Object classification for autonomous navigation systems

Abstract

A navigational control method for a mobile automation apparatus includes: controlling a depth sensor to capture depth data representing a portion of a facility containing an obstacle; identifying the obstacle from the depth data; determining a probability that the obstacle is static; based on the probability, assigning the obstacle one of a dynamic class, a static class, and at least one intermediate class; updating a map to include a position of the obstacle, and the assigned class; and selecting, based on the assigned class, a navigational control action from a first action type associated with the dynamic class and the intermediate class, and a second action type associated with the static class.

Description

BACKGROUND

Autonomous or semi-autonomous systems, such as a mobile automation apparatus configured to travel a facility (e.g. a retail facility) to collect data therein, perform various navigational actions to detect and avoid obstacles within the facility. Different navigational actions may be taken according to the nature of a given obstacle. However, changes in an obstacle's state and/or imperfect sensor information may lead to inaccurate determination of an obstacle's nature, resulting in sub-optimal navigational behavior by the apparatus.

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

The accompanying figures, where like reference numerals refer to identical or functionally similar elements throughout the separate views, together with the detailed description below, are incorporated in and form part of the specification, and serve to further illustrate embodiments of concepts that include the claimed invention, and explain various principles and advantages of those embodiments.

FIG. 1 is a schematic of a mobile automation system.

FIG. 2 depicts a mobile automation apparatus in the system of FIG. 1.

FIG. 3 is a block diagram of certain internal hardware components of the mobile automation apparatus in the system of FIG. 1.

FIG. 4 is a flowchart of a method for obstacle classification at the apparatus of FIG. 1.

FIG. 5 is a diagram illustrating an example performance of the method of FIG. 4.

FIG. 6 is a diagram illustrating another example performance of the method of FIG. 4.

FIG. 7 is a diagram illustrating a further example performance of the method of FIG. 4.

FIG. 8 is a diagram illustrating another example performance of the method of FIG. 4.

FIG. 9 is a flowchart of a method of determining a probability at block 430 of the method of FIG. 4.

Skilled artisans will appreciate that elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale. For example, the dimensions of some of the elements in the figures may be exaggerated relative to other elements to help to improve understanding of embodiments of the present invention.

The apparatus and method components have been represented where appropriate by conventional symbols in the drawings, showing only those specific details that are pertinent to understanding the embodiments of the present invention so as not to obscure the disclosure with details that will be readily apparent to those of ordinary skill in the art having the benefit of the description herein.

DETAILED DESCRIPTION

Examples disclosed herein are directed to a navigational control method for a mobile automation apparatus, comprising: controlling a depth sensor to capture depth data representing a portion of a facility containing an obstacle; identifying the obstacle from the depth data; determining a probability that the obstacle is static; based on the probability, assigning the obstacle one of (i) a dynamic class, (ii) a static class, and (iii) at least one intermediate class; updating a map to include a position of the obstacle, and the assigned class; and selecting, based on the assigned class, a navigational control action from (i) a first action type associated with the dynamic class and the intermediate class, and (ii) a second action type associated with the static class.

Additional examples disclosed herein are directed to a computing device, comprising: a depth sensor; a locomotive assembly; and a navigational controller configured to: control the depth sensor to capture depth data representing a portion of a facility containing an obstacle; identify the obstacle from the depth data; determine a probability that the obstacle is static; based on the probability, assign the obstacle one of (i) a dynamic class, (ii) a static class, and (iii) at least one intermediate class; update a map to include a position of the obstacle, and the assigned class; and select, based on the assigned class, a navigational control action from (i) a first action type associated with the dynamic class and the intermediate class, and (ii) a second action type associated with the static class.

FIG. 1 depicts a mobile automation system 100 in accordance with the teachings of this disclosure. The system 100 includes a server 101 in communication with at least one mobile automation apparatus 103 (also referred to herein simply as the apparatus 103) and at least one client computing device 104 via communication links 105, illustrated in the present example as including wireless links. In the present example, the links 105 are provided by a wireless local area network (WLAN) deployed via one or more access points (not shown). In other examples, the server 101, the client device 104, or both, are located remotely (i.e. outside the environment in which the apparatus 103 is deployed), and the links 105 therefore include wide-area networks such as the Internet, mobile networks, and the like. The system 100 also includes a dock 106 for the apparatus 103 in the present example. The dock 106 is in communication with the server 101 via a link 107 that in the present example is a wired link. In other examples, however, the link 107 is a wireless link.

The client computing device 104 is illustrated in FIG. 1 as a mobile computing device, such as a tablet, smart phone or the like. In other examples, the client device 104 is implemented as another type of computing device, such as a desktop computer, a laptop computer, another server, a kiosk, a monitor, and the like. The system 100 can include a plurality of client devices 104 in communication with the server 101 via respective links 105.

The system 100 is deployed, in the illustrated example, in a retail facility including a plurality of support structures such as shelf modules 110-1, 110-2, 110-3 and so on (collectively referred to as shelf modules 110 or shelves 110, and generically referred to as a shelf module 110 or shelf 110—this nomenclature is also employed for other elements discussed herein). Each shelf module 110 supports a plurality of products 112 (also referred to as items), which may also be referred to as items. Each shelf module 110 includes a shelf back 116-1, 116-2, 116-3 and a support surface (e.g. support surface 117-3 as illustrated in FIG. 1) extending from the shelf back 116 to a shelf edge 118-1, 118-2, 118-3. A variety of other support structures may also be present in the facility, such as pegboards, tables, and the like.

The shelf modules 110 (also referred to as sub-regions of the facility) are typically arranged in a plurality of aisles (also referred to as regions of the facility), each of which includes a plurality of modules 110 aligned end-to-end. In such arrangements, the shelf edges 118 face into the aisles, through which customers in the retail facility, as well as the apparatus 103, may travel. As will be apparent from FIG. 1, the term “shelf edge” 118 as employed herein, which may also be referred to as the edge of a support surface (e.g., the support surfaces 117) refers to a surface bounded by adjacent surfaces having different angles of inclination. In the example illustrated in FIG. 1, the shelf edge 118-3 is at an angle of about ninety degrees relative to the support surface 117-3 and to the underside (not shown) of the support surface 117-3. In other examples, the angles between the shelf edge 118-3 and the adjacent surfaces, such as the support surface 117-3, is more or less than ninety degrees.

The apparatus 103 is equipped with a plurality of navigation and data capture sensors 108, such as image sensors (e.g. one or more digital cameras) and depth sensors (e.g. one or more Light Detection and Ranging (LIDAR) sensors, one or more depth cameras employing structured light patterns, such as infrared light, or the like). The apparatus 103 is deployed within the retail facility and, via communication with the server 101 and use of the sensors 108, navigates autonomously or partially autonomously along a length 119 of at least a portion of the shelves 110.

While navigating among the shelves 110, the apparatus 103 can capture images, depth measurements and the like, representing the shelves 110 and the items 112 supported by the shelves 110 (generally referred to as shelf data or captured data). Navigation may be performed according to a frame of reference 102 established within the retail facility. The apparatus 103 therefore tracks its pose (i.e. location and orientation) in the frame of reference 102. As will be discussed in greater detail below, the apparatus 103 also detects obstacles within the facility, and executes various navigational processes to traverse certain shelves 110 for imaging as mentioned above, while avoiding such obstacles. The specific navigational processes executed in response to the presence of an obstacle in the path of the apparatus 103 depends at least in part on various attributes of the obstacle.

Obstacles may include the shelves 110, customers within the facility, shopping carts or other objects, and the like. That is, certain obstacles may be mobile, while others may be stationary. Further, certain obstacles may be mobile at some times, and stationary at other times. A mobile (i.e. dynamic) obstacle in the path of the apparatus 103 may move out of the path after a period of time, and the apparatus 103 may therefore be configured, upon encountering a dynamic obstacle, to wait for a period of time with the expectation that the current path will be cleared of the obstacle. A stationary (i.e. static) obstacle, on the other hand, is likely to remain within the path of the apparatus 103. Therefore, the apparatus 103 may be configured, if faced with a static obstacle, to generate a new navigational path to avoid the obstacle. Generation of a new path incurs a computational cost, and may also slow the completion of a data collection task by the apparatus 103, or cause premature termination of the task, e.g. if a given module 110 is rendered inaccessible by a static obstacle.

In other words, the apparatus 103 is configured to proceed differently depending on the expected behavior of the obstacle. Accurately assessing expected obstacle behavior may be difficult, however, as a result of artifacts in sensor data, and changes in the nature of the obstacles themselves. A shopping cart, for example, may be static at some times and dynamic at other times.

The apparatus 103 is therefore configured, as discussed in detail below, to implement a probabilistic model for classifying obstacles, and to employ an obstacle classification system and resulting navigational controls to mitigate the costs in time and/or computational resources that can result from attempting to navigate around static obstacles.

The server 101 includes a special purpose controller, such as a processor 120, specifically designed to control and/or assist the mobile automation apparatus 103 to navigate the facility and to capture data. The processor 120 is interconnected with a non-transitory computer readable storage medium, such as a memory 122, having stored thereon computer readable instructions for performing various functionality, including control of the apparatus 103 to navigate the modules 110 and capture shelf data, as well as post-processing of the shelf data. The memory 122 can also store data for use in the above-mentioned control of the apparatus 103 and post-processing of captured data, such as a repository 123. The repository 123 can contain, for example, a map of the facility, operational constraints for use in controlling the apparatus 103, the image and/or depth data captured by the apparatus 103, and the like.

The memory 122 includes a combination of volatile memory (e.g. Random Access Memory or RAM) and non-volatile memory (e.g. read only memory or ROM, Electrically Erasable Programmable Read Only Memory or EEPROM, flash memory). The processor 120 and the memory 122 each comprise one or more integrated circuits. In some embodiments, the processor 120 is implemented as one or more central processing units (CPUs) and/or graphics processing units (GPUs).

The server 101 also includes a communications interface 124 interconnected with the processor 120. The communications interface 124 includes suitable hardware (e.g. transmitters, receivers, network interface controllers and the like) allowing the server 101 to communicate with other computing devices—particularly the apparatus 103, the client device 104 and the dock 106—via the links 105 and 107. The links 105 and 107 may be direct links, or links that traverse one or more networks, including both local and wide-area networks. The specific components of the communications interface 124 are selected based on the type of network or other links that the server 101 is required to communicate over. In the present example, as noted earlier, a wireless local-area network is implemented within the retail facility via the deployment of one or more wireless access points. The links 105 therefore include either or both wireless links between the apparatus 103 and the mobile device 104 and the above-mentioned access points, and a wired link (e.g. an Ethernet-based link) between the server 101 and the access point.

The processor 120 can therefore obtain data captured by the apparatus 103 via the communications interface 124 for storage (e.g. in the repository 123) and subsequent processing (e.g. to detect objects such as shelved products 112 in the captured data, and detect status information corresponding to the objects). The server 101 maintains, in the memory 122, an application 125 executable by the processor 120 to perform such subsequent processing.

The server 101 may also transmit status notifications (e.g. notifications indicating that products are out-of-stock, in low stock or misplaced) to the client device 104 responsive to the determination of product status data. The client device 104 includes one or more controllers (e.g. central processing units (CPUs) and/or field-programmable gate arrays (FPGAs) and the like) configured to process notifications and other information received from the server 101. For example, the client device 104 includes a display 132 controllable to present information received from the server 101.

Turning now to FIG. 2, the mobile automation apparatus 103 is shown in greater detail. The apparatus 103 includes a chassis 201 containing a locomotive assembly 203 (e.g. one or more electrical motors driving wheels, tracks or the like). The apparatus 103 further includes a sensor mast 205 supported on the chassis 201 and, in the present example, extending upwards (e.g., substantially vertically) from the chassis 201. The mast 205 supports the sensors 108 mentioned earlier. In particular, the sensors 108 include at least one imaging sensor 207, such as a digital camera. In the present example, the mast 205 supports seven digital cameras 207-1 through 207-7 oriented to face the shelves 110.

The mast 205 also supports at least one depth sensor 209, such as a 3D digital camera capable of capturing both depth data and image data. The apparatus 103 also includes additional depth sensors, such as LIDAR sensors 211. In the present example, the mast 205 supports two LIDAR sensors 211-1 and 211-2. As shown in FIG. 2, the cameras 207 and the LIDAR sensors 211 are arranged on one side of the mast 205, while the depth sensor 209 is arranged on a front of the mast 205. That is, the depth sensor 209 is forward-facing (i.e. captures data in the direction of travel of the apparatus 103), while the cameras 207 and LIDAR sensors 211 are side-facing (i.e. capture data alongside the apparatus 103, in a direction perpendicular to the direction of travel). In other examples, the apparatus 103 includes additional sensors, such as one or more RFID readers, temperature sensors, and the like.

The mast 205 also supports a plurality of illumination assemblies 213, configured to illuminate the fields of view of the respective cameras 207. The illumination assemblies 213 may be referred to collectively as an illumination subsystem. That is, the illumination assembly 213-1 illuminates the field of view of the camera 207-1, and so on. The cameras 207 and lidars 211 are oriented on the mast 205 such that the fields of view of the sensors each face a shelf 110 along the length 119 of which the apparatus 103 is traveling. The apparatus 103 is configured to track a pose of the apparatus 103 (e.g. a location and orientation of the center of the chassis 201) in the frame of reference 102, permitting data captured by the apparatus 103 to be registered to the frame of reference 102 for subsequent processing.

Turning to FIG. 3, certain internal components of the mobile automation apparatus 103 are shown. In particular, apparatus 103 includes a special-purpose navigational controller, such as a processor 220 interconnected with a non-transitory computer readable storage medium, such as a memory 222. The memory 222 includes a combination of volatile (e.g. Random Access Memory or RAM) and non-volatile memory (e.g. read only memory or ROM, Electrically Erasable Programmable Read Only Memory or EEPROM, flash memory). The processor 220 and the memory 222 each comprise one or more integrated circuits.

The apparatus 103 also includes a communications interface 224, containing suitable hardware components to enable communication between the apparatus 103 and other computing devices, including the server 101. The interface 224 may enable, for example, wired communications via the dock 106, wireless communications via the above-mentioned WLAN, or both. Via the interface 224, the apparatus 103 may receive instructions from the server 101 to travel to certain portions of the facility to collect image and/or depth data depicting the shelves 110.

The memory 222 stores computer readable instructions for execution by the processor 220. In particular, the memory 222 stores a navigation application 228 which, when executed by the processor 220, configures the processor 220 to perform various functions discussed below in greater detail and related to the navigation of the apparatus 103 (e.g. by controlling the locomotive mechanism 203). The application 228 may also be implemented as a suite of distinct applications in other examples.

The processor 220, when so configured by the execution of the application 228, may also be referred to as a navigational controller 220. Those skilled in the art will appreciate that the functionality implemented by the processor 220 via the execution of the application 228 may also be implemented by one or more specially designed hardware and firmware components, such as FPGAs, Application-Specific Integrated Circuits (ASICs) and the like in other embodiments.

The memory 222 may also store a repository 232 containing, for example, one or more maps of the environment in which the apparatus 103 operates, for use during the execution of the application 228. The repository 232, in the examples discussed below, contains an obstacle map containing locations and other information of obstacles detected by the apparatus 103. In some examples, the repository 232 may store a global map containing information defining the shelves 110 and other substantially permanent structures in the facility, and a separate obstacle map containing other obstacles detected by the apparatus 103, such as human customers or workers and other objects.

In the present example, the apparatus 103 is configured (via the execution of the application 228 by the processor 220) to generate navigational paths to travel through the environment, for example to reach goal locations provided by the server 101. The apparatus 103 is also configured to control the locomotive mechanism 203 to travel along the above-mentioned paths. To that end, the apparatus 103 is also configured, as will be discussed below in greater detail, to detect obstacles in the surroundings of the apparatus 103. The detected obstacles (at least those not appearing in the global map mentioned above, if a global map is used) are assigned classifications, and the apparatus 103 selects navigational control actions to avoid such obstacles based in part on the classifications.

As will be apparent in the discussion below, other examples, some or all of the processing performed by the apparatus 103 may be performed by the server 101, and some or all of the processing performed by the server 101 may be performed by the apparatus 103. That is, although in the illustrated example the application 228 resides in the mobile automation apparatus 103, in other embodiments the actions performed by the apparatus 103 via execution of the application 228 may be performed by the processor 120 of the server 101, either in conjunction with or independently from the processor 220 of the mobile automation apparatus 103. As those of skill in the art will realize, distribution of navigational computations between the server 101 and the mobile automation apparatus 103 may depend upon respective processing speeds of the processors 120 and 220, the quality and bandwidth of the link 105 between the apparatus 103 and the server 101, as well as criticality level of the underlying instruction(s).

The functionality of the application 228 will now be described in greater detail. In particular, the detection and classification of obstacles will be described as performed by the apparatus 103. FIG. 4 illustrates a method 400 of obstacle classification. The method 400 will be described in conjunction with its performance by the apparatus 103.

At block 405, the processor 220 is configured to operate the depth sensor(s) (e.g. the forward-facing depth sensor 209) to capture depth data representing a portion of the facility. In particular, the captured depth data defines any objects within a field of view of the sensor 209 or other suitable depth sensor.

At block 410, the processor 220 is configured to identify any obstacles represented in the depth data. A variety of obstacle identification mechanisms may be employed at block 410, examples of which will occur to those skilled in the art. For example, various edge detection processes, clustering processes and the like may be applied to determine whether the depth data indicates the presence of an obstacle in the field of view of the depth sensor 209.

The identification of an obstacle includes the determination of various attributes of the obstacle. For example, the processor 220 determines the position of the obstacle at block, e.g. within the frame of reference 102, based on the tracked pose of the apparatus 103 itself. The processor 220 can also determine one or more dimensions of the obstacle (e.g. a width). The identification of an obstacle can further include the determination of a velocity of the obstacle, if the processor 220 has previously identified that obstacle (i.e. in previous performances of block 405).

At block 415, having identified any obstacles in the captured depth data from block 405, the processor 220 is configured to retain or discard previously detected obstacles stored in the map mentioned earlier, based on the classification assigned to those obstacles. Classification, as well as retaining mechanisms, are discussed in greater detail below. In this example performance of block 415, it is assumed that no obstacles are currently stored in the map, and no action is therefore required at block 415.

The processor 220 is then configured to process each identified obstacle via the blocks within the dashed box in FIG. 4, beginning at block 420. In general, the processing of each obstacle includes determining a probability that the obstacle is static, and assigning a classification to the obstacle based on the probability. The assigned class defines which navigational actions the apparatus 103 takes when the obstacle interrupts the path being traveled by the apparatus 103. The assigned class also affects the retention mechanism employed by the processor 220 at block 415, to either retain an obstacle that is no longer within the field of view of the depths sensor 209, or discard (i.e. “forget”) the obstacle.

The classes assigned to obstacles by the processor 220 include two primary classes, and at least one intermediate class. The primary classes include a dynamic class, assigned to obstacles that are currently in motion or likely to begin moving, and a static class, assigned to obstacles that are not currently in motion and are not likely to begin moving. The intermediate classes include at least a transitional class, assigned to obstacles for which insufficient information has been collected to assign one of the primary classes. In the present example, the intermediate classes also include an initial class, assigned to obstacles for which no previous observations exist in the map stored in the repository 232.

The class assigned to an obstacle affects the navigational control of the apparatus 103 in various ways. For example, when an obstacle is in the path of the apparatus 103, the apparatus 103 may select a first type of control action, or a second type of control action. The first type of control action includes pausing travel along the path for a brief period of time (e.g. five seconds), with the expectation that the obstacle may move out of the path. The first type of control action is therefore suitable for execution in the presence of a dynamic obstacle, which is unlikely to remain in the path of the apparatus 103 for long. The first type of control action can also include adaptive adjustments to travel along the current path, such as reducing velocity along the path without necessarily pausing travel entirely.

The second type of control action includes generating an updated path to guide the apparatus 103 around or away from the obstacle. The second type of control action is therefore suitable for execution in the presence of a static obstacle, which is likely to remain in the path of the apparatus 103. The generation of a new path can be computationally intensive, and may result in a portion of the task assigned to the apparatus 103 remaining incomplete.

In addition to selecting control actions based on detected obstacles, the apparatus 103 selects retention mechanisms to apply to the map of obstacles. Previously observed obstacles are stored in a map, and when a current observation does not result in the identification of a matching obstacle, the previous observation may be discarded or retained, based on various factors including the class of the obstacle. For example, two types of retention mechanism may be applied to an obstacle. If the obstacle is classified as a dynamic obstacle, the processor 220 may discard the obstacle from the map if the obstacle is not currently observed. If, on the other hand, the obstacle is classified as a static obstacle, the processor 220 may retain the obstacle in the map, e.g. for a predefined period of time (e.g. thirty minutes, although other time periods may also be used).

The intermediate classes are distinguished from the primary classes in that navigational actions for intermediate-classed obstacles are of the same type as dynamic obstacles, while map retention actions for intermediate-classed obstacles are of the same type as static obstacles. Thus, for obstacles for which insufficient information is available to classify as static or dynamic, the apparatus 103 may avoid unnecessarily generating a new path, while retaining the obstacle in the map in the event that path regeneration is necessary in the future.

At block 420, the processor 220 is configured to determine whether the obstacle is a newly observed obstacle. As will be apparent to those skilled in the art, the processor 220 can compare the obstacles identified at block 410 to the map, and match currently observed obstacles with previously observed obstacles. If a currently observed obstacle does not have a match in the map, the determination at block 420 is affirmative. Following an affirmative determination at block 420, the processor 220 proceeds to block 425.

Turning briefly to FIG. 5, an overhead view of a facility 500 in which the apparatus 103 is deployed. The facility includes aisles defined by sets 504-1, 504-2, and 504-3 of modules 110. The apparatus 103 is deployed to travel along the sets 504 of modules and capture images and/or depth data depicting the modules 110. In the illustrated example, the apparatus 103 is configured to travel along a path 508, e.g. to capture images of a first side of the set 504-2, and then a second side of the set 504-2. A field of view 512 of the depth sensor 209 is illustrated as encompassing a portion of an obstacle 516. The obstacle 516 may be, for example, a pallet truck used by a worker in the facility to stock shelves in the sets 504 of modules 110.

Referring again to FIG. 4, it is assumed in this example that the map is empty, and the determination at block 420 is therefore affirmative. At block 425, the processor 220 therefore sets an initial probability that the obstacle 516 is static, and also assigns an initial classification to the obstacle 516. The probability set at block 425, in the present example, is zero, indicating that the obstacle 516 is not likely to be static, or stated another way, that the obstacle 516 is likely to be dynamic. The information available at the first observation of an obstacle may be insufficient to determine whether the obstacle is static or dynamic. However, an initial assumption that the obstacle is dynamic, as at block 425, reduces the likelihood of unnecessary path regeneration, as will be apparent in the discussion below. Further, assignment of the initial class, rather than a dynamic class, to the obstacle reflects uncertainty in the actual state of the obstacle and enables the apparatus 103 to respond to the presence of the obstacle 516 differently than either static or dynamic-classified obstacles, as will also be discussed below.

Having set the probability and assigned the initial class to the obstacle 516, the processor 220 is configured to update a map at block 455. Updating the map includes storing the position of the obstacle 516 (e.g. in the frame of reference 102), as well as the class of the obstacle 516, the probability associated with the obstacle 516, and a timestamp indicating the time of the most recent observation of the obstacle 516. Returning to FIG. 5, an example map 520 is illustrated, in which the position of the obstacle 516 is indicated (the sets 504 of modules 110 are omitted for simplicity). In association with the obstacle 516, the apparatus 103 stores the above-mentioned timestamp, probability P_static, class, and one or more attributes of the obstacle 516. In this example the attributes include an observed velocity and an observed size (e.g. width) of the obstacle. The attributes of the obstacle 516 can also include other velocity-related, size-related, or positional attributes. For example, an acceleration of the obstacle 516 may be stored. In further examples, a position of the obstacle 516 relative to a permanently mapped obstacle such as a shelf module may be stored as an attribute.

Returning to FIG. 4, at block 460, the processor 220 determines whether a navigational control action is required as a result of the obstacles in the map 520. For example, the processor 220 may determine whether any obstacles in the map 520 are on the path 508, and are within a predefined proximity of the apparatus 103.

When the determination at block 460 is affirmative, at block 465 the apparatus 103 selects between the first and second types of action mentioned above. Otherwise, the processor 220 returns to block 405 to capture the next set of depth data. For example, the processor 220 may be configured to control the depth sensor 209 to capture a set of depth data at a predefined frequency, e.g. 20 Hz, with the remainder of the method 400 being performed for each set.

In this example, it is assumed that the obstacle 516 is sufficient close to the apparatus 103 to result in an affirmative determination at block 460. At block 465, because the obstacle 516 has an intermediate classification, the processor 220 selects a navigational control action of the first type, associated with dynamic obstacles. That is, the processor 220 controls the apparatus 103 to pause travel along the path 508 for a predetermined period of time. Performance of the method 400 then returns to block 405.

It is assumed, for illustrative purposes, that following the next capture of depth data at block 405, the processor 220 identifies the obstacle 516, and the obstacle 516 as shown in the map 520 is therefore retained (i.e. no retention or discarding action is selected at block 415). At block 420, the determination is negative. At block 430, the processor 220 is configured to generate an updated probability that the obstacle 516 is static. The generation of updated probabilities is discussed in greater detail further below. In general, the probability is based at least on the observed velocity of the obstacle 516. The probability may also be based on other attributes of the obstacle, such as the observed size of the obstacle (e.g. a width), as well as the additional attributes mentioned earlier such as acceleration, positions relative to other obstacles, and the like. The probability is also based, in this example, on the previous probability (i.e. stored in association with the map 520).

Referring to FIG. 6, the obstacle 516 is shown having moved away from the apparatus 103 (which itself has not moved, as a result of the control action selected at the earlier performance of block 465). An updated probability that the obstacle 516 is static is generated at block 430. As shown in FIG. 5, the updated probability has a value of 0.3. The processor 220 is configured to compare the updated probability to an upper threshold (static threshold, or ST) and a lower threshold (dynamic threshold, or DT). A class is assigned to the obstacle 516 based on such comparisons. In particular, when the probability is below the lower threshold, the obstacle 516 is classified as dynamic at block 440. When the probability is between the upper and lower thresholds, the obstacle 516 is classified as a transitional obstacle at block 445. When the probability is above the upper threshold, the obstacle 516 is classified as static at block 450.

In this example, the lower threshold is set to 0.4 (i.e. a 40% probability that the obstacle 516 is static), the processor 220 therefore proceeds to assign the dynamic class to the obstacle 516. In some examples, as shown in FIG. 4, prior to assigning the dynamic class to the obstacle 516, the processor 220 can determine at block 435 whether a count of observations of the obstacle 516 exceeds a threshold. For example, such a count may be maintained in the map. When the determination is negative, the initial class is retained instead of the dynamic class being assigned. In this example performance, the determination at block 435 is assumed to be affirmative. In other examples, a similar determination may be performed before block 445, in addition to or instead of before block 440.

Following classification of the obstacle 516 as a dynamic obstacle at block 440, the map 520 is updated at block 455. In particular, as shown in FIG. 6, the map 520 is updated to a map 520a, containing the current position of the obstacle 516, as well as an updated timestamp of the most recent observation of the obstacle 516, updated size and velocity attributes, and the above-mentioned probability and class.

At block 460, the determination is negative, e.g. because the obstacle 516 is sufficiently distant from the apparatus 103. The processor 220 therefore returns to block 405.

In a further performance of the method 400, it assumed that the obstacle 516 stops moving in the position shown in FIG. 7. Thus, as the apparatus 103 approaches the obstacle 516, the probability generated at block 430 rises to 0.75. The upper threshold is assumed to be 0.8, and from block 430 the processor 220 therefore proceeds to block 445, to assign the transitional class to the obstacle 516. The transitional class, in effect, indicates that the obstacle 516 is likely a dynamic obstacle, but is temporarily static. At block 455, the map 520a is updated to a map 520b, as shown in FIG. 7.

At block 460, the processor 220 determines that a control action is required due to the proximity of the obstacle 516 and the position of the obstacle on the path 508. Since the obstacle 516 is classified as a transitional obstacle, the selected action is to pause, e.g. for the five-second period mentioned earlier.

In subsequent observations (i.e. subsequent captures at block 405 and processing via blocks 410-455), the obstacle 516 may begin moving again, in which case the probability associated with the obstacle 516 may decrease sufficiently to be classified as a dynamic obstacle, or may remain between the upper and lower thresholds. In other examples, if the obstacle 516 remains stationary for a sufficient period of time, the probability generated at block 430 will exceed the upper threshold, and the obstacle 516 will be reclassified as a static obstacle at block 450.

Following reclassification to static at block 450, the apparatus 103 would no longer perform the “pause” control action at block 465. Instead, the processor 220 selects the second type of control action at block 465, and generates an updated path to navigate around the obstacle 516.

Thus, by employing the transitional class, the apparatus 103 may avoid prematurely generating a new path, until the likelihood that a stationary obstacle will remain stationary is sufficiently high. However, even if the probability that the obstacle 516 is static does not exceed the upper threshold, the use of the transitional class may nevertheless reduce the risk of wasteful path generation, as shown in the example illustrated in FIG. 8.

FIG. 8 shows an alternate scenario, in which the obstacle 516 moved to the other side of the aisle before becoming stationary, permitting the apparatus 103 to pass the obstacle 516 along the path 508. As illustrated, the obstacle 516 is no longer within the field of view 512 of the depth sensor 209. However, because the obstacle 516 was most recently classified as a transitional obstacle (with a probability of 0.68, as shown in the updated map 520c), the processor 220 is configured to apply a static-type retention mechanism to the obstacle 516 at block 415.

In particular, the processor 220 is configured to determine whether the timestamp in the map 520c associated with the obstacle 516 is older than a predefined timeout period. When the timestamp is more recent than the timeout period, the obstacle 516 is retained in the map, although the attributes and probability associated with the obstacle 516 are no longer updated. That is, although transitional obstacles are handled similarly to dynamic obstacles for the purpose of navigational control actions at block 465, such obstacles are handled similarly to static obstacles for map retention purposes at block 415. In some examples, different timeout periods can be applied to static obstacles and transitional obstacles, e.g. with the static timeout period being longer.

The significance of the handling of transitional obstacles at block 415 is clear in the event that a further obstacle 800 is detected, and classified as static. As seen in FIG. 8, the obstacle 800 obstructs the path 508, and given the static classification, the apparatus 103 may determine that generation of an updated path is necessary at block 465. Such a path is generated taking into account the obstacle 516, due to the presence of the obstacle 516 in the map 520c. In the absence of the transitional class and associated navigational and map retention behavior, the apparatus 103 may have simply discarded the obstacle 516 from the map 520c. In the absence of the obstacle 516, a replacement path traveling back along the aisle may have been generated, only to be obstructed by the obstacle 516 when the apparatus 103 turns around to begin executing the new path.

Turning now to FIG. 9, a method 900 of generating an updated probability at block 430 is illustrated. As noted above, the probability can be generated based on attributes such as velocity and, optionally, size. The data captured by the sensor 209 may be noisy, however, and substantial variations may therefore be present in consecutive observations of an obstacle. The method 900 enables the processor 220 to mitigate the effects of such variations on the probability generation.

In general, the probability may be generated as separate components for each attribute, e.g. a size component and a velocity component. For example, the size component may increase with size, e.g. by employing a base value above one with an exponent based on the size. The velocity component, by contrast, may decrease with velocity, e.g. by employing a base value between zero and one with an exponent based on the velocity.

The above components may be combined to produce the probability that the obstacle is static. The components may be weighted when combined, however. The method 900 also enables the processor 220 to adjust the weights applied to each component based on variability of the incoming sensor data.

At block 905, the processor 220 is configured to retrieve stored attributes (e.g. size and velocity) corresponding to the obstacle, e.g. from the map 520. At block 910, the processor 220 determines, for each attribute, a variation between the retrieved attribute and the current attribute (i.e. from the most recent performance of block 410). Obstacles are typically not expected to change in size, and are also not expected to change rapidly in velocity. Therefore, larger changes in size or velocity may indicated lower quality sensor data.

At block 915, for each attribute, the processor 220 determines whether the difference between the stored attribute and the current attribute (e.g. expressed as a fraction of the stored attribute) falls below a threshold. When the determination is negative (i.e. when the difference exceeds the threshold), the current attribute may be inaccurate due to sensor artifacts. The processor 220 is therefore configured to update the weights applied to each component at block 920. For example, the processor 220 can compare the differences between stored and current values for each attribute. The processor 220 can then select the attribute with the smallest variation, and increase the corresponding weight while decreasing the weight of the other attribute.

When the determination at block 915 is affirmative, the processor 220 does not update the weights, but instead proceeds to block 925. At block 925, the processor 220 determines whether the variation between the stored attributes and current attributes has remained below the threshold for a predetermined length of time or number of observations. For example, the processor 220 can increment a counter for each affirmative determination at block 915, and apply a threshold to the counter at block 925. When the determination at block 925 is affirmative, the current attributes are assumed to be sufficiently reliable to use in the future as the stored attributes. At block 930, the stored attributes are therefore replaced with the current attributes. It will be understood that blocks 925 and 930 are performed independently for each attribute, if multiple attributes are used.

When the determination at block 925 is negative, or after the performance of block 920 or block 930, the processor 220 determines the components as mentioned above, using the stored attributes (i.e. not necessarily the current attributes), and the weights as adjusted at block 920, or as previously set. Following the first observation of an obstacle, the weights may be initialized with random values, or with default values (e.g. 0.5 for each weight).

In other examples, blocks 905-930 may be omitted, and the component probabilities may simply be generated and combined based on the current attributes, with a fixed set of weights.

In some examples, combining the component probabilities can include summing the components. In other examples, combining the component probabilities includes summing the components, and weighting the sum with a previous probability. For example, the probability generated at block 935 can include multiplying the sum of the current components by (1-P_N−1), where P_N−1 is the result of the preceding performance of block 935, and adding the result to the square of P_N−1. As a result, the current probability acts as a modifier on the preceding probability, mitigating against significant changes to the probability over successive performances of block 430.

Variations to the above systems and methods are contemplated. In some implementations, more complex implementations of the classification decisions discussed above may be implemented. For example, certain class transitions may be prevented, such as a transition from the static class to the transitional class. In such examples, if an obstacle classified as static begins moving, the probability may be overridden and reset to zero, and the class therefore reset to dynamic, rather than returning to the transitional class. In further examples, a transition directly from the dynamic class to the static class may be prevented, such that obstacles must be reclassified first as transitional obstacles before being classified as static.

As will now be apparent to those skilled in the art, the mechanisms discussed above enable the apparatus 103 to process depth sensor data to increase the efficiency of operation of the apparatus 103, for example by reducing the incidence of unnecessary path regeneration.

In the foregoing specification, specific embodiments have been described. However, one of ordinary skill in the art appreciates that various modifications and changes can be made without departing from the scope of the invention as set forth in the claims below. Accordingly, the specification and figures are to be regarded in an illustrative rather than a restrictive sense, and all such modifications are intended to be included within the scope of present teachings.

The benefits, advantages, solutions to problems, and any element(s) that may cause any benefit, advantage, or solution to occur or become more pronounced are not to be construed as a critical, required, or essential features or elements of any or all the claims. The invention is defined solely by the appended claims including any amendments made during the pendency of this application and all equivalents of those claims as issued.

Moreover in this document, relational terms such as first and second, top and bottom, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. The terms “comprises,” “comprising,” “has”, “having,” “includes”, “including,” “contains”, “containing” or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises, has, includes, contains a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by “comprises . . . a”, “has . . . a”, “includes . . . a”, “contains . . . a” does not, without more constraints, preclude the existence of additional identical elements in the process, method, article, or apparatus that comprises, has, includes, contains the element. The terms “a” and “an” are defined as one or more unless explicitly stated otherwise herein. The terms “substantially”, “essentially”, “approximately”, “about” or any other version thereof, are defined as being close to as understood by one of ordinary skill in the art, and in one non-limiting embodiment the term is defined to be within 10%, in another embodiment within 5%, in another embodiment within 1% and in another embodiment within 0.5%. The term “coupled” as used herein is defined as connected, although not necessarily directly and not necessarily mechanically. A device or structure that is “configured” in a certain way is configured in at least that way, but may also be configured in ways that are not listed.

It will be appreciated that some embodiments may be comprised of one or more specialized processors (or “processing devices”) such as microprocessors, digital signal processors, customized processors and field programmable gate arrays (FPGAs) and unique stored program instructions (including both software and firmware) that control the one or more processors to implement, in conjunction with certain non-processor circuits, some, most, or all of the functions of the method and/or apparatus described herein. Alternatively, some or all functions could be implemented by a state machine that has no stored program instructions, or in one or more application specific integrated circuits (ASICs), in which each function or some combinations of certain of the functions are implemented as custom logic. Of course, a combination of the two approaches could be used.

Moreover, an embodiment can be implemented as a computer-readable storage medium having computer readable code stored thereon for programming a computer (e.g., comprising a processor) to perform a method as described and claimed herein. Examples of such computer-readable storage mediums include, but are not limited to, a hard disk, a CD-ROM, an optical storage device, a magnetic storage device, a ROM (Read Only Memory), a PROM (Programmable Read Only Memory), an EPROM (Erasable Programmable Read Only Memory), an EEPROM (Electrically Erasable Programmable Read Only Memory) and a Flash memory. Further, it is expected that one of ordinary skill, notwithstanding possibly significant effort and many design choices motivated by, for example, available time, current technology, and economic considerations, when guided by the concepts and principles disclosed herein will be readily capable of generating such software instructions and programs and ICs with minimal experimentation.

The Abstract of the Disclosure is provided to allow the reader to quickly ascertain the nature of the technical disclosure. It is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims. In addition, in the foregoing Detailed Description, it can be seen that various features are grouped together in various embodiments for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments require more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive subject matter lies in less than all features of a single disclosed embodiment. Thus the following claims are hereby incorporated into the Detailed Description, with each claim standing on its own as a separately claimed subject matter.

Claims

1. A navigational control method for a mobile automation apparatus, comprising: controlling a depth sensor to capture depth data representing a portion of a facility containing an obstacle;identifying the obstacle from the depth data;determining a probability that the obstacle is static;based on the probability, assigning the obstacle one of (i) a dynamic class, (ii) a static class, and (iii) at least one intermediate class;updating a map to include a position of the obstacle, and the assigned class;selecting, based on the assigned class, a navigational control action from (i) a first action type associated with the dynamic class and the intermediate class, and (ii) a second action type associated with the static class; and,retaining, based on the assigned class, the obstacle in the map according to one of (i) a first retention mechanism associated with the dynamic class, and (ii) a second retention mechanism associated with the static class and the intermediate class.

2. The method of claim 1, wherein identifying the obstacle includes detecting at least one attribute of the obstacle from the depth data; and wherein determining the probability is based on the at least one attribute.

3. The method of claim 2, wherein the at least one attribute includes a size attribute and a velocity attribute, and wherein determining the probability includes: determining a first probability component based on the size attribute;determining a second probability component based on the velocity attribute; andcombining the first and second component probabilities.

4. The method of claim 3, wherein combining the component probabilities includes applying respective weights to the component probabilities.

5. The method of claim 1, wherein the intermediate class includes a transitional class; and wherein assigning the one of the classes includes: assigning the dynamic class when the probability is below a lower threshold;assigning the static class when the probability is above an upper threshold; andassigning the transitional class when the probability is between the upper and lower thresholds.

6. The method of claim 1, wherein the intermediate class includes an initial class; and wherein the method further comprises: determining whether the map contains a previous observation of the obstacle; andwhen the map does not contain a previous observation of the obstacle, assigning the initial class to the obstacle, and setting the probability to an initial value.

7. The method of claim 1 wherein the map includes a timestamp corresponding to the most recent observation of the obstacle; and wherein first retention mechanism includes: discarding the obstacle from the map when the timestamp is older than a timeout period.

8. The method of claim 7, wherein the timeout period is one of a static timeout period corresponding to static obstacles, and a transitional timeout period corresponding to transitional obstacles, and wherein the static timeout period is greater than the transitional timeout period.

9. The method of claim 1, wherein selecting the navigational control action includes: determining whether the obstacle is in a current path of the mobile automation apparatus; andwhen the obstacle is in the current path, (i) modifying travel along the path if the obstacle is assigned the dynamic class or the at least one intermediate class, or(ii) generating an updated path to avoid the obstacle, if the obstacle is assigned the static class.

10. A computing device, comprising: a depth sensor; a locomotive assembly; anda navigational controller configured to:control the depth sensor to capture depth data representing a portion of a facility containing an obstacle;identify the obstacle from the depth data;determine a probabilitythat the obstacle is static;based on the probability, assign the obstacle one of (i) a dynamic class, (ii) a static class, and (iii) at least one intermediate class;update a map to include a position of the obstacle, and the assigned class; andselect, based on the assigned class, a navigational control action from (i) a first action type associated with the dynamic class and the intermediate class, and (ii) a second action type associated with the static class wherein: the intermediate class includes an initial class, and the navigational controller is further configured to: determine whether the map contains a previous observation of the obstacle, andwhen the map does not contain a previous observation of the obstacle, assign the initial class to the obstacle, and set the probability to an initial value.

11. The computing device of claim 10, wherein the navigational controller is configured to identify the obstacle by detecting at least one attribute of the obstacle from the depth data; and wherein determination of the probability is based on the at least one attribute.

12. The computing device of claim 11, wherein the at least one attribute includes a size attribute and a velocity attribute, and wherein the navigational controller is configured to determine the probability by: determining a first probability component based on the size attribute;determining a second probability component based on the velocity attribute; andcombining the first and second component probabilities.

13. The computing device of claim 12, wherein the navigational controller is configured to combine the component probabilities by applying respective weights to the component probabilities.

14. The computing device of claim 10, wherein the intermediate class includes a transitional class; and wherein the navigational controller is configured to assign the one of the classes by: assigning the dynamic class when the probability is below a lower threshold;assigning the static class when the probability is above an upper threshold; andassigning the transitional class when the probability is between the upper and lower thresholds.

15. The computing device of claim 10, wherein the navigational controller is further configured to: based on the assigned class, retain the obstacle in the map according to one of (i) a first retention mechanism associated with the dynamic class, and (ii) a second retention mechanism associated with the static class and the intermediate class.

16. The computing device of claim 15, wherein the map includes a timestamp corresponding to the most recent observation of the obstacle; and wherein the navigational controller is configured to perform the first retention mechanism by:discarding the obstacle from the map when the timestamp is older than a timeout period.

17. The computing device of claim 16, wherein the timeout period is one of a static timeout period corresponding to static obstacles, and a transitional timeout period corresponding to transitional obstacles, and wherein the static timeout period is greater than the transitional timeout period.

18. The computing device of claim 10, wherein the navigational controller is configured to select the navigational control action by: determining whether the obstacle is in a current path of the mobile automation apparatus; andwhen the obstacle is in the current path, (i) pausing travel along the path if the obstacle is assigned the dynamic class, or(ii) generating an updated path if the obstacle is assigned the static class.

19. A navigational control method for a mobile automation apparatus, comprising: controlling a depth sensor to capture depth data representing a portion of a facility containing an obstacle;identifying the obstacle from the depth data by detecting at least one attribute of the obstacle;determining a probability that the obstacle is static based on the at least one attribute;based on the probability, assigning the obstacle one of (i) a dynamic class, (ii) a static class, and (iii) at least one intermediate class; andupdating a map to include a position of the obstacle, and the assigned class;selecting, based on the assigned class, a navigational control action from (i) a first action type associated with the dynamic class and the intermediate class, and (ii) a second action type associated with the static class,wherein: the at least one attribute includes a size attribute and a velocity attribute, anddetermining the probability includes: determining a first probability component based on the size attribute,determining a second probability component based on the velocity attribute, andcombining the first and second component probabilities.