmit data received over a network and/or through interconnect system 202 directly to (e.g., a memory of) one or more GPU(s) 208.
The I/O ports 212 may enable the computing device 200 to be logically coupled to other devices including the I/O components 214, the presentation component(s) 218, and/or other components, some of which may be built in to (e.g., integrated in) the computing device 200. Illustrative I/O components 214 include a microphone, mouse, keyboard, joystick, game pad, game controller, satellite dish, scanner, printer, wireless device, etc. The I/O components 214 may provide a natural user interface (NUI) that processes air gestures, voice, or other physiological inputs generated by a user. In some instances, inputs may be transmitted to an appropriate network element for further processing. An NUI may implement any combination of speech recognition, stylus recognition, facial recognition, biometric recognition, gesture recognition both on screen and adjacent to the screen, air gestures, head and eye tracking, and touch recognition (as described in more detail below) associated with a display of the computing device 200. The computing device 200 may be include depth cameras, such as stereoscopic camera systems, infrared camera systems, RGB camera systems, touchscreen technology, and combinations of these, for gesture detection and recognition. Additionally, the computing device 200 may include accelerometers or gyroscopes (e.g., as part of an inertia measurement unit (IMU)) that enable detection of motion. In some examples, the output of the accelerometers or gyroscopes may be used by the computing device 200 to render immersive augmented reality or virtual reality.
The power supply 216 may include a hard-wired power supply, a battery power supply, or a combination thereof. The power supply 216 may provide power to the computing device 200 to enable the components of the computing device 200 to operate.
The presentation component(s) 218 may include a display (e.g., a monitor, a touch screen, a television screen, a heads-up-display (HUD), other display types, or a combination thereof), speakers, and/or other presentation components. The presentation component(s) 218 may receive data from other components (e.g., the GPU(s) 208, the CPU(s) 206, DPUs, etc.), and output the data (e.g., as an image, video, sound, etc.).
Referring to
The sensor data may allow detection of, or otherwise allow for identification/monitoring of, an object in the multi-sensor system's environment. For autonomous or semi-autonomous perception systems, the object may be a potential obstacle or other element that can impact/affect an aspect of safety and/or operation. The object may be unmoving (e.g., a tree, post office box, or building) or moving (e.g., a bicycle, another vehicle, or a pedestrian in motion). At block A304, method 300A includes generating a predicted state for the object detected in the environment of the multi-sensor system. For example, the perception system may generate a prediction regarding a subsequent location of the detected or identified object relative to an autonomous or semi-autonomous vehicle or machine (e.g., that includes/uses the multi-sensor system). In some embodiments, the predicted state may be represented, for example, as a relative position (e.g., a certain distance from an origin of a rig coordinate system of the multi-sensor system in a certain direction) and/or an absolute location (e.g., coordinates such as those from a global positioning system (GPS) device). In some implementations, the predicted state may indicate the subsequent “absence” of the object within the multi-sensor system's perceptible environment.
At blocks A306 and A308 of method 300A, a neural network model can be deployed. At block A306, inputs may be provided to the neural network model. The inputs may be, or may include, a set of input parameters and sensor data. The input parameters can indicate, define or represent the predicted state, and the sensor data can include, or can be based on, measurements from one or more sensors of the multi-sensor system. In some embodiments, the input parameters provided to the neural network model can be, or can include, the predicted state and an identification of a sensor or a set of sensors that provided the sensor data and/or sensor measurements which may or may not be used (e.g., by an autonomous or semi-autonomous perception system) to obtain the predicted state.
At block A308, method 300A includes obtaining a score indicative of an association between the predicted state and the sensor data. In some embodiments, the score is an association cost (which may be referred to as an “association score” in this disclosure). At block 310, method 300A includes updating data that defines, describes, or otherwise represents the environment (e.g., the current relative location of moving or unmoving objects, and the trajectories, speed, or velocities of moving objects). An update to the data representing the multi-sensor system's environment may indicate, for example, a change to a characteristic/property/aspect of the detected object. The changed characteristic/property/aspect can correspond to (in relative or absolute terms) position, trajectory, speed, velocity, outline (e.g., shape), and/or other characteristics/properties/aspects of the object.
The updated data on the environment may be used in a subsequent operation (not depicted in
In various embodiments, the neural network model is a multi-layer perceptron (MLP) model. In various embodiments, the MLP model is a feed-forward artificial neural network (ANN) that includes, for example, an input layer, one or more hidden layers, and an output layer. The MLP model may be trained using backpropagation learning, employing various supervised machine learning techniques. Beneficially, MLPs are capable of finding solutions to problems that are not linearly separable.
Advantageously, an MLP-based model can provide a suitable balance between accuracy and computational efficiency. Neural network models with more complex architectures (e.g., with a greater number of layers and artificial neurons) can potentially provide more robust association functions that tend to more often provide more accurate association costs (e.g., association costs that are more often consistent with the ground truth), but more complex neural network models can be more computationally intensive, slowing down the ability of the multi-sensor system to understand and react to changes in its environment. An MLP model can provide a desirable balance between computational demand and accuracy in predictions. In other embodiments, other neural network architectures may be employed.
Block B302 of method 300B includes providing sensor data, and predicted states based on the sensor data, to association cost functions to obtain association scores. In various embodiments, the association cost functions of block B302 can be association cost functions that do not employ machine learning models. The sensor data can be, or can include, a set of measurements from a respective group or subset of sensors of the multi-sensor system. The measurements can relate to the predicted states according to a correspondence between timestamps of the measurements and timestamps of the predicted states. The measurements can be deemed to correspond to predicted states by determining that they are contemporaneous with each other (e.g., by matching timestamps).
At block B304, method 300B includes comparing association scores (from the association cost functions of block B302) with ground truth (“GT” in
At block B306, method 300B includes matching the ground truth with the closest sensor object (e.g., an object sensed/detected by a radar or camera that is closest to the indicated ground truth) for cases that are true positives. At block B308, method 300B can include obtaining “positive” modeling data using the true positive cases. At block B310, method 300B can include obtaining “negative” modeling data. At block B312, the positive and negative modeling data can be used as training data and/or testing data to train or otherwise update the MLP or other neural network model. The neural network model can be trained or otherwise updated so that the model forms a complex association function between predictions and various sensor data, to be able to provide an association score between sensor measurements and predicted states.
In some embodiments, the generation of modeling data may comprise executing autonomous vehicle (AV) software/system (e.g., multi-sensor fusion system) on any given data such as recorded video session of the AV (or semi-autonomous or non-autonomous vehicle or machine) in any urban or freeway scenario, and saving the camera and radar measurements for each cycle. The multi-sensor fusion system's cycles can be based on camera timing, and radar data can be interpolated based on the camera cycles' timings. The input to the multi-sensor fusion system can be from multiple radars (e.g., 8 radars) and the multi-camera fusion module, and hence only one camera object. For each cycle, all detected camera objects (e.g., all objects detected by the camera perception system) and all the interpolated detected radar objects can be provided to an association function. Because multiple radar objects can be obtained for each real world object, there may be many more radar objects fed to the association function than camera objects. All of the defining camera and radar object parameters (such as velocity, position, shape, orientation, obstacle class, obstacle class confidence, etc.) can be provided to the association function. Along with the camera and radar objects, the internal predicted state of the different obstacle tracks may also be provided for each cycle. Also, fused objects formed at the end of each cycle, after prediction and update steps of the multi sensor fusion system, are saved as well. Similar to camera and radar measurement objects, the predicted state and fused objects similarly have parameters.
The saved measurement objects, predicted states, and fused objects can then be parsed and stored in accessible NumPy data objects. Human-annotated ground truth labels can be associated with the session that is run on the AV software. These human-annotated ground truth labels can contain bounding boxes (e.g., borders), classes (e.g., type of object), or other information for all obstacles that can be seen in certain frames in the given session. The manual annotation may be done at 1 frame per second (fps), such that certain intermediate frames would lack the human-annotated labels, and hence the data might not be sequential data. Once the human-annotated ground truth data is available, the saved and parsed fused objects for each timestamp for which there are annotations can be retrieved. For each timestamp, the fused objects with the ground truth data can be compared to determine the true positives and false positives. Each fused object may have a trackline identifier (ID) that is shared by all of its associated predicted states. Thus, for each TP/FP fused object, its corresponding predicted state for a given timestamp can be retrieved using the trackline identifier. All camera and radar objects detected by camera and radar systems, respectively, for that particular timestamp can also be retrieved. For a TP fused object, the associated camera/radar objects for that timestamp can be retrieved if both belong to the same obstacle class and are associated with fused objects (e.g., by the already-existing association function to be replaced by the neural network model to be trained). The ground truth associated with the TP fused object can be used to find the closest camera/radar object based on bounding shape distance. For each fused object, the process can involve storing the closest radar/camera object corresponding to the predicted state for the fused object for that timestamp with the same obstacle class and same trackline ID. These can act as positive samples for training the neural network model.
For generating the negative samples, any camera/radar object can be chosen and paired with any predicted state (ensuring the pair is not part of the positive samples described above), and the pairs can be labeled as negative samples. Any random measurement object and predicted state can act as a negative sample because they should not be associated.
Advantageously, in various embodiments, using the human annotated ground truth for finding the closest camera/radar object, instead of using the predicted state, enables finding the radar/camera objects that are closest to the ground truth and hence are the ideal measurement objects to be associated with. This way, the neural network is constrained/directed/forced to learn to choose such camera/radar objects even if other camera/radar objects are close to the given predicted state. This way, the neural network model is constrained/directed/forced to develop necessary complex functions to discern such proper measurement objects to associate with.
Because the above data creation process tends to yield many more negative samples as compared to positive samples for the neural network model, this skew can be addressed by designing appropriate loss functions suited to such a skew in the data.
The disclosed embodiment of the neural network model outputs a confidence score between 0 and 1, providing confidence of whether a pair of predicted state and measurement objects are to be associated. In some embodiments, and as a non-limiting example, a simple four-layered MLP model includes N input variables (N/2 from predicted state, and N/2 from measurement object), with 64, 32, 16 and 1 neurons in the 4 layers, respectively. The first 3 layers can be followed by Rectified Linear Unit (ReLU) activation functions, which sets all negative values to zero while retaining positive values as-is, and the last layer can include or operate as a sigmoid activation function which converts the range of output to [0,1] hence giving the confidence score.
The input variables for the MLP can be any of the parameters of the predicted state and measurement objects. Any N-dimensional input can be provided to the MLP, and the MLP can be able to determine the most important input parameters and attach weights to the parameters accordingly while weighing down not-so-important parameters. To reduce time complexity, the input can be restricted to, for example, 20 dimensions, with 10 parameters from predicted state and the other 10 parameters from measurement objects (e.g., camera/radar). In some embodiments, obstacle velocity (x axis and y axis), obstacle position (x axis and y axis), obstacle shape (height, width, length), obstacle class confidence, and obstacle orientation (x axis and y axis) from both predicted states and measurement objects can be used (resulting in 20 dimensions).
In some embodiments, the input that is assimilated can comprise different parameters with different ranges. For example, obstacle class confidence can range between 0 and 1, while velocity, position, and shape can be real floating values which can have any value. Similarly, orientation can be a unit vector with negative and positive values. With such a varied range of inputs, training the network to get the best possible association function can be more difficult, as the higher values of some input variables can lead to the network giving more significance to the input variables, even after multiplying with weights, and the resulting activation value of the MLP layers can still be high, influencing the confidence score. Various embodiments of the systems and methods can thus employ normalization on the inputs. For example, potential normalization techniques include Gaussian normalization and hyperbolic tangent (tanh) estimators. Both of these normalization techniques can involve first computing the mean and standard deviation of each of the 20 input dimensions across the training dataset for example, resulting in 20 mean values and 20 standard deviation values. In the Gaussian normalization method, each input variable can be treated as a Gaussian distribution and normalized as (xik−meank)/stdk, where xik is one of the input variables from a training sample. Here, i indicates the training sample number and k is the input variable number and hence the corresponding meank and stdk. In tanh estimators, 0.5(tanh(0.01(xik−meank)/stdk)+1) can be employed. Both these methods can set the mean of each input variable dimension to zero (after normalization). Tanh estimators can ensure that ranges of all the input variable dimensions are restricted to [−1, 1], whereas there is no such certainty with Gaussian normalization. For this reason, tanh estimators may be preferable in various embodiments.
With the MLP model designed and input assimilated and normalized, the next focus can be the loss function. Because the MLP model performs binary classification (e.g., “yes” to associate, or “no” not to associate), the process can use binary cross-entropy loss. Binary cross-entropy loss is represented by −ylog(p)+(1−y)log(1−p), where y is the ground truth (0 or 1) and p is the value predicted by the MLP model.
As suggested above, because the generated data may be more skewed towards negative samples, the MLP training may focus on negative samples more and hence suffer in classifying positive samples accurately. Accordingly, some embodiments of the systems and methods can use focal loss to reduce the loss for “easy” samples (usually negative samples when there are more negative samples in the dataset) and increase the loss for “harder” samples. The focal loss can be defined as −y(1−p)y log(p)−(1−y)(p): log(1−p). Here y is a penalty term which weighs down the loss for easy samples and weighs up the loss for harder training samples.
Once the MLP model is trained and tested, the model can be integrated into the AV software corresponding to the perception system. The existing camera and radar association functions can be replaced with the implemented MLP model. The MLP model weights and biases can be stored as matrices in an electronic file. The input variables can be assimilated from the predicted state data structure and the measurement objects data structure, and can be stored in a matrix. Matrix multiplication of the input and MLP weights can be performed and biases can be added. The ReLu activation function can be implemented and performed on the output of each layer to obtain the confidence score from the MLP. This confidence score can be used as a substitute for the association score computed using existing methods. The computed association score/cost can then be used by downstream algorithms to perform association and updates. The efficacy of the implemented MLP model-based association score computation solution can be verified by computing different standard KPIs and measuring the improvement in performance.
Compared to earlier attempted solutions, embodiments of the systems and methods can remove the need for manual iteration of determining the best distance metrics and their corresponding weights for the association functions of different sensors. Developer time spent in manually determining the best set of parameters, something that would be repeated every time there are any changes in dependencies, is reduced or eliminated. Various embodiments of the systems and methods are also capable of developing more complicated distance metrics in a data-driven way than could be generated in previous attempted solutions. The neural network model is allowed to learn complicated functions in a data-driven way to perform the task of association, and the neural network model can generate complex nonlinear functions that could not feasibly be designed and implemented manually. The approach of the systems and methods thus can provide data of varying scenarios and can use a stronger network architecture to provide complicated nonlinear functions that find correlations between varied input parameters. Moreover, the association cost/score generated by embodiments of the disclosed neural network model can be more interpretable, as it can be bound between 0 and 1 (as opposed to earlier attempted solutions where association costs are real numbers that could have any range). Further, the disclosed approach can reduce or remove the need for multiple association functions as used in earlier approaches. A single MLP model can be employed for all sensors, greatly reducing system complexity. The disclosed approach can enable accurate association of predicted states to measure obstacles for autonomous vehicle products for example.
As shown in
In at least one embodiment, grouped computing resources 414 may include separate groupings of node C.R.s 416 housed within one or more racks (not shown), or many racks housed in data centers at various geographical locations (also not shown). Separate groupings of node C.R.s 416 within grouped computing resources 414 may include grouped compute, network, memory or storage resources that may be configured or allocated to support one or more workloads. In at least one embodiment, several node C.R.s 416 including CPUs, GPUs, DPUs, and/or other processors may be grouped within one or more racks to provide compute resources to support one or more workloads. The one or more racks may also include any number of power modules, cooling modules, and/or network switches, in any combination.
The resource orchestrator 412 may configure or otherwise control one or more node C.R.s 416(1)-416(N) and/or grouped computing resources 414. In at least one embodiment, resource orchestrator 412 may include a software design infrastructure (SDI) management entity for the data center 400. The resource orchestrator 412 may include hardware, software, or some combination thereof.
In at least one embodiment, as shown in
In at least one embodiment, software 432 included in software layer 430 may include software used by at least portions of node C.R.s 416(1)-416(N), grouped computing resources 414, and/or distributed file system 438 of framework layer 420. One or more types of software may include, but are not limited to, Internet web page search software, e-mail virus scan software, database software, and streaming video content software.
In at least one embodiment, application(s) 442 included in application layer 440 may include one or more types of applications used by at least portions of node C.R.s 416(1)-416(N), grouped computing resources 414, and/or distributed file system 438 of framework layer 420. One or more types of applications may include, but are not limited to, any number of a genomics application, a cognitive compute, and a machine learning application, including training or inferencing software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.), and/or other machine learning applications used in conjunction with one or more embodiments.
In at least one embodiment, any of configuration manager 434, resource manager 436, and resource orchestrator 412 may implement any number and type of self-modifying actions based on any amount and type of data acquired in any technically feasible fashion. Self-modifying actions may relieve a data center operator of data center 400 from making possibly bad configuration decisions and possibly avoiding underutilized and/or poor performing portions of a data center.
The data center 400 may include tools, services, software or other resources to train one or more machine learning models or predict or infer information using one or more machine learning models according to one or more embodiments described herein. As discussed herein, the models may be neural network models, employing DNNs such as MLPs. For example, a machine learning model(s) may be trained by calculating weight parameters according to a neural network architecture using software and/or computing resources described above with respect to the data center 400. In at least one embodiment, trained or deployed machine learning models corresponding to one or more neural networks may be used to infer or predict information using resources described above with respect to the data center 400 by using weight parameters calculated through one or more training techniques, such as but not limited to those described herein.
In at least one embodiment, the data center 400 may use CPUs, application-specific integrated circuits (ASICs), GPUs, FPGAs, and/or other hardware (or virtual compute resources corresponding thereto) to perform training and/or inferencing using above-described resources. Moreover, one or more software and/or hardware resources described above may be configured as a service to allow users to train or performing inferencing of information, such as image recognition, speech recognition, or other artificial intelligence services.
Network environments suitable for use in implementing embodiments of the disclosure may include one or more client devices, servers, network attached storage (NAS), other backend devices, and/or other device types. The client devices, servers, and/or other device types (e.g., each device) may be implemented on one or more instances of the computing device(s) 200 of
Components of a network environment may communicate with each other via a network(s), which may be wired, wireless, or both. The network may include multiple networks, or a network of networks. By way of example, the network may include one or more Wide Area Networks (WANs), one or more Local Area Networks (LANs), one or more public networks such as the Internet and/or a public switched telephone network (PSTN), and/or one or more private networks. Where the network includes a wireless telecommunications network, components such as a base station, a communications tower, or even access points (as well as other components) may provide wireless connectivity.
Compatible network environments may include one or more peer-to-peer network environments—in which case a server may not be included in a network environment—and one or more client-server network environments—in which case one or more servers may be included in a network environment. In peer-to-peer network environments, functionality described herein with respect to a server(s) may be implemented on any number of client devices.
In at least one embodiment, a network environment may include one or more cloud-based network environments, a distributed computing environment, a combination thereof, etc. A cloud-based network environment may include a framework layer, a job scheduler, a resource manager, and a distributed file system implemented on one or more of servers, which may include one or more core network servers and/or edge servers. A framework layer may include a framework to support software of a software layer and/or one or more application(s) of an application layer. The software or application(s) may respectively include web-based service software or applications. In embodiments, one or more of the client devices may use the web-based service software or applications (e.g., by accessing the service software and/or applications via one or more application programming interfaces (APIs)). The framework layer may be, but is not limited to, a type of free and open-source software web application framework such as that may use a distributed file system for large-scale data processing (e.g., “big data”).
A cloud-based network environment may provide cloud computing and/or cloud storage that carries out any combination of computing and/or data storage functions described herein (or one or more portions thereof). Any of these various functions may be distributed over multiple locations from central or core servers (e.g., of one or more data centers that may be distributed across a state, a region, a country, the globe, etc.). If a connection to a user (e.g., a client device) is relatively close to an edge server(s), a core server(s) may designate at least a portion of the functionality to the edge server(s). A cloud-based network environment may be private (e.g., limited to a single organization), may be public (e.g., available to many organizations), and/or a combination thereof (e.g., a hybrid cloud environment).
The client device(s) may include at least some of the components, features, and functionality of the example computing device(s) 200 described herein with respect to
The disclosure may be described in the general context of computer code or machine-useable instructions, including computer-executable instructions such as program modules, being executed by a computer or other machine, such as a personal data assistant or other handheld device. Generally, program modules including routines, programs, objects, components, data structures, etc., refer to code that perform particular tasks or implement particular abstract data types. The disclosure may be practiced in a variety of system configurations, including hand-held devices, consumer electronics, general-purpose computers, more specialty computing devices, etc. The disclosure may also be practiced in distributed computing environments where tasks are performed by remote-processing devices that are linked through a communications network.
As used herein, a recitation of “and/or” with respect to two or more elements should be interpreted to mean only one element, or a combination of elements. For example, “element A, element B, and/or element C” may include only element A, only element B, only element C, element A and element B, element A and element C, element B and element C, or elements A, B, and C. In addition, “at least one of element A or element B” may include at least one of element A, at least one of element B, or at least one of element A and at least one of element B. Further, “at least one of element A and element B” may include at least one of element A, at least one of element B, or at least one of element A and at least one of element B.
The subject matter of the present disclosure is described with specificity herein to meet statutory requirements. However, the description itself is not intended to limit the scope of this disclosure. Rather, the inventors have contemplated that the claimed subject matter might also be embodied in other ways, to include different steps or combinations of steps similar to the ones described in this document, in conjunction with other present or future technologies. Moreover, although the terms “step” and/or “block” may be used herein to connote different elements of methods employed, the terms should not be interpreted as implying any particular order among or between various steps herein disclosed unless and except when the order of individual steps is explicitly described.