Driving policies determination

Abstract

A method for determining driving policies for a vehicle includes: receiving, in real-time during a trip of a vehicle, at least a set of input multimedia content elements captured by at least one sensor deployed in proximity to the vehicle, determining a plurality of possible future scenarios based at least on the set of input multimedia content elements, determining a probability score for each of the plurality of possible future scenarios, determining at least one driving policy according to at least the probability score for at least one of the plurality of possible future scenarios, and controlling the vehicle according to the at least one driving policy.

Description

TECHNICAL FIELD

The present disclosure relates generally to vehicle driving systems, and more particularly to determining driving policies based on analysis of multimedia content.

BACKGROUND

In part due to improvements in computer processing power and in location-based tracking systems such as global positioning systems, autonomous and other assisted driving systems have been developed with the aim of providing driverless control or driver-assisted control of vehicles while in movement. An autonomous vehicle includes a system for controlling the vehicle based on the surrounding environment such that the system autonomously controls functions such as acceleration, braking, steering, and the like.

Existing solutions for autonomous driving may use a global positioning system receiver, electronic maps, and the like, to determine a path from one location to another. Understandably, fatalities and injuries due to vehicles colliding with people or obstacles during the determined path are significant concerns for developers of autonomous driving systems. To this end, autonomous driving systems may utilize sensors such as cameras and radar for detecting objects to be avoided.

BRIEF DESCRIPTION OF THE DRAWINGS

The subject matter that is disclosed herein is particularly pointed out and distinctly claimed in the claims at the conclusion of the specification. The foregoing and other objects, features, and advantages of the disclosed embodiments will be apparent from the following detailed description taken in conjunction with the accompanying drawings.

FIG. 1 is a network diagram utilized to describe the various disclosed embodiments;

FIG. 2 is a schematic diagram of a driving policies' generator according to an embodiment;

FIG. 3 is a flowchart illustrating a method for generating driving policies based on multimedia content elements according to an embodiment;

FIG. 4 is a block diagram depicting the basic flow of information in the signature generator system of FIG. 1; and

FIG. 5 is a diagram showing a flow of patches generation, response vector generation, and signature generation in a large-scale speech-to-text system.

DESCRIPTION OF EXAMPLE EMBODIMENTS

Overview

A method for determining driving policies for a vehicle includes: receiving, in real-time during a trip of a vehicle, at least a set of input multimedia content elements captured by at least one sensor deployed in proximity to the vehicle, determining a plurality of possible future scenarios based at least on the set of input multimedia content elements, determining a probability score for each of the plurality of possible future scenarios, determining at least one driving policy according to at least the probability score for at least one of the plurality of possible future scenarios, and controlling the vehicle according to the at least one driving policy.

Description

It will be appreciated that as autonomous vehicles are introduced they will have to co-exist in the near future with non-autonomous vehicles, and even among autonomous vehicles, additional safety precautions may be necessary before they are safe enough for reliable, general use.

For example, some existing autonomous driving solutions engage in automatic or otherwise autonomous braking to avoid and/or minimize damage from collisions. However, such solutions may face challenges in accurately identifying obstacles. Moreover, such solutions typically employ a configured setting for braking or deceleration in response to the detection of a potential collision event. Such a setting does not necessarily account for the continued inertia and/or changes in movement of the obstacle being avoided. As a result, the automatically braking vehicle may stop unnecessarily quickly or conversely may not stop quickly enough.

Accordingly, in practice such solutions may not successfully prevent the collision event from occurring. In fact, in some cases the solutions may actually be a contributary factor in causing a collision, even if not necessarily the initially detected potential collision event. For example, stopping quickly to avoid an obstacle in front of the car may result in a rear-end collision caused by a vehicle behind the autonomously braking vehicle.

It will therefore be appreciated that such solutions typically suffer from a lack of proper consideration for the overall context for the detection of a potential collision event. It would therefore be advantageous to provide a solution for accurately detecting obstacles and determining possible scenarios respective thereof.

Reference is now made to FIG. 1 which is an exemplary network diagram 100 utilized to describe the various embodiments disclosed herein. Network diagram 100 comprises a driving control system (DCS) 120, a driving policies generator 130, a database 150, and at least one sensor 160, communicatively connected via a network 110. Network 110 may be implemented using one or more of the Internet, the world-wide-web (WWW), a local area network (LAN), a wide area network (WAN), a metro area network (MAN), and/or any other network(s) capable of enabling communication between the elements of network diagram 100.

Driving control system 120 is configured to control the vehicle (not shown), or at least a function thereof, either autonomously or at least semi-autonomously subject to driving policies determined by driving policies generator 130 in real-time during a trip of the vehicle based on sensor signals captured by sensors 160 deployed in proximity to the vehicle. While FIG. 1 depicts a single driving control system 120, it will be appreciated that in practice the overall functionality of driving control system 120 may be provided by a multiplicity of such systems. For example, driving control system 120 may interface with one or more of the vehicle's electronic control units (ECUs) to control the vehicle. Alternatively, in some implementations one or more of the vehicle's ECUs may be configured to function as additional driving control systems 120 that may communicate directly with driving policies generator 130.

In accordance with an exemplary embodiment, one, some, or all of driving control system 120, driving policies generator 130, sensor(s) 160 may be disposed in or affixed to the vehicle. It will be appreciated that the trip includes movement of the vehicle from at least a start location to a destination location. During the trip, at least visual multimedia content elements are captured by the sensors 160.

At least one of sensors 160 is configured to capture visual multimedia content elements demonstrating characteristics of at least a portion of the environment (e.g., roads, obstacles, etc.) surrounding the vehicle. In an example implementation, sensors 160 may include one or more cameras installed on a portion of a vehicle such as, but not limited to, a dashboard of the vehicle, a hood of the vehicle, a rear window of the vehicle, and the like. The visual multimedia content elements may include images, videos, and the like. The sensors 160 may be integrated in or communicatively connected to the driving control system 120 without departing from the scope of the disclosed embodiments.

In accordance with some embodiments described herein, one or more other sensors 160 may be configured to capture audio signals in the vicinity of the vehicle. For example, sensors 160 may include one or more microphones mounted in one or more locations on the vehicle's exterior. The microphone(s) may be disposed to capture ambient sounds in the driving environment, e.g., honking, shouting, brakes screeching, etc. Alternatively, or in addition, a microphone may also be located within the vehicle and disposed to capture sounds from, for example, passenger conversations, cellphone conversations, the vehicle's radio, and/or and an entertainment system (either as an integrated component of the vehicle and/or associated with an electronic device carried by one of the passengers).

Alternatively, or in addition, sensors 160 may be at least in part implemented using a vehicle's embedded electronic control units (ECUs). For example, ECUs controlling one or more of the vehicle's doors, windows, engine, power steering, seats, speed, telematics, transmission, brakes, battery, may also be leveraged to provide input regarding an environmental state for the vehicle. It will be appreciated that such ECUs may typically be connected to an onboard computer in the vehicle. In accordance with some embodiments described herein, driving control system 120 and/or driving policies generator 130 may be connected to, or even integrated as part of, the vehicle's onboard computer, such that no additional infrastructure may be necessary to receive sensor input from the vehicle's ECUs.

Alternatively, or in addition, one or more of sensors 160 may be implemented as a service, providing vehicle environment data from sensors operated independently of the vehicle. For example, a camera installed at a key intersection may provide “look ahead” images, or at least an angle or view not available to a camera mounted in or on the vehicle. In another example, the service may provide information regarding weather, traffic, and/or other environmental conditions.

Driving control system 120 may comprise an application 125 which may be implemented in software, hardware, firmware, or a combination thereof. Application 125 may be configured to send multimedia content elements and/or other environmental information captured by sensors 160 to driving policies' generator 130, and to receive policies therefrom. Application 125 may be further configured to receive collision avoidance instructions to be executed by driving control system 120 from driving policies' generator 130.

Database 150 may store a plurality of scenarios and associated potential causes and effects of collisions, probability scores, collision risk parameters, automatic braking instructions, and/or a combination thereof. Potential causes may include, but are not limited to, moving objects (e.g., other vehicles, pedestrians, animals, etc.), static objects (e.g., parked cars, buildings, boardwalks, trees, bodies of water, etc.), and/or current environmental conditions (rain, temperature, etc.). Each scenario represents a future possible series of events that leads to one or more outcomes. For example, given a pedestrian walking in proximity to an intersection in front of a vehicle in motion, one scenario may be that the pedestrian will cross the intersection, and one possible outcome may be a collision between the pedestrian and the vehicle. Another possible outcome may be that the pedestrian runs across the intersection and there is no collision. It will be appreciated that the possible outcomes listed herein are exemplary; in practice there may be other results as well.

A probability score for each of the plurality of scenarios may be stored in database 150. In accordance with embodiments described herein, the probability score may be calculated based on analysis of actual observations of the scenarios as they have occurred. For example, traffic videos may be used to identify scenarios and then to use them to populate database 150. The actual observed outcomes for each of the identified scenarios may be analyzed to calculate a probability score for each possible outcome. For example, given X occurrences observed for a given scenario with possible outcomes A and B: if outcome A is observed Y times, then its probability score may be calculated as Y/X; if outcome B is observed Z times, then its probability score may be calculated as Z/X, etc. Alternatively, or in addition, some or all of the probability scores may be input to database 150 manually. It will be appreciated that the embodiments described herein may support other methods of determining probability scores and populating database 150.

Using the abovementioned example, the speed and route of both the vehicle and the pedestrian, may be used to determine a probability score for each of the possible outcomes, where a risk score may represent the likelihood of a worst-case outcome, e.g., a collision between them the vehicle and the pedestrian. It will be appreciated that in scenarios there may be more than one undesirable outcome. For example, in a given scenario there may be a probability score reflecting the probability that the vehicle may make contact with a pedestrian or a second vehicle. In the same scenario there may also be a probability score reflecting the probability that the vehicle may hit a telephone pole when swerving to avoid the pedestrian or vehicle. In such a scenario, the risk score may be calculated to represent a combined, weighted probability for either undesirable outcome to occur.

As depicted in FIG. 1, network diagram 110 also comprises signature generator system (SGS) 140 and deep-content classification (DCC) system 170. In accordance with embodiments described herein SGS 140 and DCC system 170 are connected to the network 110 and may be utilized by the driving policies generator 130 to perform various embodiments described herein. It will be appreciated that each of SGS 140 and DCC system 170 may be connected to the Driving Policies Generator 130 either directly or through the network 110. In certain configurations, SGS 140, DCC system 170, and/or both may alternatively be embedded in driving policies generator 130.

SGS 140 is configured to generate signatures for multimedia content elements and comprises a plurality of computational cores, each computational core having properties that are at least partially statistically independent of each other core, where the properties of each core are set independently of the properties of each other core. Generation of signatures by the signature generator system is described further herein below with respect to FIGS. 4 and 5.

Deep content classification system 170 is configured to create, automatically and in an unsupervised fashion, concepts for a wide variety of multimedia content elements. To this end, deep content classification system 170 may be configured to inter-match patterns between signatures for a plurality of multimedia content elements and to cluster the signatures based on the inter-matching. Deep content classification system 170 may be further configured to generate reduced signature clusters by reducing the number of signatures in a cluster to a minimum while maintaining matching and enabling generalization to new multimedia content elements. Metadata of the multimedia content elements is collected to form, together with each of the reduced signature clusters, a concept. An example deep content classification system is described further in U.S. Pat. No. 8,266,185, which is assigned to the common assignee, the contents of which are hereby incorporated by reference. It will be appreciated that the configuration of SGS 140 and DCC 170 in network diagram 110 may be exemplary; the embodiments described herein may also support the use of other methods/systems to identify/classify the input multimedia content elements and/or to associate them with the scenarios and probability scores stored in database 150.

In an embodiment, driving policies generator 130 may be configured to send the input multimedia content elements to signature generator system 140, to deep content classification system 170, or both. Additionally, driving policies generator 130 may also be configured to receive a plurality of signatures generated for the input multimedia content elements from signature generator system 140, to receive a plurality of signatures (e.g., signature reduced clusters) of concepts matched to the input multimedia content elements from deep content classification system 170, or both. Alternatively, or in addition, driving policies' generator 130 may be configured to generate the plurality of signatures, identify the plurality of signatures (e.g., by determining concepts associated with the signature reduced clusters matching the multimedia content element to be tagged), or a combination thereof.

It will be appreciated that each signature is associated with at least one concept. A concept may be understood as a description of the content for which the signature was generated. Each concept comprises a collection of signatures representing associated multimedia content elements and metadata describing the concept. As a non-limiting example, a ‘Superman concept’ is a signature-reduced cluster of signatures describing elements (such as multimedia elements) related to, e.g., a Superman cartoon, and a set of metadata providing textual representation of the Superman concept. As another example, metadata of a concept represented by the signature generated for a picture showing a bouquet of red roses may be “flowers”. In yet another example, metadata of a concept represented by the signature generated for a picture showing a bouquet of wilted roses may be “wilted flowers”.

In accordance with embodiments described herein, driving policies generator 130 may be configured to determine a context of the input multimedia content elements based on their signatures, associated concepts, and/or both. Determination of the context allows for contextually matching between a potential cause of collision according to the input multimedia content elements and a predetermined potential cause of collision shown in a reference multimedia content element. Determining contexts of multimedia content elements is described further in U.S. patent application Ser. No. 13/770,603, assigned to the common assignee, the contents of which are hereby incorporated by reference.

In accordance with embodiments described herein, driving policies generator 130 may be configured to obtain, in real-time, or at least near real-time, input multimedia content elements from driving control system 120 that are captured by sensors 160 during the trip. It will be appreciated that at least some of the input multimedia content elements may be visual multimedia content elements showing obstacles that may lead to one or more future scenarios based on the current vehicle movement and the characteristics of the obstacle. For each future scenario, driving policies generator 130 may generate a probability score indicating the probability of that scenario occurring. The probability score may be generated by comparing one or more historical scenarios to each of the future possible scenarios that may occur based at least on the vehicle's movement and the obstacle's expected behavior. The obstacle may be, for example, pedestrians, animals, other vehicles, etc. It will be appreciated that the embodiments described herein may also support the use of other environmental factors as input when generating probability scores. For example, some of the historical scenarios may be associated with poor visibility and/or unsafe road conditions. These additional environmental factors may also be factored into the generation of probability scores, e.g., the presence of fog and/or slippery road conditions may increase the likelihood of an undesirable outcome.

In accordance with embodiments described herein, driving policies generator 130 may be configured to determine whether any reference multimedia content element matches the input multimedia content elements and, if so, to determine a probability score accordingly, e.g., per a probability score stored in database 150 for a scenario associated with the matched reference multimedia content element. For example, if the matched reference multimedia content element is a puddle on the road, the probability scores may reflect the likelihood for each possible outcome (e.g., how likely is it that the car may skid, the engine may be flooded, there may be no effect, etc.) as per the stored probability scores in database 150.

In accordance with embodiments described herein, determining whether there is a matching reference multimedia content element includes generating at least one signature for each input multimedia content element and comparing the generated input multimedia content signatures to signatures of the reference multimedia content elements. It will be appreciated that driving policies generator 130 may be configured to send the input multimedia content elements to SGS 140, to deep content classification system 170, and/or both. Driving policies generator 130 may further be configured to receive the generated signatures (e.g., from SGS 140) and/or at least one concept matching the input multimedia content elements (e.g., from DCC system 170). It will also be appreciated that the functionality attributed herein to SGS 140 and DCC 170 may be exemplary; the embodiments described herein may also support the use of other methods/systems determining whether there is a matching reference multimedia content element.

Each reference multimedia content element in database 150 is a previously captured multimedia content element representing an obstacle or other potential future scenario associated therewith. Each reference multimedia content element may be associated with at least one probability score, a risk score (e.g., reflecting one or more undesirable outcomes with possible damage that may occur), predetermined collision avoidance instructions, and the like. The collision avoidance instructions include, but are not limited to, one or more instructions for controlling the vehicle, e.g., to avoid an accident due to colliding with one or more obstacles.

Each reference multimedia content element may further be associated with a portion of a vehicle so as to indicate the location on the vehicle from which the reference multimedia content element was captured. To this end, in some embodiments, a reference multimedia content element may only match an input multimedia content element if, in addition to any signature matching, the reference multimedia content element is associated with the same or a similar portion of the vehicle (e.g., a portion on the same side of the vehicle).

As a non-limiting example, input multimedia content elements showing a dog approaching the car from 5 feet away that were captured by a camera deployed on a hood of the car may only match a reference multimedia content element showing a dog approaching the car from approximately 5 feet away that was captured by a camera deployed on the hood or other area on the front side of the car. The same reference multimedia content element may not match the input multimedia content element if the input multimedia content element was captured from a camera facing the rear of the vehicle. However, it will be appreciated that there may be a scenario in database 150 where the vehicle is in reverse (i.e., moving backwards) with a dog approaching from the rear. Accordingly, if the input multimedia content element showing a dog approaching from the rear is accompanied by additional information indicating that the vehicle is in reverse, it may still match a reference multimedia content element in database 150.

In accordance with embodiments described herein, when a potential collision is detected, driving policies generator 130 may be configured to generate an alert. The alert may indicate the potential cause of collision shown in the input multimedia content elements (i.e., a potential cause of collision associated with the matching reference multimedia content element), the at least one collision parameter associated with the matching reference multimedia content element (as stored in database 150), or both. The alert may further include one or more collision avoidance instructions to be executed by driving control system 120 to avoid the collision. The collision avoidance instructions may be instructions for configuring one or more portions of the driving control system such as, but not limited to, a braking system, a steering system, and the like.

In accordance with embodiments described herein, driving policies generator 130 may be configured to send the generated alert and/or the collision avoidance instructions to the driving control system 120. Alternatively, driving policies generator 130 may include driving control system 120 in an integrated module and may be further configured to control the vehicle based on the collision avoidance instructions.

It will be appreciated that only one driving control system 120 and one application 125 are described herein above with reference to FIG. 1 for the sake of simplicity and without limitation on the disclosed embodiments. Multiple driving control systems 120 may provide multimedia content elements via multiple applications 125, and appropriate driving decisions may be provided to each such instance of driving control system 120, without departing from the scope of the disclosure.

It should be noted that any of driving control system 120, sensors 160, driving policies generator 130, and database 150 may be integrated without departing from the scope of the disclosure.

Reference is now made to FIG. 2 which is an example schematic diagram of driving policies generator 130 according to an embodiment. Driving policies generator 130 comprises processing circuitry 210 coupled to memory 220, storage 230, and network interlace 240. The components of driving policies' generator 130 may be communicatively connected via bus 250.

Processing circuitry 210 may be Instantiated as one or more hardware logic components and circuits. For example, and without limitation, illustrative types of hardware logic components that may be used include field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), Application-specific standard products (ASSPs), system-on-a-chip systems (SOCs), general-purpose microprocessors, microcontrollers, digital signal processors (DSPs), and the like, or any other hardware logic components capable of performing calculations and/or other manipulations of information. In accordance with embodiments described herein, processing circuitry 210 may be implemented as an array of at least partially statistically independent computational cores. The properties of each computational core may be set independently of those of each other core, as described further herein above.

Memory 220 may be volatile (e.g., RAM, etc.), non-volatile (e.g., ROM, flash memory, etc.), or a combination thereof. It will be appreciated that computer readable instructions to implement one or more embodiments disclosed herein may be stored in storage 230.

Memory 220 may also be configured to store software. Software shall be construed broadly to mean any type of instructions, whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise. Instructions may include code (e.g., in source code format, binary code format, executable code format, or any other suitable format of code). The instructions, when executed by processing circuitry 210, instruct the processing circuitry 210 to perform the various processes described herein. Specifically, the instructions, when executed, facilitate at least the generation by processing circuitry 210 of driving alerts based on multimedia content as described herein.

Storage 230 may be magnetic storage, optical storage, and the like, and may be implemented, for example, as flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVDs), or any other suitable medium which may be used to store the desired information.

Network interface 240 enables driving policies generator 130 to communicate with the signature generator system 140 and/or DCC 170 for the purpose of, for example, sending multimedia content elements, receiving signatures, and the like. Further, network interface 240 enables driving policies' generator 130 to obtain multimedia content elements from sensors 160 as well in order to determine a plurality of possible scenarios that may occur in light of the obstacles and the vehicle's movement.

It should be understood that the embodiments described herein are not limited to the specific architecture illustrated in FIG. 2, and other architectures may be equally used without departing from the scope of the disclosed embodiments. In particular, driving policies generator 130 may further include an integrated signature generator system functionally similar to SGS 140 that may be configured to generate signatures as described herein without departing from the scope of the disclosed embodiments.

Reference is now made to FIG. 3 which depicts a flowchart 300 illustrating an exemplary method to be performed by driving policies generator 130 to generate driving policies based on multimedia content according to embodiments described herein. For example, as discussed hereinabove, driving policies generator 130 may generate driving policies based on multimedia content elements captured by sensors 160 (e.g., a camera) deployed in proximity to a vehicle such that the sensor signals indicate at least some features of the environment around the vehicle. The multimedia content elements may be captured during a trip, where the trip includes movement of the vehicle from a beginning location to a destination location.

Driving policies generator 130 may receive (step S310), input multimedia content elements (MMCEs) during a vehicle's trip. As described herein, the input multimedia content elements may be captured by sensors 160 deployed in proximity to the vehicle and may be, for example, received from sensors 160 via driving control system 120 which may be communicatively connected to sensors 160. The input multimedia content elements may be received in real-time, or near real-time, during the trip, thereby facilitating real-time provision of alerts to autonomous or assisted driving systems.

At least some of the input multimedia content elements may comprise obstacles shown therein. Such obstacles may include, both moving objects (e.g., pedestrians, other vehicles, animals, etc.) or stationary objects (e.g., signs, bodies of water, parked vehicles, statues, buildings, walls, etc.).

SGS 140 generates (step S320) signatures of the input multimedia content elements. The generation of signatures by SGS 140 is further described herein below with respect of FIGS. 4 and 5.

In accordance with embodiments described herein, step S320 may include SGS 140 generating (or at least causing generation) of at least one signature for each input multimedia content element and comparing the generated signatures for the input multimedia content element to signatures of the plurality of reference multimedia content elements. The reference multimedia content element signatures may be generated and stored in database 150 in a preprocessing step and/or during previous iterations of flowchart 300.

In accordance with embodiments described herein, step S320 may include generating the signatures via a plurality of at least partially statistically independent computational cores, where the properties of each core are set independently of the properties of the other cores. In another embodiment, S320 includes sending the multimedia content element(s) to signature generator system 140, to deep content classification system 170, or both, and receiving the plurality of signatures in return. Signature generator system 140 may comprise a plurality of at least statistically independent computational cores as described further herein. Deep content classification system 170 may be configured to autonomously define concepts for a wide variety of multimedia content elements in an unsupervised fashion.

In accordance with embodiments described herein, step S320 may include driving policies generator 130 querying DCC system 170 using the generated signatures to identify at least one concept matching the input multimedia content elements. The metadata of the matching concept may be used for correlation between a first signature (e.g., from the input MMCE) and at least a second signature (e.g., from a reference MMCE).

Driving policies generator 130 may determine (step S330), a plurality of future possible scenarios based on matches between the input and reference multimedia content elements. It will be appreciated that the vehicle's movement and the identification of one or more obstacles that may potentially collide with the vehicle may be factored into such a determination.

Driving policies generator 130 may determine (step S340) a probability score for each future scenario. The probability score can be determined by comparing the future scenarios to historical scenarios stored in the database 150 and determining the respective probability and characteristics associated therewith. It will be appreciated that at least some of the probability scores stored in database 150 may be determined and stored either in a preprocessing step and/or in previous iterations of flowchart 300; i.e., it will be appreciated that database may be at least partially populated with probability scores prior to a given iteration of flowchart 300. Driving policies generator 130 may then determine (step S350) driving policies for the future scenarios in light of their associated probability scores. The determination of the driving policies is further described hereinabove with respect of FIG. 1.

If additional input multimedia content elements have been received (step S360), process control may return to step S310. Otherwise, execution of flowchart 300 may terminate. Accordingly, execution of flowchart 300 may continue until the trip is completed by, for example, arriving at the destination location, the vehicle stopping at or near the destination location, and the like.

As a non-limiting example, input video may be received from a dashboard camera mounted on a car and facing forward such that the dashboard camera captures video of the environment in front of the car showing a pedestrian entering a crosswalk. The captured input video is obtained in real-time and analyzed to generate signatures therefore. The generated signatures are compared to signatures of reference videos. Based on the comparison, a plurality of possible future scenarios may be determined. Then, a probability of a collision between the vehicle and the pedestrian may be determined in light of at least current/expected movement of the vehicle and the pedestrian. A driving policy may then be determined based on the determined probability of collision. As an example, in case of high probability, a driving policy may be to slow down while arriving close to the crosswalk. In the case of an autonomous vehicle, the driving policy may be to brake, thereby avoiding collision with the pedestrian.

Reference is now made to FIGS. 4 and 5 which together illustrate an exemplary implementation for the generation of signatures for the multimedia content elements by SGS 140. An exemplary high-level description of the process for large scale matching is depicted in FIG. 4. In this example, the matching is for video content.

Video content segments 2 from a master database (DB) 6 and a target DB 1 are processed in parallel by a large number of independent computational cores 3 that constitute an architecture for generating the signatures (hereinafter the “architecture”). Further details on the computational cores generation are provided below. The independent cores 3 generate a database 5 of robust signatures 4 for video content-segments 2 from target DB 1 and a database 8 of robust signatures 7 for video content-segments 2 from master DB 6. An exemplary and non-limiting process of signature generation for an audio component is shown in detail in FIG. 4. Finally, target robust signatures 4 and/or signatures are effectively matched, by a matching algorithm 9, to master robust signatures 7 and/or signatures database to find matches between the two databases.

To demonstrate an example of the signature generation process, it is assumed, merely for the sake of simplicity and without limitation on the generality of the disclosed embodiments, that the signatures are based on a single frame, leading to certain simplification of the computational cores generation. The matching system is extensible for signatures generation capturing the dynamics in-between the frames. In an embodiment the driving policies generator 130 may be configured with a plurality of computational cores to perform matching between signatures.

The signatures generation process is now described with reference to FIG. 5. The first step in the process of signatures generation from a given speech-segment is to breakdown the speech-segment to K patches 14 of random length P and random position within the speech segment 12. The breakdown is performed by the patch generator component 21. The value of the number of patches K, random length P and random position parameters is determined based on optimization, considering the tradeoff between accuracy rate and the number of fast matches required in the flow process of driving policies generator 130 and SGS 140. Thereafter, all the K patches are injected in parallel into multiple computational cores 3 to generate K response vectors 22, which are fed into a signature generator system 23 to produce a database of robust signatures and signatures 4.

In order to generate robust signatures, i.e., signatures that are robust to additive noise L (where L is an integer equal to or greater than 1) by the computational cores 3, a frame ‘i’ is injected into the cores 3. Then, cores 3 generate two binary response vectors: which is a signature vector, and {right arrow over (RS)} which is a robust signature vector.

For generation of signatures robust to additive noise, such as White-Gaussian-Noise, scratch, etc., but not robust to distortions typical in video elements, e.g., crop, shift and rotation, etc., a core Ci={n_i}(1≤i≤L) may consist of a single leaky integrate-to-threshold unit (LTU) node or more nodes. The node n_i equations are:

$V_i=\sum _j{w}_{ij}{k}_j$ $n_i=\theta \left(\mathrm{Vi}-{\mathrm{Th}}_x\right)$

where, θ is a Heaviside step function; w_ij is a coupling node unit (CNU) between node i and image component j (for example, grayscale value of a certain pixel j); kj is an image component T (for example, grayscale value of a certain pixel j); Th_x is a constant Threshold value, where ‘x’ is ‘S’ for signature and ‘RS’ for robust signature; and V_i is a Coupling Node Value.

The Threshold values Th_x are set differently for signature generation and for robust signature generation. For example, for a certain distribution of V_i values (for the set of nodes), the thresholds for signature (Th_S) and robust signature (Th^RS) are set apart, after optimization, according to at least one or more of the following criteria:

For: V_i>Th_RS

1−P(V>Th_S)−1−(1−e)¹<<1,

i.e., given that 1 nodes (cores) constitute a robust signature of a certain image I, the probability that not all of these 1 nodes will belong to the signature of same, but noisy image, p is sufficiently low (according to a system's specified accuracy).

2:1−P(V_i>Th_RS)≈1/L

i.e., approximately 1 out of the total L nodes can be found to generate a robust signature according to the above definition.

• • 3: Both robust signature and signature are generated for certain frame i.

It should be understood that even though the generation of a signature uses lossless compression, where the characteristics of the compressed data are maintained, the uncompressed data of the underlying multimedia content element may not be reconstructed; a generated signature may be unidirectional in that the lossless compression may be used on only a portion of the multimedia content element such that the entire element may not be reconstructed by merely reversing the lossless compression. However, the use of lossless compression facilitates the purpose of comparison to another signature without the need of comparison to the original data. The detailed description of the Signature generation can be found in U.S. Pat. Nos. 8,326,775 and 8,312,031, assigned to the common assignee, which are hereby incorporated by reference.

A computational core generation is a process of definition, selection, and tuning of the parameters of the cores for a certain realization in a specific system and application. The process may be based on several design considerations, such as:

• • (a) The cores are designed so as to obtain maximal independence, i.e., the projection from a signal space should generate a maximal pair-wise distance between any two cores' projections into a high-dimensional space.
  • (b) The cores are optimally designed for the type of signals, i.e., the cores are maximally sensitive to the spatio-temporal structure of the injected signal, for example, and in particular, sensitive to local correlations in time and space. Thus, in some cases a core represents a dynamic system, such as in state space, phase space, edge of chaos, etc., which may be uniquely used herein to exploit their maximal computational power.
  • (c) The cores are optimally designed with regard to invariance to a set of signal distortions, of interest in relevant applications.

A detailed description of computational core generation and the process for configuring such cores is discussed in more detail in U.S. Pat. No. 8,655,801 which is assigned to the common assignee, the contents of which are hereby incorporated by reference.

It is important to note that the embodiments disclosed herein are only examples of the many advantageous uses of the innovative teachings herein. In general, statements made in the specification of the present application do not necessarily limit any of the various claimed inventions. Moreover, some statements may apply to some inventive features but not to others. In general, unless otherwise indicated, singular elements may be in plural and vice versa with no loss of generality. In the drawings, like numerals refer to like parts through several views.

It should be noted that various embodiments described herein are discussed with respect to autonomous driving decisions and systems are merely for simplicity and without limitation on the disclosed embodiments. The embodiments disclosed herein are equally applicable to other assisted driving systems such as, for example, accident detection systems, lane change warning systems, and the like. In such example implementations, the automated driving decisions may be generated, e.g., only for specific driving events.

The various embodiments disclosed herein may be implemented as hardware, firmware, software, or any combination thereof. Moreover, the software is preferably implemented as an application program tangibly embodied on a program storage unit or computer readable medium consisting of parts, or of certain devices and/or a combination of devices. The application program may be uploaded to, and executed by, a machine comprising any suitable architecture. Preferably, the machine is implemented on a computer platform having hardware such as one or more central processing units (“CPUs”), a memory, and input/output interfaces. The computer platform may also include an operating system and microinstruction code. The various processes and functions described herein may be either part of the microinstruction code or part of the application program, or any combination thereof, which may be executed by a CPU, whether or not such a computer or processor is explicitly shown. In addition, various other peripheral units may be connected to the computer platform such as an additional data storage unit and a printing unit. Furthermore, a non-transitory computer readable medium is any computer readable medium except for a transitory propagating signal.

The term “non-transitory”, as used herein, is a limitation of the medium itself (i.e., tangible, not a signal) as opposed to a limitation on data storage persistency (e.g., RAM vs. ROM).

All examples and conditional language recited herein are intended for pedagogical purposes to aid the reader in understanding the disclosed embodiments and the concepts contributed by the inventor to furthering the art and are to be construed as being without limitation to such specifically recited examples and conditions. Moreover, all statements herein reciting principles, aspects, and embodiments of the invention, as well as specific examples thereof, are intended to encompass both structural and functional equivalents thereof. Additionally, it is intended that such equivalents include both currently known equivalents as well as equivalents developed in the future, i.e., any elements developed that perform the same function, regardless of structure.

It should be understood that any reference to an element herein using a designation such as “first,” “second,” and so forth does not generally limit the quantity or order of those elements. Rather, these designations are generally used herein as a convenient method of distinguishing between two or more elements or instances of an element. Thus, a reference to first and second elements does not mean that only two elements may be employed there or that the first element must precede the second element in some manner. Also, unless stated otherwise, a set of elements comprises one or more elements.

As used herein, the phrase “at least one of” followed by a listing of items means that any of the listed items can be utilized individually, or any combination of two or more of the listed items can be utilized. For example, if a system is described as including “at least one of A, B, and C,” the system can include A alone; B alone; C alone; A and B in combination; B and C in combination; A and C in combination; or A, B, and C in combination.

Claims

1. A method for determining driving policies for a vehicle, comprising: receiving, in real-time during a trip of a vehicle and during a current iteration of the method, at least a set of input multimedia content elements captured by at least one sensor deployed in proximity to the vehicle;determining, a plurality of possible future scenarios based at least on the set of input multimedia content elements; wherein at least two possible future scenarios of the plurality of possible future scenarios have undesirable outcomes;determining a probability score for each of the plurality of possible future scenarios to provide a plurality of probability scores;generating a risk score that represents a combined, weighted probability score of the at least two possible future scenarios;determining at least one driving policy, wherein the determining comprises selecting the at least one driving policy to decrease risk associated with the risk score; andcontrolling the vehicle according to the at least one driving policy.

2. The method of claim 1, further comprising: generating, at least one signature for each one multimedia content element of the set of input multimedia content elements, wherein the plurality of possible future scenarios is determined by at least matching the at least one signature to at least one reference signature associated with at least one of the plurality of future scenarios; wherein the at least one reference signature is generated during the previous iteration of the method.

3. The method of claim 2, wherein each of the at least one reference signature is associated with one or more concepts, wherein each concept of the one or more concepts is a collection of associated signatures and metadata.

4. The method of claim 1, wherein the set of multimedia content elements includes at least one of: an image; a video, or at least an audio signal.

5. The method of claim 1, wherein the least two possible future scenarios of the plurality of possible future scenarios comprise (a) a possible future scenario of colliding with an object captured in the set of input multimedia content elements have undesirable outcomes, and (b) colliding with another object when driving to avoid the colliding with the object.

6. The method of claim 1, wherein the determining a plurality of possible future scenarios is also based on an environmental state for the vehicle as indicated by non-image based information provided by at least one electronic control unit (ECU) in the vehicle.

7. The method of claim 6, wherein the at least one ECU controls at least one of the vehicle's doors, windows, engine, power steering, seats, speed, telematics, transmission, brakes, or battery.

8. The method of claim 1, wherein the determining a plurality of possible future scenarios is also based on an environmental state for the vehicle as indicated by information provided by a service external to the vehicle.

9. The method of claim 1, wherein the determining of a probability score of at least one possible future scenario is calculated during a previous iteration of the method; wherein the previous iteration of the method is executed during the trip of the vehicle.

10. The method according to claim 1 wherein the set of input multimedia content elements are indicative of a scenario and wherein the determining of the probability score for each of the plurality of possible future scenarios comprises comparing between numbers of occurrences of each of the plurality of possible future scenarios given the scenario.

11. A vehicle driving system, comprising: processing circuitry;at least one sensor operative to capture at least a set of input multimedia content elements in real-time during a trip of a vehicle;a driving policies generator application to be executed by the processing circuitry and operative to:determine, during a current iteration, a plurality of possible future driving scenarios based at least on the set of input multimedia content elements, wherein at least two possible future scenarios of the plurality of possible future scenarios have undesirable outcomes;determine a probability score for each of the plurality of possible future driving scenarios to provide a plurality of probability scores;generate a risk score that represents a combined, weighted probability score of the at least two possible future scenarios;anddetermine at least one driving policy by selecting the at least one driving policy to decrease risk associated with the risk score; anda driving control system operative to control the vehicle according to the at least one driving policy.

12. The system of claim 11, further comprising: a signature generating system to be executed by the processing circuitry to generate at least one signature for each multimedia content element of the set of input multimedia content elements, wherein the driving policies generator application is further operative to determine the plurality of possible future driving scenarios by at least matching the at least one signature to at least one reference signature associated with at least one of the plurality of future scenarios; and wherein the at least one reference signature is generated during the previous iteration of the method.

13. The system of claim 12, wherein each reference signature is associated one or more concepts, wherein each concept of the one or more concepts is a collection of associated signatures and metadata.

14. The system of claim 11, wherein each multimedia content element of the set of input multimedia content elements is at least one of: an image, or a video, or at least an audio signal.

15. The system of claim 11, wherein the driving policies generator application is also operative to determine the plurality of possible future driving scenarios based at least in part on an environmental state for the vehicle as indicated by non-image based information provided by at least one electronic control unit (ECU) in the vehicle.

16. The system of claim 15, wherein the at least one ECU controls at least one of the vehicle's doors, windows, engine, power steering, seats, speed, telematics, transmission, brakes, or battery.

17. The system of claim 11, wherein the driving policies generator application is also operative to determine the plurality of possible future driving scenarios based at least in part on an environmental state for the vehicle as indicated by information provided by a service external to the vehicle.

18. The system of claim 11, wherein the least two possible future scenarios of the plurality of possible future scenarios comprise (a) a possible future scenario of colliding with an object captured in the set of input multimedia content elements have undesirable outcomes, and (b) colliding with another object when driving to avoid the colliding with the object.

19. The system of claim 11, wherein the determining of a probability score of at least one possible future scenario is calculated during a previous iteration of the method; wherein the previous iteration of the method is executed during the trip of the vehicle.

20. A vehicle driving system comprising: means for receiving, in real-time during a trip of a vehicle, at least a set of input multimedia content elements captured by at least one sensor deployed in proximity to the vehicle;means for determining, during a current iteration, a plurality of possible future scenarios based at least on the set of input multimedia content elements; wherein at least two possible future scenarios of the plurality of possible future scenarios have undesirable outcomes;means for determining a probability score for each of the plurality of possible future scenarios to provide a plurality of probability scores;means for generating a risk score that represents a combined, weighted probability score of the at least two possible future scenarios;means for determining at least one driving policy wherein the determining comprises selecting the at least one driving policy to decrease risk associated with the risk score; andmeans for controlling the vehicle according to the at least one driving policy.