CN117152709A

CN117152709A - Computer system, computer-implemented method, and computer-readable medium

Info

Publication number: CN117152709A
Application number: CN202310024784.0A
Authority: CN
Inventors: E·M·沃尔夫; O·贝基波姆; A·朗; S·沃拉; B·赫劳; E·C·格里戈雷; 姜成
Original assignee: Motional AD LLC
Current assignee: Motional AD LLC
Priority date: 2022-05-31
Filing date: 2023-01-09
Publication date: 2023-12-01

Abstract

The invention relates to a computer system, a computer-implemented method and a computer-readable medium. Methods of motion prediction in autonomous vehicles using fused composite and camera images are provided. The method may include obtaining pairs of data, wherein each pair of data reflects: data corresponding to a composite image representing a bird's eye view of an area surrounding the vehicle and identifying the object, and data corresponding to a camera image depicting the object. The machine learning model may be trained based on the data pairs to produce a trained model that predicts motion of objects within the data pairs based on the composite image and the camera image of the data pairs. Systems and computer program products are also provided.

Description

Computer system, computer-implemented method, and computer-readable medium

Technical Field

The present disclosure relates to computer systems, computer implemented methods, and computer readable media.

Background

Accurate perception of the environment of a vehicle is critical to ensuring safe and effective vehicle operation. However, combining the various available sensor modalities into an accurate and useful perceived understanding is often challenging.

Disclosure of Invention

A computer system, comprising: one or more computer-readable storage devices configured to store computer-executable instructions; and one or more computer processors configured to execute the computer-executable instructions, wherein execution of the computer-executable instructions causes the computer system to: obtaining a set of data pairs, each data pair comprising: first data corresponding to a composite image representing a bird's eye view of an area generated based on sensor data of a vehicle in the area, wherein the composite image identifies an object in the area; and second data corresponding to a camera image representing a viewpoint of the vehicle in the area, wherein the camera image depicts the object; training a machine learning model based on the set of data pairs to produce a trained model, wherein the machine learning model includes at least one convolutional neural network to process the set of data pairs, and wherein the machine learning model accepts as input a given data pair of the set of data pairs and provides as output a predicted motion of an object corresponding to the given data pair; and transmitting the trained model to a destination vehicle, wherein the destination vehicle is configured to apply the trained model to sensor data of the destination vehicle to predict movement of a target object identified within the sensor data.

Drawings

FIG. 1 is an example environment in which a vehicle including one or more components of an autonomous system may be implemented;

FIG. 2 is a diagram of one or more systems of a vehicle including an autonomous system;

FIG. 3 is a diagram of components of one or more devices and/or one or more systems of FIGS. 1 and 2;

FIG. 4A is a diagram of certain components of an autonomous system;

FIG. 4B is a diagram of an implementation of a neural network;

FIGS. 4C and 4D are diagrams illustrating example operations of a convolutional neural network;

5-9 are diagrams of example implementations of machine learning models for fusing composite images from a perception system and data from additional sensor modalities (such as images from cameras, etc.);

FIG. 10 depicts an example routine for training a machine learning model to fuse data of the output of the sensing system and additional sensor modalities; and

FIG. 11 depicts a routine for predicting object motion or planning vehicle motion using a trained machine learning model.

Detailed Description

In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. It will be apparent, however, that the embodiments described in this disclosure may be practiced without these specific details. In some instances, well-known structures and devices are shown in block diagram form in order to avoid unnecessarily obscuring aspects of the present disclosure.

In the drawings, for ease of description, specific arrangements or sequences of illustrative elements (such as those representing systems, devices, modules, blocks of instructions, and/or data elements, etc.) are illustrated. However, those of skill in the art will understand that a specific order or arrangement of elements illustrated in the drawings is not intended to require a specific order or sequence of processes, or separation of processes, unless explicitly described. Furthermore, the inclusion of a schematic element in a figure is not intended to mean that such element is required in all embodiments nor that the feature represented by such element is not included in or combined with other elements in some embodiments unless explicitly described.

Furthermore, in the drawings, connecting elements (such as solid or dashed lines or arrows, etc.) are used to illustrate a connection, relationship or association between or among two or more other schematic elements, the absence of any such connecting element is not intended to mean that no connection, relationship or association exists. In other words, some connections, relationships, or associations between elements are not illustrated in the drawings so as not to obscure the present disclosure. Further, for ease of illustration, a single connection element may be used to represent multiple connections, relationships, or associations between elements. For example, if a connection element represents a communication of signals, data, or instructions (e.g., "software instructions"), those skilled in the art will understand that such element may represent one or more signal paths (e.g., buses) that may be required to effect the communication.

Although the terms "first," "second," and/or "third," etc. may be used to describe various elements, these elements should not be limited by these terms. The terms "first," second, "and/or third," etc. are used merely to distinguish one element from another element. For example, a first contact may be referred to as a second contact, and similarly, a second contact may be referred to as a first contact, without departing from the scope of the described embodiments. Both the first contact and the second contact are contacts, but they are not the same contacts.

The terminology used in the description of the various embodiments described herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in the specification of the various embodiments described and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, and may be used interchangeably with "one or more than one" or "at least one," unless the context clearly indicates otherwise. It will also be understood that the term "and/or" as used herein refers to and includes any and all possible combinations of one or more of the associated listed items. It will be further understood that the terms "comprises," "comprising," "includes," "including" and/or "having," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

As used herein, the terms "communication" and "communicating" refer to at least one of the receipt, transmission, and/or provision of information (or information represented by, for example, data, signals, messages, instructions, and/or commands, etc.). For one unit (e.g., a device, system, component of a device or system, and/or a combination thereof, etc.) to communicate with another unit, this means that the one unit is capable of directly or indirectly receiving information from and/or sending (e.g., transmitting) information to the other unit. This may refer to a direct or indirect connection that is wired and/or wireless in nature. In addition, two units may communicate with each other even though the transmitted information may be modified, processed, relayed and/or routed between the first unit and the second unit. For example, a first unit may communicate with a second unit even if the first unit passively receives information and does not actively transmit information to the second unit. As another example, if at least one intervening unit (e.g., a third unit located between the first unit and the second unit) processes information received from the first unit and transmits the processed information to the second unit, the first unit may communicate with the second unit. In some embodiments, a message may refer to a network packet (e.g., a data packet, etc.) that includes data.

As used herein, the term "if" is optionally interpreted to mean "when …", "at …", "in response to being determined to" and/or "in response to being detected", etc., depending on the context. Similarly, the phrase "if determined" or "if [ a stated condition or event ] is detected" is optionally interpreted to mean "upon determination …", "in response to determination" or "upon detection of [ a stated condition or event ]" and/or "in response to detection of [ a stated condition or event ]" or the like, depending on the context. Furthermore, as used herein, the terms "having," "having," or "owning," and the like, are intended to be open-ended terms. Furthermore, unless explicitly stated otherwise, the phrase "based on" is intended to mean "based, at least in part, on".

Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the various embodiments described. It will be apparent, however, to one of ordinary skill in the art that the various embodiments described may be practiced without these specific details. In other instances, well-known methods, procedures, components, circuits, and networks have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.

General overview

In some aspects and/or embodiments, the systems, methods, and computer program products described herein include and/or enable fusion of camera images with the output of a perception system of an autonomous vehicle for purposes such as planning or motion prediction. In general, autonomous vehicles have various sensor modalities such as cameras, lidar, or radar for attempting to perceive the environment of the autonomous vehicle. These sensations are then used as inputs to other systems, such as planning systems, to control the operation of the vehicle. Thus, accurate perception of the environment of a vehicle is critical to ensuring safe and efficient vehicle operation. However, combining the various available sensor modalities into an accurate and useful perceived understanding is often challenging. In practice, it often occurs that one modality is better than the other modality or even a combination of other modalities. For example, one mechanism for providing contextual understanding of the environment is to generate a Bird's Eye View (BEV) composite image of the area surrounding the vehicle, where the Bird's Eye View (BEV) composite image models, for example, the vehicle and objects surrounding the vehicle from a view directly above the vehicle. When producing such BEV composite images, it is often the case that using one modality alone, such as lidar, produces more accurate results than trying to combine multiple modalities. This may occur, for example, due to increased accuracy of one modality or difficulty in projecting data from other modalities into the relevant view. However, this modality may not capture important contexts that most humans intuitively understand that are important to accurate vehicle operation. For example, lidar alone may not be able to capture important signals such as the presence (or absence) of brake lights or turn signals on other vehicles, where these signals may be indicative of possible movements of other vehicles, representing important data for planning the actions of autonomous vehicles. Accordingly, autonomous vehicles often need to decide between reduced accuracy due to the inclusion of multiple sensor modalities when perceiving their environment and lost context due to the limited modalities.

Embodiments of the present disclosure provide solutions to these problems by providing a fusion of the output of the additional sensor modality with the output of the sensing system in the following manner: capturing context from those additional sensor modalities is provided without interfering with the accuracy of the sensing system. Specifically, as disclosed herein, the machine learning model may be trained to take as input the output of the perception system, such as BEV images, and data from additional sensor modalities, such as camera images. For example, the machine learning model may include one or more convolutional neural networks having as inputs the composite BEV image and the raw camera image. The machine learning model may then output information for motion prediction or planning. For example, where the BEV image represents an object in the environment of a vehicle (e.g., other vehicles) and the original camera image depicts the object, the output of the machine learning model may represent the predicted motion of the object. Illustratively, if the object is another vehicle and the camera image captures that the brake lights of the other vehicle are illuminated, the model may output a prediction that the vehicle is likely to stop. In this way, the data captured within the BEV image may be supplemented with data from additional sensor modalities. Because the data from the additional sensor modalities is combined with the output of the sensing system (e.g., BEV image) after its generation, the data of the additional sensor modalities does not interfere with the sensing system as might be introduced to generate the initial output of the sensing system. Thus, these embodiments enable capturing context from additional sensor modalities in a manner that overcomes the challenges of simply including data of the additional sensor modalities in a perception system.

In addition to motion prediction of objects, embodiments of the present disclosure may also provide motion planning at autonomous vehicles. Illustratively, it may be desirable for the vehicle to operate as far as possible in a manner that would be operated by a skilled human driver. Accordingly, a human driver may utilize the context such as brake lights, turn signals, etc., captured in some, but not all, of the sensor modalities. Accordingly, embodiments of the present disclosure may include a machine learning model that utilizes a combination of data of the perception system output (e.g., synthetic BEV images) and additional sensor modalities to provide planning actions of the vehicle in addition to or as an alternative to motion prediction of the perception object.

Those skilled in the art will appreciate in view of this disclosure that embodiments disclosed herein improve the ability of a computing system (such as a computing device included within or supporting operation of an autonomous vehicle) to conduct object motion prediction or vehicle planning. Moreover, the presently disclosed embodiments address technical problems inherent in computing systems; in particular, it is difficult to accurately combine data of multiple sensor modalities in a manner that captures various available contextual information across these modalities. These technical problems are addressed by the various technical solutions described herein that include the use of a machine learning model that is trained to combine the output of a perception system (such as a synthetic BEV image, etc.) and data of additional sensor modalities (such as a camera image, etc.) to produce predicted object motions or planning actions. Accordingly, the present disclosure generally represents improvements to computer vision systems and computing systems.

The foregoing aspects and many of the attendant advantages of this disclosure will become more readily appreciated as the same become better understood by reference to the following description, when taken in conjunction with the accompanying drawings.

Referring now to FIG. 1, an example environment 100 is illustrated in which a vehicle that includes an autonomous system and a vehicle that does not include an autonomous system operate in the example environment 100. As illustrated, environment 100 includes vehicles 102a-102n, objects 104a-104n, routes 106a-106n, areas 108, vehicle-to-infrastructure (V2I) devices 110, a network 112, a remote Autonomous Vehicle (AV) system 114, a queue management system 116, and a V2I system 118. The vehicles 102a-102n, the vehicle-to-infrastructure (V2I) device 110, the network 112, the Autonomous Vehicle (AV) system 114, the queue management system 116, and the V2I system 118 are interconnected via wired connections, wireless connections, or a combination of wired or wireless connections (e.g., establishing a connection for communication, etc.). In some embodiments, the objects 104a-104n are interconnected with at least one of the vehicles 102a-102n, the vehicle-to-infrastructure (V2I) device 110, the network 112, the Autonomous Vehicle (AV) system 114, the queue management system 116, and the V2I system 118 via a wired connection, a wireless connection, or a combination of wired or wireless connections.

The vehicles 102a-102n (individually referred to as vehicles 102 and collectively referred to as vehicles 102) include at least one device configured to transport cargo and/or personnel. In some embodiments, the vehicle 102 is configured to communicate with the V2I device 110, the remote AV system 114, the queue management system 116, and/or the V2I system 118 via the network 112. In some embodiments, the vehicle 102 comprises a car, bus, truck, train, or the like. In some embodiments, the vehicle 102 is the same as or similar to the vehicle 200 (see fig. 2) described herein. In some embodiments, vehicles 200 in a group of vehicles 200 are associated with an autonomous queue manager. In some embodiments, the vehicles 102 travel along respective routes 106a-106n (individually referred to as routes 106 and collectively referred to as routes 106), as described herein. In some embodiments, one or more vehicles 102 include an autonomous system (e.g., the same or similar to autonomous system 202).

The objects 104a-104n (individually referred to as objects 104 and collectively referred to as objects 104) include, for example, at least one vehicle, at least one pedestrian, at least one rider, and/or at least one structure (e.g., building, sign, hydrant, etc.), and the like. Each object 104 is stationary (e.g., at a fixed location and for a period of time) or moves (e.g., has a velocity and is associated with at least one trajectory). In some embodiments, the object 104 is associated with a respective location in the region 108.

Routes 106a-106n (individually referred to as routes 106 and collectively referred to as routes 106) are each associated with (e.g., define) a series of actions (also referred to as tracks) that connect the states along which the AV can navigate. Each route 106 begins in an initial state (e.g., a state corresponding to a first space-time location and/or speed, etc.) and ends in a final target state (e.g., a state corresponding to a second space-time location different from the first space-time location) or target area (e.g., a subspace of acceptable states (e.g., end states)). In some embodiments, the first state includes one or more places where the one or more individuals are to pick up the AV, and the second state or zone includes one or more places where the one or more individuals pick up the AV are to be off. In some embodiments, the route 106 includes a plurality of acceptable state sequences (e.g., a plurality of spatiotemporal site sequences) associated with (e.g., defining) a plurality of trajectories. In an example, the route 106 includes only high-level actions or imprecise status places, such as a series of connecting roads indicating a change of direction at a roadway intersection, and the like. Additionally or alternatively, the route 106 may include more precise actions or states such as, for example, specific target lanes or precise locations within a lane region, and target speeds at these locations, etc. In an example, the route 106 includes a plurality of precise state sequences along at least one high-level action with a limited look-ahead view to an intermediate target, where a combination of successive iterations of the limited view state sequences cumulatively corresponds to a plurality of trajectories that collectively form a high-level route that terminates at a final target state or zone.

The area 108 includes a physical area (e.g., a geographic area) that the vehicle 102 may navigate. In an example, the region 108 includes at least one state (e.g., a country, a province, an individual state of a plurality of states included in a country, etc.), at least a portion of a state, at least one city, at least a portion of a city, etc. In some embodiments, the area 108 includes at least one named thoroughfare (referred to herein as a "road"), such as a highway, interstate, park, city street, or the like. Additionally or alternatively, in some examples, the area 108 includes at least one unnamed road, such as a roadway, a section of a parking lot, a section of an open space and/or undeveloped area, a mud path, and the like. In some embodiments, the roadway includes at least one lane (e.g., a portion of the roadway through which the vehicle 102 may traverse). In an example, the road includes at least one lane associated with (e.g., identified based on) the at least one lane marker.

A Vehicle-to-infrastructure (V2I) device 110 (sometimes referred to as a Vehicle-to-Everything (V2X) device) includes at least one device configured to communicate with the Vehicle 102 and/or the V2I system 118. In some embodiments, V2I device 110 is configured to communicate with vehicle 102, remote AV system 114, queue management system 116, and/or V2I system 118 via network 112. In some embodiments, V2I device 110 includes a Radio Frequency Identification (RFID) device, a sign, a camera (e.g., a two-dimensional (2D) and/or three-dimensional (3D) camera), a lane marker, a street light, a parking meter, and the like. In some embodiments, the V2I device 110 is configured to communicate directly with the vehicle 102. Additionally or alternatively, in some embodiments, the V2I device 110 is configured to communicate with the vehicle 102, the remote AV system 114, and/or the queue management system 116 via the V2I system 118. In some embodiments, V2I device 110 is configured to communicate with V2I system 118 via network 112.

Network 112 includes one or more wired and/or wireless networks. In an example, the network 112 includes a cellular network (e.g., a Long Term Evolution (LTE) network, a third generation (3G) network, a fourth generation (4G) network, a fifth generation (5G) network, a Code Division Multiple Access (CDMA) network, etc.), a Public Land Mobile Network (PLMN), a Local Area Network (LAN), a Wide Area Network (WAN), a Metropolitan Area Network (MAN), a telephone network (e.g., a Public Switched Telephone Network (PSTN)), a private network, an ad hoc network, an intranet, the internet, a fiber-optic based network, a cloud computing network, etc., and/or a combination of some or all of these networks, etc.

The remote AV system 114 includes at least one device configured to communicate with the vehicle 102, the V2I device 110, the network 112, the queue management system 116, and/or the V2I system 118 via the network 112. In an example, the remote AV system 114 includes a server, a group of servers, and/or other similar devices. In some embodiments, the remote AV system 114 is co-located with the queue management system 116. In some embodiments, the remote AV system 114 participates in the installation of some or all of the components of the vehicle (including autonomous systems, autonomous vehicle computing, and/or software implemented by autonomous vehicle computing, etc.). In some embodiments, the remote AV system 114 maintains (e.g., updates and/or replaces) these components and/or software over the life of the vehicle.

The queue management system 116 includes at least one device configured to communicate with the vehicle 102, the V2I device 110, the remote AV system 114, and/or the V2I system 118. In an example, the queue management system 116 includes a server, a server group, and/or other similar devices. In some embodiments, the queue management system 116 is associated with a carpool company (e.g., an organization for controlling operation of multiple vehicles (e.g., vehicles that include autonomous systems and/or vehicles that do not include autonomous systems), etc.).

In some embodiments, the V2I system 118 includes at least one device configured to communicate with the vehicle 102, the V2I device 110, the remote AV system 114, and/or the queue management system 116 via the network 112. In some examples, the V2I system 118 is configured to communicate with the V2I device 110 via a connection other than the network 112. In some embodiments, V2I system 118 includes a server, a server farm, and/or other similar devices. In some embodiments, the V2I system 118 is associated with a municipality or private institution (e.g., a private institution for maintaining the V2I device 110, etc.).

The number and arrangement of elements illustrated in fig. 1 are provided as examples. There may be additional elements, fewer elements, different elements, and/or differently arranged elements than those illustrated in fig. 1. Additionally or alternatively, at least one element of environment 100 may perform one or more functions described as being performed by at least one different element of fig. 1. Additionally or alternatively, at least one set of elements of environment 100 may perform one or more functions described as being performed by at least one different set of elements of environment 100.

Referring now to fig. 2, a vehicle 200 (which may be the same as or similar to vehicle 102 of fig. 1) includes an autonomous system 202, a powertrain control system 204, a steering control system 206, and a braking system 208 or is associated with autonomous system 202, powertrain control system 204, steering control system 206, and braking system 208. In some embodiments, the vehicle 200 is the same as or similar to the vehicle 102 (see fig. 1). In some embodiments, autonomous system 202 is configured to confer autonomous driving capability to vehicle 200 (e.g., implement at least one driving automation or steering-based function, feature, and/or device, etc., that enables vehicle 200 to operate partially or fully without human intervention, including, but not limited to, fully autonomous vehicles (e.g., vehicles that forgo reliance on human intervention, such as vehicles operated at 5-level ADS, etc.), highly autonomous vehicles (e.g., vehicles that forgo reliance on human intervention, in some cases, such as vehicles operated at 4-level ADS, etc.), and/or conditional autonomous vehicles (e.g., vehicles that forgo reliance on human intervention, in limited situations, such as vehicles operated at 3-level ADS, etc.), etc. In one embodiment, the autonomous system 202 includes the operations or strategic functions required to operate the vehicle 200 in road traffic and perform some or all of the Dynamic Driving Tasks (DDTs) on a continuous basis. In another embodiment, the autonomous system 202 includes an Advanced Driving Assistance System (ADAS) with driving support features. The autonomous system 202 supports various levels of driving automation ranging from unmanned automation (e.g., level 0) to full driving automation (e.g., level 5). For a detailed description of fully autonomous vehicles and highly autonomous vehicles, reference may be made to SAE International Standard J3016, classification and definition of on-road automotive autopilot system related terms (SAE International's Standard J3016: taxonomy and Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems), which is incorporated by reference in its entirety. In some embodiments, the vehicle 200 is associated with an autonomous queue manager and/or a carpooling company.

The autonomous system 202 includes a sensor suite that includes one or more devices such as a camera 202a, liDAR sensor 202b, radar (radar) sensor 202c, and microphone 202 d. In some embodiments, autonomous system 202 may include more or fewer devices and/or different devices (e.g., ultrasonic sensors, inertial sensors, GPS receivers (discussed below), and/or odometry sensors for generating data associated with an indication of the distance that vehicle 200 has traveled, etc.). In some embodiments, the autonomous system 202 uses one or more devices included in the autonomous system 202 to generate data associated with the environment 100 described herein. The data generated by the one or more devices of the autonomous system 202 may be used by the one or more systems described herein to observe the environment (e.g., environment 100) in which the vehicle 200 is located. In some embodiments, autonomous system 202 includes a communication device 202e, an autonomous vehicle calculation 202f, and a safety controller 202g.

The camera 202a includes at least one device configured to communicate with the communication device 202e, the autonomous vehicle calculation 202f, and/or the safety controller 202g via a bus (e.g., the same or similar to the bus 302 of fig. 3). The camera 202a includes at least one camera (e.g., a digital camera using a light sensor such as a Charge Coupled Device (CCD), thermal camera, infrared (IR) camera, event camera, etc.) to capture images including physical objects (e.g., cars, buses, curbs, and/or people, etc.). In some embodiments, camera 202a generates camera data as output. In some examples, camera 202a generates camera data including image data associated with the image. In this example, the image data may specify at least one parameter corresponding to the image (e.g., image characteristics such as exposure, brightness, etc., and/or an image timestamp, etc.). In such examples, the image may be in a format (e.g., RAW, JPEG, and/or PNG, etc.). In some embodiments, the camera 202a includes a plurality of independent cameras configured (e.g., positioned) on the vehicle to capture images for stereoscopic (stereo vision) purposes. In some examples, camera 202a includes a plurality of cameras that generate and transmit image data to autonomous vehicle computing 202f and/or a queue management system (e.g., a queue management system that is the same as or similar to queue management system 116 of fig. 1). In such an example, the autonomous vehicle calculation 202f determines a depth to one or more objects in the field of view of at least two cameras of the plurality of cameras based on image data from the at least two cameras. In some embodiments, camera 202a is configured to capture images of objects within a distance (e.g., up to 100 meters and/or up to 1 kilometer, etc.) relative to camera 202 a. Thus, the camera 202a includes features such as sensors and lenses that are optimized for sensing objects at one or more distances relative to the camera 202 a.

In an embodiment, camera 202a includes at least one camera configured to capture one or more images associated with one or more traffic lights, street signs, and/or other physical objects that provide visual navigation information. In some embodiments, camera 202a generates traffic light data associated with one or more images. In some examples, the camera 202a generates TLD (traffic light detection) data associated with one or more images including formats (e.g., RAW, JPEG, and/or PNG, etc.). In some embodiments, the camera 202a that generates TLD data differs from other systems described herein that include cameras in that: the camera 202a may include one or more cameras having a wide field of view (e.g., wide angle lens, fisheye lens, and/or lens having a viewing angle of about 120 degrees or greater, etc.) to generate images related to as many physical objects as possible.

Light detection and ranging (LiDAR) sensor 202b includes at least one device configured to communicate with communication device 202e, autonomous vehicle computation 202f, and/or security controller 202g via a bus (e.g., the same or similar bus as bus 302 of fig. 3). LiDAR sensor 202b includes a system configured to emit light from a light emitter (e.g., a laser emitter). Light emitted by the LiDAR sensor 202b includes light outside the visible spectrum (e.g., infrared light, etc.). In some embodiments, during operation, light emitted by the LiDAR sensor 202b encounters a physical object (e.g., a vehicle) and is reflected back to the LiDAR sensor 202b. In some embodiments, the light emitted by LiDAR sensor 202b does not penetrate the physical object that the light encounters. LiDAR sensor 202b also includes at least one light detector that detects light emitted from the light emitter after the light encounters a physical object. In some embodiments, at least one data processing system associated with the LiDAR sensor 202b generates an image (e.g., a point cloud and/or a combined point cloud, etc.) representative of objects included in the field of view of the LiDAR sensor 202b. In some examples, at least one data processing system associated with the LiDAR sensor 202b generates images representing boundaries of the physical object and/or surfaces (e.g., topology of surfaces) of the physical object, etc. In such an example, the image is used to determine the boundary of a physical object in the field of view of the LiDAR sensor 202b.

The radio detection and ranging (radar) sensor 202c includes at least one device configured to communicate with the communication device 202e, the autonomous vehicle calculation 202f, and/or the safety controller 202g via a bus (e.g., the same or similar bus as the bus 302 of fig. 3). The radar sensor 202c includes a system configured to emit (pulsed or continuous) radio waves. The radio waves emitted by the radar sensor 202c include radio waves within a predetermined frequency spectrum. In some embodiments, during operation, radio waves emitted by the radar sensor 202c encounter a physical object and are reflected back to the radar sensor 202c. In some embodiments, the radio waves emitted by the radar sensor 202c are not reflected by some objects. In some embodiments, at least one data processing system associated with radar sensor 202c generates signals representative of objects included in the field of view of radar sensor 202c. For example, at least one data processing system associated with radar sensor 202c generates images representing boundaries of physical objects and/or surfaces (e.g., topology of surfaces) of physical objects, etc. In some examples, the image is used to determine boundaries of physical objects in the field of view of radar sensor 202c.

Microphone 202d includes at least one device configured to communicate with communication device 202e, autonomous vehicle computing 202f, and/or security controller 202g via a bus (e.g., the same or similar bus as bus 302 of fig. 3). Microphone 202d includes one or more microphones (e.g., array microphone and/or external microphone, etc.) that capture an audio signal and generate data associated with (e.g., representative of) the audio signal. In some examples, microphone 202d includes transducer means and/or the like. In some embodiments, one or more systems described herein may receive data generated by microphone 202d and determine a position (e.g., distance, etc.) of an object relative to vehicle 200 based on an audio signal associated with the data.

The communication device 202e includes at least one device configured to communicate with a camera 202a, a LiDAR sensor 202b, a radar sensor 202c, a microphone 202d, an autonomous vehicle calculation 202f, a security controller 202g, and/or a DBW (drive-by-wire) system 202 h. For example, communication device 202e may include the same or similar devices as communication interface 314 of fig. 3. In some embodiments, the communication device 202e comprises a vehicle-to-vehicle (V2V) communication device (e.g., a device for enabling wireless communication of data between vehicles).

The autonomous vehicle calculation 202f includes at least one device configured to communicate with the camera 202a, the LiDAR sensor 202b, the radar sensor 202c, the microphone 202d, the communication device 202e, the security controller 202g, and/or the DBW system 202 h. In some examples, the autonomous vehicle computing 202f includes devices such as client devices, mobile devices (e.g., cellular phones and/or tablet computers, etc.), and/or servers (e.g., computing devices including one or more central processing units and/or graphics processing units, etc.), among others. In some embodiments, the autonomous vehicle calculation 202f is the same as or similar to the autonomous vehicle calculation 400 described herein. Additionally or alternatively, in some embodiments, the autonomous vehicle computing 202f is configured to communicate with an autonomous vehicle system (e.g., an autonomous vehicle system that is the same as or similar to the remote AV system 114 of fig. 1), a queue management system (e.g., a queue management system that is the same as or similar to the queue management system 116 of fig. 1), a V2I device (e.g., a V2I device that is the same as or similar to the V2I device 110 of fig. 1), and/or a V2I system (e.g., a V2I system that is the same as or similar to the V2I system 118 of fig. 1).

The safety controller 202g includes at least one device configured to communicate with the camera 202a, the LiDAR sensor 202b, the radar sensor 202c, the microphone 202d, the communication device 202e, the autonomous vehicle calculation 202f, and/or the DBW system 202 h. In some examples, the safety controller 202g includes one or more controllers (electrical and/or electromechanical controllers, etc.) configured to generate and/or transmit control signals to operate one or more devices of the vehicle 200 (e.g., the powertrain control system 204, the steering control system 206, and/or the braking system 208, etc.). In some embodiments, the safety controller 202g is configured to generate control signals that override (e.g., override) control signals generated and/or transmitted by the autonomous vehicle calculation 202 f.

The DBW system 202h includes at least one device configured to communicate with the communication device 202e and/or the autonomous vehicle calculation 202 f. In some examples, the DBW system 202h includes one or more controllers (e.g., electrical and/or electromechanical controllers, etc.) configured to generate and/or transmit control signals to operate one or more devices of the vehicle 200 (e.g., the powertrain control system 204, the steering control system 206, and/or the braking system 208, etc.). Additionally or alternatively, one or more controllers of the DBW system 202h are configured to generate and/or transmit control signals to operate at least one different device of the vehicle 200 (e.g., turn signal lights, headlights, door locks, and/or windshield wipers, etc.).

The powertrain control system 204 includes at least one device configured to communicate with the DBW system 202 h. In some examples, the powertrain control system 204 includes at least one controller and/or actuator, etc. In some embodiments, the powertrain control system 204 receives control signals from the DBW system 202h, and the powertrain control system 204 causes the vehicle 200 to perform longitudinal vehicle movements, such as starting forward movement, stopping forward movement, starting backward movement, stopping backward movement, accelerating in a certain direction, decelerating in a certain direction, or performing lateral vehicle movements, such as making a left turn and/or making a right turn. In an example, the powertrain control system 204 increases, maintains the same, or decreases the energy (e.g., fuel and/or electricity, etc.) provided to the motor of the vehicle, thereby rotating or not rotating at least one wheel of the vehicle 200.

The steering control system 206 includes at least one device configured to rotate one or more wheels of the vehicle 200. In some examples, the steering control system 206 includes at least one controller and/or actuator, etc. In some embodiments, steering control system 206 rotates the two front wheels and/or the two rear wheels of vehicle 200 to the left or right to turn vehicle 200 to the left or right. In other words, steering control system 206 causes the necessary activities to adjust the y-axis component of vehicle motion.

The braking system 208 includes at least one device configured to actuate one or more brakes to slow and/or hold the vehicle 200 stationary. In some examples, the braking system 208 includes at least one controller and/or actuator configured to cause one or more calipers associated with one or more wheels of the vehicle 200 to close on a respective rotor of the vehicle 200. Additionally or alternatively, in some examples, the braking system 208 includes an Automatic Emergency Braking (AEB) system and/or a regenerative braking system, or the like.

In some embodiments, the vehicle 200 includes at least one platform sensor (not explicitly illustrated) for measuring or inferring a property of the state or condition of the vehicle 200. In some examples, the vehicle 200 includes platform sensors such as a Global Positioning System (GPS) receiver, an Inertial Measurement Unit (IMU), a wheel speed sensor, a wheel brake pressure sensor, a wheel torque sensor, an engine torque sensor, and/or a steering angle sensor, among others. Although the braking system 208 is shown in fig. 2 as being located proximal to the vehicle 200, the braking system 208 may be located anywhere in the vehicle 200.

Referring now to fig. 3, a schematic diagram of an apparatus 300 is illustrated. As illustrated, the apparatus 300 includes a processor 304, a memory 306, a storage component 308, an input interface 310, an output interface 312, a communication interface 314, and a bus 302. In some embodiments, the device 300 corresponds to at least one device of the vehicle 102, the remote AV system 114, the queue management system 116, the vehicle-to-infrastructure system 118, and/or the network 112. In some embodiments, one or more devices of the vehicle 102, the remote AV system 114, the queue management system 116, the vehicle-to-infrastructure system 118 and/or the network 112, and/or one or more devices of the network 112 (e.g., one or more devices of the system of the network 112) include at least one device 300 and/or at least one component of the device 300. As shown in fig. 3, the apparatus 300 includes a bus 302, a processor 304, a memory 306, a storage component 308, an input interface 310, an output interface 312, and a communication interface 314.

Bus 302 includes components that permit communication between the components of device 300. In some cases, processor 304 includes a processor (e.g., a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), and/or an Acceleration Processing Unit (APU), etc.), a microphone, a Digital Signal Processor (DSP), and/or any processing component that may be programmed to perform at least one function (e.g., a Field Programmable Gate Array (FPGA), and/or an Application Specific Integrated Circuit (ASIC), etc.). Memory 306 includes Random Access Memory (RAM), read Only Memory (ROM), and/or another type of dynamic and/or static storage device (e.g., flash memory, magnetic and/or optical memory, etc.) that stores data and/or instructions for use by processor 304.

The storage component 308 stores data and/or software related to operation and use of the apparatus 300. In some examples, storage component 308 includes a hard disk (e.g., magnetic disk, optical disk, magneto-optical disk, and/or solid state disk, etc.), a Compact Disk (CD), a Digital Versatile Disk (DVD), a floppy disk, a magnetic cassette tape, a magnetic tape, a CD-ROM, RAM, PROM, EPROM, FLASH-EPROM, NV-RAM, and/or another type of computer-readable medium, and a corresponding drive.

Input interface 310 includes components that permit device 300 to receive information, such as via user input (e.g., a touch screen display, keyboard, keypad, mouse, buttons, switches, microphone, and/or camera, etc.). Additionally or alternatively, in some embodiments, the input interface 310 includes sensors (e.g., global Positioning System (GPS) receivers, accelerometers, gyroscopes, and/or actuators, etc.) for sensing information. Output interface 312 includes components (e.g., a display, a speaker, and/or one or more Light Emitting Diodes (LEDs), etc.) for providing output information from device 300.

In some embodiments, the communication interface 314 includes transceiver-like components (e.g., a transceiver and/or separate receivers and transmitters, etc.) that permit the device 300 to communicate with other devices via a wired connection, a wireless connection, or a combination of a wired connection and a wireless connection. In some examples, the communication interface 314 permits the device 300 to receive information from and/or provide information to another device. In some of the examples of the present invention, communication interface 314 includes an ethernet interface, an optical interface, a coaxial interface an infrared interface, a Radio Frequency (RF) interface, a Universal Serial Bus (USB) interface, An interface and/or a cellular network interface, etc.

In some embodiments, the apparatus 300 performs one or more of the processes described herein. The apparatus 300 performs these processes based on the processor 304 executing software instructions stored by a computer readable medium, such as the memory 305 and/or the storage component 308. A computer-readable medium (e.g., a non-transitory computer-readable medium) is defined herein as a non-transitory memory device. Non-transitory memory devices include storage space located within a single physical storage device or distributed across multiple physical storage devices.

In some embodiments, the software instructions are read into memory 306 and/or storage component 308 from another computer-readable medium or from another device via communication interface 314. The software instructions stored in memory 306 and/or storage component 308, when executed, cause processor 304 to perform one or more of the processes described herein. Additionally or alternatively, hardwired circuitry is used in place of or in combination with software instructions to perform one or more processes described herein. Thus, unless explicitly stated otherwise, the embodiments described herein are not limited to any specific combination of hardware circuitry and software.

Memory 306 and/or storage component 308 includes a data store or at least one data structure (e.g., database, etc.). The apparatus 300 is capable of receiving information from, storing information in, communicating information to, or searching information stored in a data store or at least one data structure in the memory 306 or storage component 308. In some examples, the information includes network data, input data, output data, or any combination thereof.

In some embodiments, apparatus 300 is configured to execute software instructions stored in memory 306 and/or a memory of another apparatus (e.g., another apparatus that is the same as or similar to apparatus 300). As used herein, the term "module" refers to at least one instruction stored in memory 306 and/or a memory of another device that, when executed by processor 304 and/or a processor of another device (e.g., another device that is the same as or similar to device 300), causes device 300 (e.g., at least one component of device 300) to perform one or more processes described herein. In some embodiments, the modules are implemented in software, firmware, hardware, and/or the like.

The number and arrangement of components illustrated in fig. 3 are provided as examples. In some embodiments, apparatus 300 may include additional components, fewer components, different components, or differently arranged components than those illustrated in fig. 3. Additionally or alternatively, a set of components (e.g., one or more components) of the apparatus 300 may perform one or more functions described as being performed by another component or set of components of the apparatus 300.

Referring now to fig. 4A, an example block diagram of an autonomous vehicle computation 400 (sometimes referred to as an "AV stack") is illustrated. As illustrated, autonomous vehicle computation 400 includes a perception system 402 (sometimes referred to as a perception module), a planning system 404 (sometimes referred to as a planning module), a positioning system 406 (sometimes referred to as a positioning module), a control system 408 (sometimes referred to as a control module), and a database 410. In some embodiments, the perception system 402, the planning system 404, the positioning system 406, the control system 408, and the database 410 are included in and/or implemented in an automated navigation system of the vehicle (e.g., the autonomous vehicle calculation 202f of the vehicle 200). Additionally or alternatively, in some embodiments, the perception system 402, the planning system 404, the positioning system 406, the control system 408, and the database 410 are included in one or more independent systems (e.g., one or more systems identical or similar to the autonomous vehicle calculation 400, etc.). In some examples, the perception system 402, the planning system 404, the positioning system 406, the control system 408, and the database 41 are included in one or more independent systems located in the vehicle and/or at least one remote system as described herein. In some embodiments, any and/or all of the systems included in autonomous vehicle computing 400 are implemented in software (e.g., software instructions stored in memory), computer hardware (e.g., by microprocessors, microcontrollers, application Specific Integrated Circuits (ASICs), and/or Field Programmable Gate Arrays (FPGAs), etc.), or a combination of computer software and computer hardware. It will also be appreciated that in some embodiments, the autonomous vehicle computing 400 is configured to communicate with a remote system (e.g., an autonomous vehicle system that is the same as or similar to the remote AV system 114, a queue management system 116 that is the same as or similar to the queue management system 116, and/or a V2I system that is the same as or similar to the V2I system 118, etc.).

In some embodiments, the perception system 402 receives data associated with at least one physical object in the environment (e.g., data used by the perception system 402 to detect the at least one physical object) and classifies the at least one physical object. In some examples, perception system 402 receives image data captured by at least one camera (e.g., camera 202 a) that is associated with (e.g., represents) one or more physical objects within a field of view of the at least one camera. In such examples, the perception system 402 classifies at least one physical object based on one or more groupings of physical objects (e.g., bicycles, vehicles, traffic signs, and/or pedestrians, etc.). In some embodiments, based on the classification of the physical object by the perception system 402, the perception system 402 transmits data associated with the classification of the physical object to the planning system 404.

In some embodiments, planning system 404 receives data associated with a destination and generates data associated with at least one route (e.g., route 106) along which a vehicle (e.g., vehicle 102) may travel toward the destination. In some embodiments, the planning system 404 receives data (e.g., the data associated with the classification of the physical object described above) from the perception system 402 periodically or continuously, and the planning system 404 updates at least one trajectory or generates at least one different trajectory based on the data generated by the perception system 402. In other words, the planning system 404 may perform strategic function related tasks required to operate the vehicle 102 in road traffic. Strategic efforts involve maneuvering vehicles in traffic during a journey, including but not limited to deciding whether and when to exceed other vehicles, lane changes, or selecting appropriate rates, accelerations, decelerations, etc. In some embodiments, planning system 404 receives data associated with an updated position of a vehicle (e.g., vehicle 102) from positioning system 406, and planning system 404 updates at least one track or generates at least one different track based on the data generated by positioning system 406.

In some embodiments, the positioning system 406 receives data associated with (e.g., representative of) a location of a vehicle (e.g., the vehicle 102) in an area. In some examples, the positioning system 406 receives LiDAR data associated with at least one point cloud generated by at least one LiDAR sensor (e.g., liDAR sensor 202 b). In some examples, the positioning system 406 receives data associated with at least one point cloud from a plurality of LiDAR sensors, and the positioning system 406 generates a combined point cloud based on each point cloud. In these examples, the positioning system 406 compares the at least one point cloud or combined point cloud to a two-dimensional (2D) and/or three-dimensional (3D) map of the area stored in the database 410. The location system 406 then determines the location of the vehicle in the area based on the location system 406 comparing the at least one point cloud or combined point cloud to the map. In some embodiments, the map includes a combined point cloud for the region generated prior to navigation of the vehicle. In some embodiments, the map includes, but is not limited to, a high-precision map of roadway geometry, a map describing road network connection properties, a map describing roadway physical properties (such as traffic rate, traffic flow, number of vehicles and bicycle traffic lanes, lane width, type and location of lane traffic direction or lane markings, or combinations thereof, etc.), and a map describing spatial locations of roadway features (such as crosswalks, traffic signs or various types of other travel signals, etc.). In some embodiments, the map is generated in real-time based on data received by the perception system.

In another example, the positioning system 406 receives Global Navigation Satellite System (GNSS) data generated by a Global Positioning System (GPS) receiver. In some examples, positioning system 406 receives GNSS data associated with a location of a vehicle in an area, and positioning system 406 determines a latitude and longitude of the vehicle in the area. In such examples, the positioning system 406 determines the location of the vehicle in the area based on the latitude and longitude of the vehicle. In some embodiments, the positioning system 406 generates data associated with the position of the vehicle. In some examples, based on the positioning system 406 determining the location of the vehicle, the positioning system 406 generates data associated with the location of the vehicle. In such examples, the data associated with the location of the vehicle includes data associated with one or more semantic properties corresponding to the location of the vehicle.

In some embodiments, control system 408 receives data associated with at least one trajectory from planning system 404, and control system 408 controls the operation of the vehicle. In some examples, the control system 408 receives data associated with at least one trajectory from the planning system 404, and the control system 408 controls operation of the vehicle by generating and transmitting control signals to operate a powertrain control system (e.g., the DBW system 202h and/or the powertrain control system 204, etc.), a steering control system (e.g., the steering control system 206), and/or a braking system (e.g., the braking system 208). For example, the control system 408 is configured to perform operational functions such as lateral vehicle motion control or longitudinal vehicle motion control. Lateral vehicle motion control causes the required activity to adjust the y-axis component of vehicle motion. Longitudinal vehicle motion control causes the required activity to adjust the x-axis component of vehicle motion. In an example, where the trajectory includes a left turn, the control system 408 transmits a control signal to cause the steering control system 206 to adjust the steering angle of the vehicle 200, thereby causing the vehicle 200 to turn left. Additionally or alternatively, the control system 408 generates and transmits control signals to cause other devices of the vehicle 200 (e.g., headlights, turn signal lights, door locks, and/or windshield wipers, etc.) to change state.

In some embodiments, the perception system 402, the planning system 404, the localization system 406, and/or the control system 408 implement at least one machine learning model (e.g., at least one multi-layer perceptron (MLP), at least one Convolutional Neural Network (CNN), at least one Recurrent Neural Network (RNN), at least one automatic encoder and/or at least one transformer, etc.). In some examples, the perception system 402, the planning system 404, the positioning system 406, and/or the control system 408 implement at least one machine learning model alone or in combination with one or more of the above systems. In some examples, the perception system 402, the planning system 404, the positioning system 406, and/or the control system 408 implement at least one machine learning model as part of a pipeline (e.g., a pipeline for identifying one or more objects located in an environment, etc.). Examples of implementations of the machine learning model are included below with respect to fig. 4B-4D.

Database 410 stores data transmitted to, received from, and/or updated by sensing system 402, planning system 404, positioning system 406, and/or control system 408. In some examples, database 410 includes a storage component (e.g., the same or similar to storage component 308 of fig. 3) for storing data and/or software related to operations and using at least one system of autonomous vehicle computing 400. In some embodiments, database 410 stores data associated with 2D and/or 3D maps of at least one region. In some examples, database 410 stores data associated with 2D and/or 3D maps of a portion of a city, portions of multiple cities, counties, states, and/or countries (states) (e.g., countries), etc. In such examples, a vehicle (e.g., the same or similar vehicle as vehicle 102 and/or vehicle 200) may drive along one or more drivable regions (e.g., single lane roads, multi-lane roads, highways, remote roads, and/or off-road roads, etc.) and cause at least one LiDAR sensor (e.g., the same or similar LiDAR sensor as LiDAR sensor 202 b) to generate data associated with an image representative of an object included in a field of view of the at least one LiDAR sensor.

In some embodiments, database 410 may be implemented across multiple devices. In some examples, database 410 is included in a vehicle (e.g., the same or similar to vehicle 102 and/or vehicle 200), an autonomous vehicle system (e.g., the same or similar to remote AV system 114), a queue management system (e.g., the same or similar to queue management system 116 of fig. 1), and/or a V2I system (e.g., the same or similar to V2I system 118 of fig. 1), etc.

Referring now to FIG. 4B, a diagram of an implementation of a machine learning model is illustrated. More specifically, a diagram illustrating an implementation of Convolutional Neural Network (CNN) 420. For purposes of illustration, the following description of CNN 420 will be with respect to the implementation of CNN 420 by sensing system 402. However, it will be appreciated that in some examples, CNN 420 (e.g., one or more components of CNN 420) is implemented by other systems (such as planning system 404, positioning system 406, and/or control system 408, etc.) other than sensing system 402 or in addition to sensing system 402. Although CNN 420 includes certain features as described herein, these features are provided for illustrative purposes and are not intended to limit the present disclosure.

CNN420 includes a plurality of convolutional layers including a first convolutional layer 422, a second convolutional layer 424, and a convolutional layer 426. In some embodiments, CNN420 includes a sub-sampling layer 428 (sometimes referred to as a pooling layer). In some embodiments, the sub-sampling layer 428 and/or other sub-sampling layers have dimensions that are smaller than the dimensions of the upstream system (i.e., the amount of nodes). By means of the sub-sampling layer 428 having a dimension smaller than that of the upstream layer, the CNN420 merges the amount of data associated with the initial input and/or output of the upstream layer, thereby reducing the amount of computation required by the CNN420 to perform the downstream convolution operation. Additionally or alternatively, CNN420 incorporates the amount of data associated with the initial input by way of sub-sampling layer 428 being associated with (e.g., configured to perform) at least one sub-sampling function (as described below with respect to fig. 4C and 4D).

Based on the perception system 402 providing respective inputs and/or outputs associated with each of the first convolution layer 422, the second convolution layer 424, and the convolution layer 426 to generate respective outputs, the perception system 402 performs convolution operations. In some examples, the perception system 402 implements the CNN420 based on the perception system 402 providing data as input to a first convolution layer 422, a second convolution layer 424, and a convolution layer 426. In such examples, based on the perception system 402 receiving data from one or more different systems (e.g., one or more systems of a vehicle that is the same or similar to the vehicle 102, a remote AV system that is the same or similar to the remote AV system 114, a queue management system that is the same or similar to the queue management system 116, and/or a V2I system that is the same or similar to the V2I system 118, etc.), the perception system 402 provides data as input to the first convolution layer 422, the second convolution layer 424, and the convolution layer 426. The following detailed description of the convolution operation is included with respect to fig. 4C.

In some embodiments, the perception system 402 provides data associated with an input (referred to as an initial input) to a first convolution layer 422, and the perception system 402 generates data associated with an output using the first convolution layer 422. In some embodiments, the perception system 402 provides as input the output generated by the convolutional layers to the different convolutional layers. For example, the perception system 402 provides the output of the first convolution layer 422 as an input to the sub-sampling layer 428, the second convolution layer 424, and/or the convolution layer 426. In such examples, the first convolution layer 422 is referred to as an upstream layer and the sub-sampling layer 428, the second convolution layer 424, and/or the convolution layer 426 are referred to as downstream layers. Similarly, in some embodiments, the perception system 402 provides the output of the sub-sampling layer 428 to the second convolution layer 424 and/or the convolution layer 426, and in this example, the sub-sampling layer 428 will be referred to as an upstream layer and the second convolution layer 424 and/or the convolution layer 426 will be referred to as a downstream layer.

In some embodiments, the perception system 402 processes data associated with the input provided to the CNN 420 before the perception system 402 provides the input to the CNN 420. For example, based on the sensor data (e.g., image data, liDAR data, radar data, etc.) being normalized by the perception system 402, the perception system 402 processes data associated with the input provided to the CNN 420.

In some embodiments, the perception system 402 generates an output based on the CNN 420 performing convolution operations associated with each of the convolution layers. In some examples, CNN 420 generates an output based on the perception system 402 performing convolution operations associated with the various convolution layers and the initial input. In some embodiments, the perception system 402 generates an output and provides the output to the fully connected layer 430. In some examples, the perception system 402 provides the output of the convolutional layer 426 to the fully-connected layer 430, where the fully-connected layer 430 includes data associated with a plurality of characteristic values referred to as F1, F2. In this example, the output of convolution layer 426 includes data associated with a plurality of output characteristic values representing predictions.

In some embodiments, based on the perception system 402 identifying the feature value associated with the highest likelihood as the correct prediction of the plurality of predictions, the perception system 402 identifies the prediction from the plurality of predictions. For example, where fully connected layer 430 includes eigenvalues F1, F2,..fn, and F1 is the largest eigenvalue, perception system 402 identifies the prediction associated with F1 as the correct prediction of the plurality of predictions. In some embodiments, the perception system 402 trains the CNN 420 to generate predictions. In some examples, based on perception system 402 providing training data associated with the predictions to CNN 420, perception system 402 trains CNN 420 to generate the predictions.

Referring now to fig. 4C and 4D, diagrams illustrating example operations of CNN440 utilizing perception system 402 are illustrated. In some embodiments, CNN440 (e.g., one or more components of CNN 440) is the same as or similar to CNN 420 (e.g., one or more components of CNN 420) (see fig. 4B).

At step 450, perception system 402 provides data associated with the image as input to CNN440 (step 450). For example, as illustrated, the perception system 402 provides data associated with an image to the CNN440, where the image is a grayscale image represented as values stored in a two-dimensional (2D) array. In some embodiments, the data associated with the image may include data associated with a color image represented as values stored in a three-dimensional (3D) array. Additionally or alternatively, the data associated with the image may include data associated with an infrared image and/or a radar image, or the like.

In step 455, cnn440 performs a first convolution function. For example, based on CNN440 providing a value representing an image as input to one or more neurons (not explicitly illustrated) included in first convolution layer 442, CNN440 performs a first convolution function. In this example, the value representing the image may correspond to the value of the region (sometimes referred to as receptive field) representing the image. In some embodiments, each neuron is associated with a filter (not explicitly illustrated). The filter (sometimes referred to as a kernel) may be represented as an array of values corresponding in size to the values provided as inputs to the neurons. In one example, the filter may be configured to identify edges (e.g., horizontal lines, vertical lines, and/or straight lines, etc.). In successive convolutional layers, filters associated with neurons may be configured to continuously identify more complex patterns (e.g., arcs and/or objects, etc.).

In some embodiments, CNN 440 performs a first convolution function based on CNN 440 multiplying the value provided as input to each of the one or more neurons included in first convolution layer 442 with the value of the filter corresponding to each of the same one or more neurons. For example, CNN 440 may multiply the value provided as an input to each of the one or more neurons included in first convolutional layer 442 by the value of the filter corresponding to each of the one or more neurons to generate a single value or array of values as an output. In some embodiments, the collective outputs of the neurons of the first convolutional layer 442 are referred to as convolutional outputs. In some embodiments, where the individual neurons have the same filter, the convolved output is referred to as a signature.

In some embodiments, CNN 440 provides the output of each neuron of first convolutional layer 442 to neurons of a downstream layer. For clarity, the upstream layer may be a layer that transfers data to a different layer (referred to as a downstream layer). For example, CNN 440 may provide the output of each neuron of first convolutional layer 442 to a corresponding neuron of the sub-sampling layer. In an example, CNN 440 provides the output of each neuron of first convolutional layer 442 to a corresponding neuron of first sub-sampling layer 444. In some embodiments, CNN 440 adds bias values to the aggregation of all values provided to the various neurons of the downstream layer. For example, CNN 440 adds bias values to the aggregation of all values provided to the individual neurons of first sub-sampling layer 444. In such an example, CNN 440 determines the final value to be provided to each neuron of first sub-sampling layer 444 based on the aggregation of all values provided to each neuron and the activation function associated with each neuron of first sub-sampling layer 444.

At step 460, cnn440 performs a first sub-sampling function. For example, CNN440 may perform a first sub-sampling function based on CNN440 providing the values output by first convolutional layer 442 to the corresponding neurons of first sub-sampling layer 444. In some embodiments, CNN440 performs a first sub-sampling function based on the aggregate function. In an example, the CNN440 performs a first sub-sampling function based on the CNN440 determining the largest input (referred to as the max-pooling function) of the values provided to a given neuron. In another example, the CNN440 performs a first sub-sampling function based on the CNN440 determining an average input (referred to as an average pooling function) in the values provided to a given neuron. In some embodiments, based on CNN440 providing values to the various neurons of first sub-sampling layer 444, CNN440 generates an output, sometimes referred to as a sub-sampled convolutional output.

At step 465, cnn440 performs a second convolution function. In some embodiments, CNN440 performs a second convolution function in a similar manner as how CNN440 performs the first convolution function described above. In some embodiments, CNN440 performs a second convolution function based on CNN440 providing as input the value output by first sub-sampling layer 444 to one or more neurons (not explicitly illustrated) included in second convolution layer 446. In some embodiments, as described above, each neuron of the second convolution layer 446 is associated with a filter. As described above, the filter(s) associated with the second convolution layer 446 may be configured to identify more complex patterns than the filter associated with the first convolution layer 442.

In some embodiments, CNN 440 performs a second convolution function based on CNN 440 multiplying the value provided as input to each of the one or more neurons included in second convolution layer 446 with the value of the filter corresponding to each of the one or more neurons. For example, CNN 440 may multiply the value provided as an input to each of the one or more neurons included in second convolution layer 446 with the value of the filter corresponding to each of the one or more neurons to generate a single value or array of values as an output.

In some embodiments, CNN 440 provides the output of each neuron of second convolutional layer 446 to neurons of a downstream layer. For example, CNN 440 may provide the output of each neuron of first convolutional layer 442 to a corresponding neuron of the sub-sampling layer. In an example, CNN 440 provides the output of each neuron of first convolutional layer 442 to a corresponding neuron of second sub-sampling layer 448. In some embodiments, CNN 440 adds bias values to the aggregation of all values provided to the various neurons of the downstream layer. For example, CNN 440 adds bias values to the aggregation of all values provided to the individual neurons of second sub-sampling layer 448. In such an example, CNN 440 determines the final value provided to each neuron of second sub-sampling layer 448 based on the aggregation of all values provided to each neuron and the activation function associated with each neuron of second sub-sampling layer 448.

At step 470, cnn440 performs a second sub-sampling function. For example, CNN440 may perform a second sub-sampling function based on CNN440 providing the values output by second convolution layer 446 to corresponding neurons of second sub-sampling layer 448. In some embodiments, CNN440 performs a second sub-sampling function based on CNN440 using an aggregation function. In an example, as described above, CNN440 performs a first sub-sampling function based on CNN440 determining the maximum or average input of the values provided to a given neuron. In some embodiments, CNN440 generates an output based on CNN440 providing values to individual neurons of second sub-sampling layer 448.

At step 475, cnn440 provides the output of each neuron of second sub-sampling layer 448 to full connection layer 449. For example, CNN440 provides the output of each neuron of second sub-sampling layer 448 to fully connected layer 449, such that fully connected layer 449 generates an output. In some embodiments, the fully connected layer 449 is configured to generate an output associated with the prediction (sometimes referred to as classification). The prediction may include an indication that an object included in the image provided as input to CNN440 includes an object and/or a set of objects, etc. In some embodiments, the perception system 402 performs one or more operations and/or provides data associated with predictions to the different systems described herein.

Referring now to fig. 5-9, diagrams of example implementations of a machine learning model for fusing composite images from a perception system (such as perception system 402) and data from additional sensor modalities (such as images from camera 202a, etc.) for purposes such as object motion prediction or vehicle planning are shown. In particular, FIG. 5 depicts a first embodiment of a machine learning model that fuses the output of the sensing system and data from the additional sensor modalities, wherein features learned from the data of the additional sensor modalities and the output from the sensing system are linked and processed by a neural network for prediction or planning. Fig. 6 and 7 depict a second embodiment of a machine learning model that fuses the output of the sensing system and data from the additional sensor modalities, wherein a combined image combining both the output of the sensing system and the annotation drawn from the output learned from the data of the additional sensor modalities is processed by the neural network for prediction or planning. Fig. 8 and 9 depict a third embodiment of a machine learning model that fuses the output of the sensing system and data from additional sensor modalities, wherein the characterized annotation to the output of the sensing system is learned from the data from the additional sensor modalities while training the model for prediction or planning. Various embodiments are described in turn.

As shown in fig. 5, the machine learning model 500 takes as input both the output of the sensing system (such as the sensing system 402, etc.) and data from the additional sensor modalities. In fig. 5, the output of the perception system is a composite BEV image 502. For example, the image may depict the environment surrounding the vehicle, objects within the environment (e.g., as shapes within the image), and computed or detected trajectories of the objects. For example, the shape and trajectory may be color coded to indicate attributes of the shape or trajectory (such as an intended object category (e.g., pedestrian, bicycle, car, truck, etc.), velocity, pose, or acceleration, etc.), certainty regarding the assigned attributes, and so forth. Thus, the BEV image 502 may represent an understanding generated by the perception system of the environment. In fig. 5, the data from the additional sensor modality is the raw image 504 from the camera. Illustratively, where the model 500 is used to predict the motion of an object indicated within the BEV image 502, the original image 504 may depict a view of the vehicle in the direction of the object, and thus the object. Although only one original image 504 is shown in fig. 5, in some embodiments, the model 500 may accept multiple images. For example, where the model 500 is used to plan motion for a vehicle, the model 500 may accept camera images from multiple viewpoints (e.g., front, rear, side, etc.) of the vehicle. Furthermore, while in fig. 5 a camera image is used as an example of an additional sensor modality, data from other sensor modalities may additionally or alternatively be used. For example, the raw image 504 may be replaced with data from a radar sensor 202c, a microphone 202d, or other sensor modalities.

In fig. 5, the composite image 502 and the original image 504 are fed through a filter denoted CNN _s And CNN _c Is provided. CNNs each function similarly to the networks described in fig. 4B-4D. Illustratively, each CNN is operable to take a respective image as input and provide a learned feature of the image as output, where the feature contributes to the output of the machine learning model 500 (e.g., the predicted motion of the object or the planning action of the vehicle). Thus, CNN _s Outputting the set of composite image features 506, and CNN _c A set of original image features 508 is output.

The corresponding features 506 and 508 are then coupled to a status input 510. The particular state input 510 may vary, for example, depending on the desired output of the model 500. For example, where model 500 is used to predict the motion of an object in the environment of a vehicle, state input 510 may reflect a known or estimated state of the object. Where model 500 is used to plan the actions of a vehicle, state input 510 may reflect the state of the vehicle, or the state of one or more objects in the environment of the vehicle, or a combination thereof. An example of state information such as velocity, acceleration, yaw rate, etc. is depicted in fig. 5. Other non-limiting examples of status information are pitch, roll, attitude (e.g., as a combination of pitch, roll, and yaw), and speed (e.g., including a direction vector). For the state of the vehicle, the state information may include additional data such as steering wheel angle, braking force, engine power, traction data, and the like.

The concatenated data is then fed to a generator, which generates a set of trajectories 512 that may represent a set of possible trajectories based on the concatenated data. Illustratively, where the model 500 is used to predict the motion of an object in the environment of a vehicle, the set 512 may reflect the potential trajectories of the object. Where model 500 is used to plan the actions of a vehicle, collection 512 may include potential trajectories of the vehicle. Track set 512 may illustratively be generated based on state information 510, such as by applying a potential set of modifications (e.g., deceleration, acceleration, and steering) to state information 510 to generate possible tracks. The trajectory set 512 may thus represent potential outputs of the model 500. In some embodiments, the track set may be replaced with other outputs such as possible track-independent actions of the vehicle (e.g., braking, steering, accelerating, etc.).

To distinguish between potential outputs, the join data is further fed through a dense layer to generate a set 514 of pattern probabilities representing probabilities of respective potential outputs in the set 512. For example, the dense layer may operate to evaluate each potential output in the set 512 from the join data and assign probabilities that the potential outputs are correct. Thus, the model 500 may select the output of the set 512 (e.g., the potential output with the highest probability) as the output of the model 500.

Although the track set 512 is shown in fig. 5 as being generated via a generator, in some embodiments the track set 512 may alternatively be generated via a dense layer (e.g., along with the pattern probabilities 514).

Training of model 500 may occur based on collected data reflecting the correct output of model 500. For example, when applied to motion prediction for an object, a training data set may be generated, where the training data set includes a composite image for identifying the object, a camera image for depicting the object, and object motion observed after the generation of the composite image and the capture of the camera image. For example, the observed motion may be observed based on subsequent sensor data (e.g., subsequent lidar data indicating movement of the object). Model 500 may then be trained such that model 500 produces an output that matches the observed motion. When applied to planning for a vehicle, the training data set may include a composite image of the environment of the vehicle, one or more camera images depicting the environment, and observed movements of a skilled human operator. Model 500 may then be trained such that model 500 produces an output that matches the observed actions of the human operator.

Thereafter, the trained model 500 may be applied during operation of the vehicle. The application of the trained model 500 is sometimes referred to as "reasoning" and the trained model 500 is utilized to produce an unknown output such as a predicted motion of the object (where the motion is not yet known) or a planning motion of the vehicle (where the motion is not yet known). For example, by entering the composite image 502, one or more camera raw images 504 of the object, and state information 510 of the object generated by a perception system of the vehicle (e.g., the perception system 402), the trained model 500 may be used to predict the motion of the object around the vehicle during non-training operation of the vehicle, thereby producing as output object motion predicted to match the actual motion of the object. Similarly, the trained model 500 may be used to plan the motion of the vehicle by importing as output a composite image 502 generated by the perception system of the vehicle (e.g., the perception system 402), one or more camera raw images 504 of the vehicle's surroundings, and state information 510 of objects in the vehicle and/or surroundings, thereby producing a vehicle motion that is predicted to match the motion of a skilled human driver.

As discussed above, by combining the composite image 502 with data of additional sensor modalities, such as the original image 504 of the camera, the model 500 may consider additional contextual information within the data of the additional sensor modalities, where such information may be difficult or impossible to directly incorporate into the composite image 502 without negatively affecting the accuracy of the composite image 502. For example, while the composite image 502 may include a predicted trajectory of the object, the image 502 may not have been generated based on data of additional sensor modalities, and thus may lack contextual information such as the presence of signal lights, limb positions, etc., which would otherwise be used by a human to predict motion and/or planning actions. Thus, the inclusion of this data of additional sensor modalities may significantly improve motion prediction or planning. In some embodiments, model 500 may thus be implemented as part of planning system 404. In other embodiments, the model 500 may be incorporated into the perception system 402. For example, the system 402 may update the composite image 502 with the predicted motion of the object as the model 500 output.

While the application of model 500 in motion prediction is discussed above with respect to a single object, the models disclosed herein may additionally or alternatively be applied to predicting motion of multiple objects simultaneously. Such a configuration may enable the model to capture relevant context information for multiple objects. For example, such a configuration may enable the model to more accurately predict how signals for one object modify the motion of other objects (e.g., how the presence of a brake light or turn signal on one vehicle may modify the motion of other vehicles).

As described above, fig. 6-9 depict additional embodiments of machine learning models that fuse the output of the sensing system with data from additional sensor modalities. Additional embodiments include some elements that are similar or identical to the embodiment of fig. 5, and thus a description of these elements will not be repeated with respect to fig. 6-9. Reference numerals of fig. 5 are repeated in fig. 6-9 to indicate similar or identical elements to the embodiment of fig. 5.

With reference to fig. 6-7, additional embodiments of machine learning models that fuse the output of the sensing system with data from additional sensor modalities will be described. In particular, the models of fig. 6-7 are aggregated models including the first model 600 of fig. 6 and the second model 700 of fig. 7, wherein the first model 600 is used to generate a rendered annotation of a composite image based on data from one or more additional sensor modalities and the second model 700 is used to make motion prediction and/or planning based on the output of the first model 600.

In contrast to the model 500 of FIG. 5, the aggregate model of FIGS. 6-7 may provide intermediate representations of data from additional sensor modalities as labels "drawn" on the composite image. In this context, "rendering" refers to adding metadata to a composite image that is used to indicate the properties of objects within the image. For example, the metadata may indicate that the vehicle has turned on a turn signal or brake light, that the pedestrian has assumed a particular posture or is facing in a certain direction, and so on. Instead of providing the original camera image for processing, these rendering annotations may be combined with the composite image for processing over a network to generate an output. For example, where the BEV composite image is represented as a set of channels, the callout may be represented as an additional channel of each potential feature (e.g., brake lights, signal lights, pedestrian pose, pedestrian facing direction, etc.).

As in fig. 5, to produce the rendering annotations, the composite image 502 and the original image 504 of the camera are fed through a respective neural network CNN _s And CNN _c . In one embodiment, the corresponding neural network may be the same as the neural network of fig. 5. In other embodiments, the attributes of the network may differ depending on the function of the network to generate the draw labels. For example, the super parameters of the network may be different when used to generate the drawing annotations.

As in fig. 5, the corresponding network generates a feature set: specifically, the composite image 502 is converted into a composite image feature set 506, and the camera raw image 504 is converted into a raw image feature set 508. Features 506 and 508 may then be joined. However, unlike fig. 5, the join data is then fed through a decoder for producing a combined image 602 that combines the composite image 502 with the set of rendering annotations learned based on the join data. (although not explicitly shown in FIG. 6, the composite image 502 may also be fed to the decoder in a manner similar to the U-net architecture.)

The model 600 of fig. 6 may be trained based on a manually generated set of drawn annotations. For example, a human may annotate the set of synthetic images 502 based on the respective original images 504 to produce a manually created combined image 602. Thereafter, the model 600 may be trained to create a combined image 602 from the set of composite images 502 and corresponding original images 504 that corresponds to the manually created combined image 602. During inference, the model 600 may thus generate a new combined image 602 based on the new composite image features 506 and the original image features 508. Although a single model is shown in fig. 6, in some embodiments, multiple instances of model 600 may be created. For example, each instance may generate a particular type of annotation (e.g., turn signal, brake light, pedestrian gesture, etc.), and the annotations of each model may be combined into a combined image 602 having multiple types of annotations.

The combined image 602 generated by the model 600, which fuses the composite image 502 and the original image 504 of the camera, may then be used as an input to the second machine learning model 700. Specifically, as shown in fig. 7, the combined image 602 may be fed through a convolutional neural network to generate an image feature set 704. Image features 704 may then be concatenated with state information 510 and fed into a generator and dense layer to produce a set of trajectories 512 and pattern probabilities 514 in a manner similar to the set of trajectories 512 and pattern probabilities 514 of FIG. 5. As described above with respect to fig. 5, the combination of the set 512 and the probabilities 514 may indicate an output of the model 700, wherein the output may represent, for example, a predicted motion of the object, a planning action of the vehicle, or a combination thereof. Model 700 may be trained similar to model 500 of fig. 5, as modified to reflect the input of combined image 602, rather than the input of composite image 502 and original image 504. During inference, models 600 and 700 may be implemented in combination as an aggregate machine learning model for planning or motion prediction purposes. For example, the new composite image 502 and the original image 504 (e.g., not previously included in the training dataset) may be fed into the model 600 to produce a combined image 602, where the combined image 602 may then be fed into the model 700 to produce the predicted motion or planning action.

Thus, models 600 and 700 may provide outputs similar to model 500 of fig. 5, enabling the vehicle to utilize contextual information from additional sensor modalities that may otherwise be lost in generating composite image 502. However, while FIG. 5 produces an output from a concatenation of composite image and raw image data, FIGS. 6-7 provide for the separate creation of rendering annotations, which may enable the aggregate model to be more specific to certain signals as controlled, for example, by training of model 600. For example, the model 600 may be specially trained to recognize certain signals, such as brake lights, turn signals, etc., which may not be recognized during the more general training of the model 500.

Referring to fig. 8-9, yet another embodiment of a machine learning model that fuses the output of the sensing system and data from additional sensor modalities will be described. In particular, fig. 8 depicts a preprocessing pipeline 800 that pertains to how information from the BEV composite image 502 may be used to preprocess data, such as the camera image 504, from additional sensor modalities, prior to use in a machine learning model. Fig. 9 depicts how such preprocessing is incorporated into a model 900 to learn annotations in connection with desired outputs such as motion prediction or planning.

Referring to fig. 8, a preprocessing pipeline 800 is shown that enables data, such as camera image 504, from additional sensor modalities to be preprocessed to reduce the amount of information that must be processed in conjunction with the data, which can increase both the rate of a trained model (e.g., by reducing the data to be processed) and the accuracy of the model (e.g., by focusing on related data within a larger data set). In particular, while the above description relates to an original camera image for depicting an object, it is possible that most or all of the information related to the object is contained within a particular portion of the image. For example, camera images for capturing other vehicles may also capture various objects in addition to other vehicles. The portions of the image that capture these other objects may not be related to the signals provided by the vehicle and may actually prevent the machine learning model from learning such signals related to the vehicle. For example, portions of the image depicting other objects may prevent the machine learning model from detecting whether a particular vehicle has illuminated a brake light of the vehicle, turned on a turn signal, and so forth.

Thus, in some embodiments, the composite image 502 or data from a perception system such as the perception system 402 may be used to crop the original camera image 504 to produce a cropped image 802, which may be used in place of the image 504 for machine learning applications. For example, the composite image 502 or other data from the perception system may identify the location of the object depicted in the original camera image 504. Thus, these sites may be projected onto the original camera image 504 to locate the object within the original camera image 504 and produce a cropped image that separates the object from within the original camera image. When training a model to make motion predictions for an object, an image of the object after cropping, rather than a complete image of a camera that otherwise depicts the object, may be used as an input to the model. Similarly, when training a model for motion planning of an autonomous vehicle, a set of cropped images for each object in the vicinity of the vehicle may be used instead of complete images from each associated camera.

To produce the cropped image 802, information contained within the composite image 502 may be projected onto the camera image 504. For example, where the composite image 502 indicates that a particular object is located at a position relative to the vehicle, that position may be projected onto the camera image 504 as the intended position of the particular object. Image 504 may then be cropped to contain the location (e.g., include a buffer around the location) to produce cropped image 802. Thus, the cropped image 802 may represent the portion of the original camera image 504 that contains the relevant object.

Thereafter, in the machine learning model disclosed herein, the cropped image 802 may be used in place of the original image 504. For example, the cropped image may be represented by CNN in fig. 8 _R To produce a learned feature set 806 for the cropped image 802, wherein the learned feature set 806 may then be used to produce annotations 810 drawn on the BEV image 502.

One example machine learning model architecture that utilizes the crop of FIG. 8 is shown in FIG. 9, where FIG. 9 depicts a model 900 that learns annotations in conjunction with desired output, such as motion prediction or planning. As shown in FIG. 9, the model 900 will be created as discussed in FIG. 8 The constructed cropped image 802 is used as an input. The cropped image 802 may be represented by CNN in fig. 9 _R To produce learned features 806 for the cropped image 802. These learned features 806 may then be passed through different dense networks to produce signal probabilities 902 and reduced features 904, respectively. In fig. 9, signal probabilities 902 illustratively represent probabilities of various learned signals present within cropped image 802. As described below, the probabilities 902 of these learned signals may be used in conjunction with the trajectory pattern probabilities 514 to train the model 900.

The reduced features 904 may then be used to perform feature rendering of the composite image 502 to produce the combined image 602. Feature rendering may occur similarly to that discussed above with respect to fig. 6, such as by passing the reduced features 904 and the composite image 502 through a decoder to generate the composite image 602, and so on. Although the combined image 602 is shown in fig. 9 as being generated based on the composite image 502 and the reduced feature 904, in some examples, the combined image 602 may be generated based on the composite image 502 and the learned feature 806, omitting the dense network between the learned feature 806 and the reduced feature 904.

Thereafter, the combined image 602 may be used with the state information 510 to generate a set of trajectories and trajectory pattern probabilities 514 in a manner similar to that discussed above with respect to fig. 7.

In fig. 9, the relevant features of the cropped image 802 may be learned during training of the model 900 by establishing a loss function of the network that combines the signal probabilities 902 learned from the cropped image 802 with the pattern probabilities 514. For example, the penalty of model 900 may be equal to the penalty of pattern probability 514 plus the sum of the penalty of each signal within signal probability 902 (e.g., where each penalty is calculated as a cross entropy penalty). In some examples, the sum of the losses of the individual signals may be weighted by an adjustable super parameter. In other embodiments, the loss function may be based on a pattern probability 514 that is independent of the signal probability 902.

Thus, the model 900 may be trained to learn features within the data of the additional sensor modalities and to make motion predictions or vehicle plans based on those learned features. Thus, the output of model 900 may be similar to model 700 of FIG. 7, without relying on manual labeling of composite image 502. This may enable the model 900 to capture signals that would not otherwise be captured during manual annotation, thereby increasing the overall accuracy of the model 900.

Referring now to fig. 10 and 11, a flow chart of a process for fusing the output of the sensing system and data of additional sensor modalities for purposes such as motion prediction and planning is shown. Specifically, FIG. 10 depicts a routine 1000 for training a machine learning model to fuse the output of the perception system and data of additional sensor modalities, while FIG. 11 depicts a routine 1100 for utilizing the trained machine learning model to predict object motion or to plan the motion of a vehicle. Routines 1000 and 1100 may be implemented, for example, by apparatus 300. Illustratively, the routines 1000 and 1100 may be implemented by the apparatus 300 included within the autonomous vehicle 200. In some cases, routine 1000 may be implemented by a device 300 external to autonomous vehicle 200, such as queue management system 116, and then the trained model generated via routine 1000 may be loaded onto device 300 of autonomous vehicle 200 for use during operation of vehicle 200.

Routine 1000 begins at block 1002, where device 300 obtains a composite image for identifying objects within the environment of a vehicle. For example, the image may be a BEV image of the area generated based on sensor data of the vehicles in the area. The identified objects may include any object sensed by the vehicle, such as other motorized vehicles, non-motorized vehicles (such as bicycles, etc.), pedestrians, etc. For example, the image may be generated based on a perception system of the vehicle, such as perception system 402. In one embodiment, the perception system 402 generates a composite image based on lidar data that indicates that the object is in the region.

Additionally, at block 1004, the apparatus 300 obtains a camera image for depicting the object in the composite image. For example, the image may be an image of a camera pointing in the direction of the object from the viewpoint of the vehicle. In one embodiment, the composite image and the camera image are arranged into data pairs, wherein each pair comprises first data corresponding to the composite image and second data corresponding to the camera image, the camera image being obtained simultaneously with the sensor data used to generate the composite image. For example, data pairs may be collected during operation of one or more test vehicles under the operation of a skilled human operator. While fig. 10 discusses a camera image as an example of data for an additional sensor modality, other additional sensor modalities may additionally or alternatively be used.

Thereafter, at block 1006, the apparatus 300 trains a machine learning model based on the fusion of the composite image and the camera image for planning or prediction purposes. The trained model may correspond to the model described above with reference to fig. 5-9. As discussed for example with respect to fig. 5, the model may include, for example, a first convolutional neural network for processing first data of a given data pair, a second convolutional neural network for processing second data of the given data pair, and a dense layer for generating predicted motion of an object corresponding to the given data pair. As discussed above, the model may take as input additional information such as status information of one or more objects identified within the composite image (e.g., captured concurrently with the data used to generate the composite image). For example, the model may take as input state information including velocity, acceleration, or pose of the object. As another example, the model may include a convolutional neural network that takes as input first data corresponding to a composite image of a given data pair, labeled according to second data corresponding to a camera image of the given data pair. The callout may include any metadata related to an object such as a lighted brake light, a lighted turn signal, a wheel position, a limb position, a joint position, and the like. Labeling may occur, for example, based on the application of a second machine learning model, such as model 600 of fig. 6. Alternatively, labeling may occur based on the application of other convolutional neural networks that produce the labeling, where the other neural networks are trained simultaneously with the first convolutional neural network (e.g., as discussed with reference to FIG. 9). As discussed above, the machine learning model may accept camera data as raw images in some instances. Additionally or alternatively, the model may accept the cropped camera data, such as by projecting the object indicated within the composite image into the raw camera data and cropping the raw camera data around the object to produce cropped camera data, or the like.

As described above, the machine learning model may be trained for purposes such as object motion prediction or vehicle motion planning. For example, where a model is trained based on the motion of an observed object captured within a data pair, the model may be transmitted to a destination vehicle for predicting the motion of an additional object sensed by the vehicle. In the case of training a model based on observed movements of a skilled human operator, the model may be transmitted to the destination vehicle for planning autonomous movements of the destination vehicle. In some instances, routine 1000 may be implemented multiple times to generate multiple machine learning models, such as a first model for object motion prediction and a second model for motion planning.

As described above, FIG. 11 depicts a routine 1100 for predicting object motion or planning vehicle actions using a trained machine learning model, where the routine 1100 may be implemented within the vehicle 200, for example.

Routine 1100 begins at block 1102, where the vehicle 200 obtains a trained machine learning model, such as the model generated via routine 1000 of FIG. 10. As discussed with respect to fig. 10, the model may be trained for specific purposes such as object motion detection or motion planning.

Thereafter, at blocks 1104 and 1106, the vehicle 200 obtains input data input to the trained model. Specifically, at block 1104, the vehicle 200 obtains a composite image of an area of the vehicle, wherein the image identifies objects (such as other vehicles, pedestrians, etc.) in the area of the vehicle. Additionally, at block 1106, the vehicle 200 obtains a camera image for depicting the object, wherein the image may be captured, for example, by a camera of the vehicle 200 pointing in the direction of the object. The composite image and the camera image may be captured in the same or similar manner as the data pairs used to train the machine learning model are captured. For example, the vehicle 200 may be the same as or have a similar configuration as the vehicle that captured the data used to train the model.

In some embodiments, the trained model may use additional information as input, such as state information of the object or state information of the vehicle. Thus, such state information may also be obtained at the vehicle in a manner similar to how such state information is obtained for training the model.

Thereafter, at block 1108, a trained machine learning model is applied to the composite image and the camera image (and potentially additional inputs such as state information). For example, the images may be passed through one or more trained convolutional neural networks and/or dense layers to generate a set of trajectory pattern probabilities for indicating a predicted motion of the object or a planned trajectory of the vehicle. Thus, at block 1110, the model may output the predicted motion or planned route. This output can then be used to control subsequent operations of the vehicle. For example, the predicted motion of the object or the planned route may be used as input to a planning system (e.g., planning system 404) to control subsequent movements of the vehicle. As discussed above, because data of additional sensor modalities may capture contextual information, such as the presence of light, sound, gestures, etc., that is not otherwise contained within the composite image, and because such data is captured within the trained model of routine 1100, the predicted motion or planned route may have greater accuracy than an alternative predicted or planned route. For example, routine 1100 may enable vehicle 200 to more accurately predict braking of other vehicles due to the presence of brake lights, lane changes of other vehicles due to the presence of turn signals, movement of pedestrians into crosswalks based on pose or facing directions, and the like. Similarly, assuming that such a skilled human operator may consider signals such as those described above, routine 1100 may enable vehicle 200 to more accurately simulate the actions of the skilled human operator. Thus, routine 1100 may provide a safer and more accurate autonomous vehicle.

In the foregoing specification, aspects and embodiments of the disclosure have been described with reference to numerous specific details that may vary from implementation to implementation. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense. The sole and exclusive indicator of the scope of the application, and what is intended by the applicants to be the scope of the application, is the literal and equivalent scope of the claims, including any subsequent amendments, that are issued from the present application in the specific form of the claims. Any definitions expressly set forth herein for terms contained in such claims shall govern the meaning of such terms as used in the claims. In addition, when the term "further comprises" is used in the preceding description or the appended claims, the phrase may be followed by additional steps or entities, or sub-steps/sub-entities of the previously described steps or entities.

Claims

1. A computer system, comprising:

one or more computer-readable storage devices configured to store computer-executable instructions; and

one or more computer processors configured to execute the computer-executable instructions, wherein execution of the computer-executable instructions causes the computer system to:

Obtaining a set of data pairs, each data pair comprising:

first data corresponding to a composite image representing a bird's eye view of an area generated based on sensor data of a vehicle in the area, wherein the composite image identifies an object in the area; and

second data corresponding to a camera image representing a viewpoint of the vehicle in the area, wherein the camera image depicts the object;

training a machine learning model based on the set of data pairs to produce a trained model, wherein the machine learning model includes at least one convolutional neural network to process the set of data pairs, and wherein the machine learning model accepts as input a given data pair of the set of data pairs and provides as output a predicted motion of an object corresponding to the given data pair; and

transmitting the trained model to a destination vehicle, wherein the destination vehicle is configured to apply the trained model to sensor data of the destination vehicle to predict movement of a target object identified within the sensor data.

2. The computer system of claim 1, wherein the at least one convolutional neural network comprises a first convolutional neural network for processing first data of the given data pair, a second convolutional neural network for processing second data of the given data pair, and a dense layer for generating a predicted motion of an object corresponding to the given data pair.

3. The computer system of claim 2, wherein the first convolutional neural network generates a composite image feature, wherein the second convolutional neural network generates a camera image feature, and wherein the dense layer takes as input a concatenation of the composite image feature and the camera image feature and outputs a predicted motion of an object corresponding to the given pair of data.

4. A computer system according to claim 3, wherein the dense layer further takes as input state information relating to an object corresponding to the given pair of data and outputs a predicted motion of the object corresponding to the given pair of data.

5. The computer system of claim 4, wherein the status information includes one or more of a velocity of an object corresponding to the given data pair, an acceleration of an object corresponding to the given data pair, and a pose of an object corresponding to the given data pair.

6. The computer system of claim 1, wherein the at least one convolutional neural network is a first convolutional neural network, wherein the first convolutional neural network takes as input first data corresponding to a composite image of the given data pair, labeled from the second data corresponding to a camera image of the given data pair.

7. The computer system of claim 6, wherein the at least one convolutional neural network generates an output that is provided to a dense layer to provide a predicted motion of an object corresponding to the given pair of data.

8. The computer system of claim 6, wherein the first data is annotated to indicate at least one of a brake light illuminated, a turn signal illuminated, a wheel position, a limb position, and a joint position on the subject of the given data pair.

9. The computer system of claim 6, wherein the first data is annotated via application of a second machine learning model to the first data of the given data pair and the second data of the given data pair.

10. The computer system of claim 6, wherein the machine learning model includes a second convolutional neural network to generate annotations to the first data of the given data pair.

11. The computer system of claim 10, wherein the first convolutional neural network is trained simultaneously with the second convolutional neural network during training of the machine learning model.

12. The computer system of claim 11, wherein the second convolutional neural network outputs to a dense layer for providing signal probabilities, and wherein the first convolutional neural network is trained with a loss function, wherein the first convolutional neural network takes signal probabilities as inputs and outputs predicted motion of an object corresponding to the given data pair.

13. The computer system of claim 10, wherein the second convolutional neural network accepts as input a cropped portion of the second data of the given data pair, the cropped portion selected based on a position of an object corresponding to the given data pair within the second data of the given data pair, the position within the second data of the given data pair indicated by a position of an object corresponding to the given data pair within the first data of the given data pair.

14. The computer system of claim 1, wherein execution of the computer-executable instructions further causes the computer system to:

Training a second machine learning model based on the set of data pairs to produce a second trained model, wherein the second machine learning model includes at least one convolutional neural network to process the set of data pairs, and wherein the second machine learning model accepts as input a given data pair of the set of data pairs and provides as output a planning motion of a vehicle corresponding to the given data pair; and

transmitting the second trained model to a destination vehicle, wherein the destination vehicle is configured to apply the second trained model to sensor data of the destination vehicle to plan movement of the destination vehicle.

15. The computer system of claim 1, wherein the object within each data pair of the set of data pairs is at least one of a pedestrian, a motor vehicle, and a bicycle.

16. The computer system of claim 1, wherein the sensor data of the vehicle used to generate the composite image comprises at least one of lidar data associated with a point cloud and radar data associated with a radar image.

17. A computer-implemented method, comprising:

obtaining a set of data pairs, each data pair comprising:

18. The computer-implemented method of claim 17, further comprising:

19. One or more non-transitory computer-readable media comprising computer-executable instructions that, when executed by a computing system comprising a processor, cause the computing system to:

obtaining a set of data pairs, each data pair comprising:

20. The one or more non-transitory computer-readable media of claim 19, wherein the computer-executable instructions further cause the computing system to: