Hybrid data driven actor prediction using reachable set and supervised learning

Abstract

A method for predicting occupancy of a corridor by an actor includes: receiving input data; predicting an occupancy set of the actor using the reachability analysis and the input data; using the machine learning model and the input data to determine a corridor occupation constraint condition of the actor; and determining the occupancy corridor of the actor using the occupancy corridor constraints of the actor using the machine learning model and the occupancy set of the actor using the strength accessibility analysis. Further, the method includes controlling movement of the host vehicle based on the occupancy corridor of the actor.

Description

Technical Field

[0001] The present disclosure relates to methods and systems for actor prediction, and more particularly, to a hybrid data-driven actor prediction system and method using reachable sets and supervised learning. Background Art

[0002] This introduction generally presents the background of the present disclosure. To the extent described in this introduction, the work of the presently named inventors and aspects of the specification that may not constitute prior art at the time of filing are neither expressly nor impliedly admitted to be prior art to the present disclosure.

[0003] Self-driving cars use models to predict the locations of other actors on the road. State-of-the-art (SOTA) actor prediction models are data-driven prediction models. These data-driven prediction models are not necessarily robust and output a limited number of hypotheses that do not cover the space of possible future locations of the actors well. Other approaches include actor prediction based on reachability theory, which tends to be overly conservative. Specifically, reachability-based actor prediction models are mostly useful in the short term and do not scale well in complex, cluttered, and congested driving environments due to their conservative nature. Therefore, there is a need to develop a method and system for predicting the location or occupancy of an actor on a map that is conservative in the short term. Summary of the invention

[0004] The present disclosure describes a method for predicting an actor's occupied corridor. The method includes receiving input data. The input data includes map data and initial trajectory data. The map data includes a map of an area around an actor. The initial trajectory data includes an initial trajectory of the actor. The method also includes: using an exact or approximate accessibility analysis (such as Hamilton-Jacobi accessibility, a sampling-based optimal control method, a polyhedron or polyhedron set-based method, regional accessibility, ellipsoidal accessibility, etc.) and the input data to predict multiple occupancy sets of the actor. In addition, the method includes: using a machine learning model and input data to determine the actor's occupied corridor; using the actor's occupied corridor constraints using the machine learning model and the actor's occupied corridor using the accessibility theory to determine the actor's occupied corridor. In addition, the method includes: controlling the movement of the main vehicle based on the actor's occupied corridor. The main vehicle is an autonomous vehicle. The method described in this paragraph improves autonomous vehicle technology by accurately and reasonably predicting the position of actors on the road, thereby allowing the main vehicle to fully maneuver to avoid contact with the actor.

[0005] In some aspects of the present disclosure, the method may include: receiving perception data from a plurality of sensors of a host vehicle. The perception data includes information about an area around an actor. In addition, the method includes fusing the perception data with map data to generate scene data. The scene data includes a nominal route path and a road centerline of a road. The actor is moving on the road. The initial trajectory data includes a longitudinal acceleration and a lateral acceleration of the actor, the reachability analysis uses the longitudinal acceleration and the lateral acceleration of the actor and an assumption model, including a kinematic model, a dynamic model, or a machine learning model, and the actor's occupancy set is a set of forward reachable positions within a predetermined time. The method also includes: determining that the actor's forward reachable set does not intersect with other traffic actors within a predetermined time period, and determining that the actor's forward reachable set is within an allowed road position. Using the actor's output acceleration, an occupancy corridor is determined. The actor's occupancy corridor constraint includes a maximum acceleration of the actor and a minimum acceleration of the actor. The method may include: determining that the trajectory of the host vehicle intersects with the actor's occupancy corridor. The method may include: removing the trajectory of the host vehicle that intersects with the actor's occupancy corridor. The method may include determining that a trajectory of the host vehicle does not intersect an occupied corridor of the actor. The method may include controlling the host vehicle to move along a trajectory of the host vehicle that does not intersect an occupied corridor of the actor. Implementations of the described techniques may include hardware, methods or processes, or computer software on a computer accessible medium.

[0006] The present disclosure also describes a system for predicting actor-occupied corridors. The system includes a sensor and a controller in communication with the sensor. The controller includes a processor and a non-transitory computer-readable medium and is programmed to execute the above method.

[0007] The present disclosure also describes a tangible, non-transitory machine-readable medium comprising machine-readable instructions that, when executed by a processor, cause the processor to perform the above-described method.

[0008] Further areas of applicability will become apparent from the detailed description provided hereinafter.It should be understood that the description and specific examples are intended for purposes of illustration only and are not intended to limit the scope of the present disclosure.

[0009] The above features and advantages and other features and advantages of the presently disclosed systems and methods are apparent from the detailed description, including the claims and exemplary embodiments, when taken in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The present disclosure will be more fully understood from the detailed description and accompanying drawings provided below, in which:

[0011] Figure 1is a block diagram depicting a host vehicle including a hybrid data-driven actor prediction system using reachable sets and supervised learning.

[0012] Figure 2 Schematic diagram of a host vehicle and actors moving along a road.

[0013] Figure 3 is a flowchart of a hybrid data-driven actor prediction approach using reachable sets and supervised learning.

[0014] Figure 4 is a flowchart of a method for actor prediction using reachable set computation.

[0015] Figure 5 is a flowchart of a method for training a reasoner. DETAILED DESCRIPTION

[0016] Reference will now be made in detail to several examples of the present disclosure, which are illustrated in the accompanying drawings. Whenever possible, the same or similar reference numerals are used in the drawings and the description to refer to the same or similar parts or steps.

[0017] Reference Figure 1 , the main vehicle 10 generally includes a chassis 12, a body 14, front wheels and rear wheels 17 and can be referred to as a vehicle system. In the depicted embodiment, the main vehicle 10 includes two front wheels 17a and two rear wheels 17b. The body 14 is arranged on the chassis 12 and substantially surrounds the components of the main vehicle 10. The body 14 and the chassis 12 can together form a frame. Each wheel 17 is rotatably coupled to the chassis 12 near a corresponding corner of the body 14. The main vehicle 10 includes a front axle 19 coupled to the front wheels 17a and a rear axle 25 coupled to the rear wheels 17b.

[0018] In various embodiments, the host vehicle 10 may be an autonomous vehicle, and the system 98 is incorporated into the host vehicle 10. The system 98 may be referred to as a hybrid data-driven actor prediction system using reachable sets and supervised learning. The host vehicle 10 is, for example, a vehicle that is automatically controlled to transport passengers (e.g., vehicle occupants 15) from one location to another. The host vehicle 10 may be constructed as a truck, a sedan, a coupe, a sports utility vehicle (SUV), a recreational vehicle (RV), etc. In one embodiment, the host vehicle 10 may be a so-called secondary, tertiary, tertiary, or tertiary automation system. The tertiary system represents "high automation" and refers to the driving mode-specific performance of the autonomous driving system in various aspects of the dynamic driving task, even if the human driver does not respond appropriately to the intervention request. The tertiary system represents "full automation" and refers to the full-time performance of the autonomous driving system in various aspects of the dynamic driving task under a variety of road and environmental conditions that the human driver can manage. In a level 3 vehicle, the vehicle system performs the entire dynamic driving task (DDT) in the area in which it is designed to do so. The vehicle operator is responsible for DDT takeover only when the host vehicle 10 essentially "asks" the driver to take over when a problem occurs or the vehicle is about to leave the area in which it can operate. In a Level 2 vehicle, the system provides steering, braking / acceleration support, lane centering and adaptive cruise control.

[0019] As shown, the main vehicle 10 generally includes a propulsion system 20, a transmission system 22, a steering system 24, a braking system 26, a sensor system 28, an actuator system 30, at least one data storage device 32, at least one controller 34, and a communication system 36. In various embodiments, the propulsion system 20 may include an electric motor such as a traction motor and / or a fuel cell propulsion system. The main vehicle 10 may also include a battery (or battery pack) 21 electrically connected to the propulsion system 20. Therefore, the battery 21 is configured to store electrical energy and provide electrical energy to the propulsion system 20. In certain embodiments, the propulsion system 20 may include an internal combustion engine. The transmission system 22 is configured to transfer power from the propulsion system 20 to the wheels 17 according to a selectable speed ratio. According to various embodiments, the transmission system 22 may include a stepped automatic transmission, a continuously variable transmission, or other suitable transmission. The braking system 26 is configured to provide braking torque to the wheels 17. In various embodiments, the braking system 26 may include friction brakes, wire control brakes, regenerative braking systems (e.g., motors), and / or other suitable braking systems. The steering system 24 affects the position of the wheels 17 and may include a steering wheel 33. Although depicted as including a steering wheel 33 for purposes of illustration, in some embodiments contemplated within the scope of the present disclosure, the steering system 24 may not include a steering wheel 33 .

[0020] The sensor system 28 includes one or more sensors 40 (i.e., sensing devices) that sense observable conditions of the external environment and / or the internal environment of the host vehicle 10. The sensors 40 communicate with the controller 34 and may include, but are not limited to, one or more radars, one or more light detection and ranging (LIDAR) sensors, one or more proximity sensors, one or more odometers, one or more ground penetrating radar (GPR) sensors, one or more steering angle sensors, a global navigation satellite system (GNSS) transceiver (e.g., one or more global positioning system (GPS) transceivers), one or more tire pressure sensors, one or more cameras, one or more gyroscopes, one or more accelerometers, one or more inclinometers, one or more speed sensors, one or more three-dimensional (3D) depth sensors, one or more ultrasonic sensors, one or more inertial measurement units (IMUs), thermal imaging sensors, one or more microphones 31, and / or other sensors. Each sensor 40 is configured to generate a signal indicating a sensed observable condition (i.e., sensor data) of the external environment and / or the internal environment of the host vehicle 10. Because the sensor system 28 provides sensor data to the controller 34, the sensor system 28 and its sensors 40 are considered to be information sources (or simply sources).

[0021] The actuator system 30 includes one or more actuator devices 42 that control one or more vehicle features, such as, but not limited to, the propulsion system 20, the transmission system 22, the steering system 24, and the braking system 26. In various embodiments, the vehicle functions may also include interior and / or exterior vehicle functions, such as, but not limited to, doors, trunk, and cabin functions such as air, music, lighting, etc.

[0022] The data storage device 32 stores data for automatically controlling the host vehicle 10. In various embodiments, the data storage device 32 stores a map defined by the navigable environment. In various embodiments, the defined map may be predefined by a remote system and obtained from the remote system. For example, the defined map may be assembled by the remote system and transmitted to the host vehicle 10 (wirelessly and / or by wire) and stored in the data storage device 32. The data storage device 32 may be part of the controller 34, rerouted from the controller 34, or part of the controller 34 and part of the rerouting system.

[0023] The controller 34 includes at least one processor 44 and a non-transitory computer readable storage device or medium 46. The processor 44 can be a custom or commercially available processor, a central processing unit (CPU), a graphics processing unit (GPU), an auxiliary processor of several processors associated with the sensing device controller 34, a semiconductor-based microprocessor (in the form of a microchip or chipset), a macroprocessor, a combination thereof, or generally a device for executing instructions. The computer readable storage device or medium 46 can include volatile and non-volatile storage devices such as read-only memory (ROM), random access memory (RAM), and keep alive memory (KAM). KAM is a persistent or non-volatile memory that can be used to store various operating variables when the processor 44 is turned off. The computer readable storage device or medium 46 can be implemented using a variety of storage devices, such as PROM (programmable read-only memory), EPROM (electrical PROM), EEPROM (electrically erasable PROM), flash memory, or any other electrical, magnetic, optical, or combination storage device capable of storing data used by the controller 34 when controlling the host vehicle 10, some of which represents executable instructions. The controller 34 of the host vehicle 10 may be referred to as a vehicle controller and may be programmed to perform the method 100 described in detail below. Figure 3 ) and / or method 200( Figure 4 ).

[0024] The instructions may include one or more rerouting programs, each of which includes an ordered list of executable instructions for implementing logical functions. When executed by the processor 44, the instructions receive and process signals from the sensor system 28, execute logic, calculations, methods and / or algorithms for automatically controlling components of the host vehicle 10, and generate control signals to the actuator system 30 to automatically control components of the host vehicle 10 based on the logic, calculations, methods and / or algorithms. Although Figure 1 , a single controller 34 is shown, but embodiments of the host vehicle 10 may include multiple controllers 34 that communicate via a suitable communication medium or combination of communication mediums and cooperate to process sensor signals, execute logic, calculations, methods and / or algorithms, and generate control signals to automatically control functions of the host vehicle 10. In various embodiments, one or more instructions of the controller 34 are included in the system 98.

[0025] The host vehicle 10 includes a user interface 23, which may be a touch screen 43 ( Figure 3). The user interface 23 may include, but is not limited to, an alarm, such as one or more speakers 27 for providing audible sounds, tactile feedback in a vehicle seat or other object, one or more displays 29, one or more microphones 31 (e.g., a microphone array), and / or other devices suitable for providing notifications to a vehicle user of the host vehicle 10. The user interface 23 is in electronic communication with the controller 34 and is configured to receive input by a user (e.g., a vehicle operator or a vehicle passenger). For example, the user interface 23 may include a touch screen and / or buttons configured to receive input from a person (e.g., a vehicle occupant 15). Thus, the controller 34 is configured to receive input from a user via the user interface 23. Although Figure 1 2 is shown as part of the user interface 23, but other microphones 31 may be part of the sensor system 28. The microphone 31 is configured to capture voice commands from the vehicle occupant 15 in the passenger compartment 39 of the host vehicle 10. However, it is contemplated that the microphone 31 may be coupled to other components of the host vehicle 10. Regardless of its exact location, the microphone 31 communicates with the controller 34.

[0026] The host vehicle 10 may include one or more displays 29 configured to display one or more images as described below. For example, the display 29 may be configured to present information to one or more vehicle occupants 15 within a passenger compartment 39 of the host vehicle 10. The display 29 is in communication with the controller 34. Thus, the controller 34 is configured to control the operation of the display 29.

[0027] The communication system 36 communicates with the controller 34 and is configured to wirelessly transmit information to other entities (e.g., remote entities 48) and wirelessly receive information from other entities using, for example, vehicle-to-everything (V2X) technology. As a non-limiting example, the communication system 36 can send and / or receive information from other vehicles ("V2V" communication), infrastructure ("V2I" communication), remote systems at remote call centers (e.g., ON-STAR of GENERAL MOTORS), and / or personal electronic devices (e.g., mobile phones). Some remote entities 48 may also be configured to wirelessly transmit information to other entities (e.g., remote vehicles 48) and wirelessly receive information from other entities using, for example, V2X technology. In the present disclosure, a remote vehicle 48 capable of wirelessly transmitting information to other entities (e.g., remote vehicles 48) is referred to as a remote vehicle with V2X capabilities. In some embodiments, the communication system 36 is a wireless communication system configured to communicate using the IEEE 802.11 standard or via a wireless local area network (WLAN) using cellular data communications. However, additional or alternative communication methods such as dedicated short range communication (DSRC) channels are also considered to be within the scope of the present disclosure. DSRC channels refer to unidirectional or bidirectional short-range to medium-range wireless communication channels designed specifically for automotive use and a set of corresponding protocols and standards. Therefore, the communication system 36 may include one or more antennas and / or communication transceivers 37 for receiving and / or transmitting signals, such as collaborative sensing messages (CSM). The communication system 36 is configured to wirelessly transmit information between the main vehicle 10 and another vehicle (i.e., the remote entity 48). In addition, the communication system 36 is configured to wirelessly transmit information between the main vehicle 10 and infrastructure.

[0028] refer to Figure 2In several driving scenarios, the host vehicle 10 may be traveling on a road 54 while other actors 56 (e.g., another vehicle, pedestrian, cyclist, etc.) are moving along the same road 54. In such driving scenarios, it is desirable to predict the occupancy corridor 58 of the actor 56 in the short term (e.g., 0 to 5 seconds) to allow the host vehicle 10 to avoid contact with the actor 56. Specifically, the host vehicle 10 may select a path 62 taking into account the centerline 60 of the road 54 and the predicted occupancy corridor 58 of the actor 56. The presently disclosed hybrid data-driven actor prediction system 98 combines reachable set analysis with observed real-world or simulation data. In doing so, the system 98 predicts a set of tightly bound observed trajectories of the actors 56 that lead to similar characteristics under variable driving scenarios and conditions. In other words, actors 56 with similar characteristics will have similar predicted occupancy corridors 58. In system 98, the prediction of occupancy corridors 58 is a function of various complex features, namely: actor aggressiveness, scene uncertainty, road geometry, weather conditions, vehicle class, etc. System 98 saves computing power and energy by utilizing machine learning models to compute open-loop occupancy corridor predictions faster than other systems. System 98 is also less conservative than other methods because occupancy corridor predictions are a function of actor aggressiveness, scene uncertainty, vehicle class / type, map geometry, and other complex features, and are used for short-term ranges. System 98 is more conservative than purely data-driven methods because system 98 outputs sets, envelopes, and / or regions rather than singleton trajectories.

[0029] Figure 3 is a flow chart of a hybrid data-driven actor prediction method 100 using reachable sets and supervised learning. The method 100 begins at box 102 and box 104. At box 102, the controller 34 receives map data. As a non-limiting example, the map data includes a map of the area surrounding the actor 56, the road centerline 60, and the nominal path route of the actor 56. At box 104, the controller 34 receives initial trajectory data of the actor 56. The initial trajectory data includes an initial trajectory of the actor 56. The map data and the initial trajectory data can be collectively referred to as input data. Therefore, the controller 34 initially receives the input data. Then, the method 100 proceeds to box 106.

[0030] At box 106, scene fusion occurs. Specifically, at box 106, the controller 34 receives perception data from sensors 40 (e.g., cameras, lidar, etc.) of the host vehicle 10 and / or the actor 56. The perception data includes information about the area around the actor 56. The perception data is then fused with the map data to generate scene data. The scene data includes the nominal route path of the actor 56, the road centerline 60 of the road 54, the road geometry of the road 54, etc. As described above, the host vehicle 10 and / or the actor 56 may be moving on the road 54. Next, the method 100 proceeds to box 108. At box 108, the controller 34 uses the accessibility analysis and the input data to predict multiple occupancy sets of the actor 56 within a predetermined time period in the future. In the present disclosure, the term "occupancy" refers to the position of the actor 56 on the road. Therefore, the term "occupancy set" refers to multiple possible positions of the actor 56 on the road 54 at a specific point in time.

[0031] Figure 4 is a flow chart of a method 200 for predicting the occupancy set of an actor 56 within a predetermined time period in the future using reachability analysis and input data. The input data includes the state of the actor 56 (i.e., the actor state) and the actor state uncertainty. Given the actor state and the state uncertainty, the controller 34 defines the initial occupancy set of the actor 56 at the initial time as the area occupied by the actor 56 on the road 54 at the initial time (i.e., the first time step). Once the initial occupancy set is determined, the method 200 proceeds to box 202. At box 202, the controller 34 uses reachability theory to determine (i.e., calculate) the next forward reachable set (i.e., the forward occupancy set of the actor 56). The method 200 then proceeds to box 204. At box 204, the controller 34 determines whether the occupancy set (i.e., the forward reachable set) of the actor 56 determined in box 202 intersects with other traffic actors and / or is within a permitted road position (e.g., within the edge line 64 of the road 54) within the predetermined time period ( Figure 2 ). If the occupancy set of the actor 56 does not intersect with other traffic actors and / or is within the allowed road position, the controller 34 designates the occupancy set as the valid occupancy set and then proceeds to box 206. At box 206, the controller 34 designates the new occupancy set as the current occupancy set in order to determine the next occupancy set for the actor 56. The method 100 then returns to box 202. Thus, in the method 200, the controller 34 performs continuous forward occupancy set predictions over a predetermined period of time in the future. These sets are occupancy sets that represent locations that the actor 56 can reach within the allotted time.

[0032] Return to Figure 3At box 110, the occupancy set prediction determined at box 108 is fed to the reasoner. The reasoner also receives the initial trajectory data of the actor 56 determined at box 104 and the scene data generated at box 106. In addition to the scene data, the reasoner may also receive (as input data) the trajectory data of the actor 56, the actor aggressiveness, the actor type (e.g., sedan, SUV, truck, bicycle, pedestrian, etc.), road conditions, weather conditions, nominal predictions and the confidence (i.e., uncertainty) of these nominal predictions, map data, and scene uncertainty (e.g., perception and / or location uncertainty). The reasoner selects hyperparameters that control the occupancy set prediction based on the scene data. As non-limiting examples, the hyperparameters include the prediction range length (e.g., time range) of the occupancy set prediction, the time step, the braking time, the collision time, and / or the time to the nearest actor. The prediction range length (i.e., how far into the future) of the occupancy set prediction can be variable. For example, the prediction range length of the occupancy set prediction can be a function of the estimated time to collision with the nearest actor and the uncertainty of the nominal prediction assumptions. As an output, the reasoner uses the machine learning model and the input data to determine the occupancy corridor constraints for the actor 56. In other words, the reasoner outputs hyperparameters that constrain the occupancy corridor prediction. As non-limiting examples, the occupancy corridor constraints for the actor 56 include the maximum acceleration of the actor 56, the minimum acceleration of the actor 56, the prediction range length and / or time step (e.g., coarse or fine) of the occupancy set prediction. Supervised learning is used to train function approximators, such as neural networks, etc. Therefore, the reasoner can be a machine learning model, such as a neural network, etc.

[0033] Figure 5is a flow chart of a method 300 for training the above-mentioned reasoner. The method 300 is performed offline and uses supervised learning to learn the mapping of input to output using simulated and / or observed driving data. The method 300 starts at box 302. At box 302, the controller 34 receives input data. As described above, the input data may include trajectory data of the actor 56, actor aggressiveness, actor type (e.g., sedan, SUV, truck, bicycle, pedestrian, etc.), road conditions, weather conditions, nominal predictions and confidence (i.e., uncertainty) of these nominal predictions, map data, scene data, and scene uncertainty (e.g., perception and / or position uncertainty). Then, the method 300 proceeds to box 304. At box 304, the reasoner receives input data (i.e., simulated and / or observed driving data) and outputs the acceleration of the actor 56. Next, the method 300 proceeds to box 306. At box 306, the occupancy predictor receives the acceleration of the actor 56 and outputs the occupied corridor 58 of the actor 56. The method 300 then proceeds to boxes 308 and 310. At box 308, the ground truth trajectory of the actor 56 is received. The ground truth trajectory of the actor 56 can be based on simulated and / or observed driving data. The ground truth trajectory of the actor 56 is input into the loss function. The loss function is a mathematical function that measures how well a machine learning model performs on a given data set. The machine learning algorithm (i.e., the reasoner) minimizes the value of the loss function, which is achieved by adjusting the model parameters during the training process. The loss function is used to update the machine learning parameters of the reasoner (e.g., neural network parameters). Therefore, at box 304, the updated machine learning parameters are fed to the reasoner again. The training method 300 is repeated until the value of the loss function reaches a predetermined minimum.

[0034] Return to Figure 3 , the method 300 proceeds to block 112 after block 110. As described above, the reasoner outputs hyperparameters for controlling occupancy set prediction based on scene data. At block 112, the occupancy predictor receives the hyperparameters for controlling occupancy set prediction and determines (e.g., calculates) the occupancy set of the front target actor 56 of interest in time. Next, the method 200 proceeds to block 114.

[0035] At box 114, the controller 34 receives the occupancy set and uncertainty corridor determined at box 108 and the occupancy corridor determined at box 112. Using this information, the predictor resolver (at box 114) blends the nominal (i.e., baseline) machine learning predictions and the occupancy set together. The predictor resolver performs two tasks. First, the predictor resolver ensures that the data-driven occupancy corridor determination (i.e., machine learning determination) has the same data structure as the occupancy set prediction determined at box 108. Second, the predictor resolver also corrects infeasible nominal predictions through projection and convex optimization. For example, a nominal prediction that was outside the occupancy corridor before a predetermined time period is infeasible and / or inconsistent with the training data. The correction mechanism finds the trajectory that is closest to the occupancy set prediction that is within the occupancy region for the initial time step and minimizes the error for the remaining time. To do this, the prediction solver can use an optimization cost function defined by the following equation:

[0036]

[0037] So that the following conditions are met

[0038] ψ(i)∈O i ,i=1,…,N

[0039] X=[x(0),x(1),…,x(N)] T

[0040] Ψ=[ψ(0),ψ(1),…,ψ(N)] T

[0041] in:

[0042] J is the optimization cost function;

[0043] X is the waypoint predicted by the corrected trajectory;

[0044] Ψ is the waypoint predicted by the infeasible trajectory;

[0045] i is the time index;

[0046] N is the number of time steps; and

[0047] O i is the occupancy set at time index i.

[0048] At box 114, the predictor parser outputs the occupancy corridor 58 of the actor 56. The method 100 then proceeds to box 116. At box 116, the occupancy corridor 58 of the actor 56 is fed to the behavior planning algorithm of the host vehicle 10. The controller 34 of the host vehicle 10 controls the movement of the host vehicle 10 based on the occupancy corridor 58 of the actor 56. For example, the controller 34 can determine candidate trajectories for the host vehicle 10 and compare these candidate trajectories with the occupancy corridor 58 of the actor 56. If the candidate trajectory of the host vehicle 10 intersects the occupancy corridor 58 of the actor 56, the controller 34 removes the candidate trajectory. If the candidate trajectory of the host vehicle 10 intersects the long-term data-driven occupancy prediction of the actor 56, the controller 34 allows the candidate trajectory in the short term. Furthermore, if the candidate trajectory of the host vehicle 10 does not intersect the occupancy corridor 58 of the actor 56 or the long-term data-driven occupancy prediction, the candidate trajectory is deemed acceptable and the controller 34 automatically controls the host vehicle 10 to move along the candidate trajectory.

[0049] Although exemplary embodiments are described above, it is not intended that these embodiments describe all possible forms included in the claims. The words used in the specification are descriptive rather than restrictive, and it is understood that various changes can be made without departing from the spirit and scope of the present disclosure. As previously mentioned, the features of the various embodiments can be combined to form further embodiments of the currently disclosed system and method that may not be explicitly described or illustrated. Although various embodiments can be described as having advantages or advantages over other embodiments or prior art embodiments relative to one or more desired features, a person of ordinary skill in the art recognizes that one or more features or characteristics can be compromised to achieve the desired overall system properties, depending on the specific application and implementation. These properties may include, but are not limited to, cost, strength, durability, life cycle cost, marketability, appearance, packaging, size, applicability, weight, manufacturability, ease of assembly, etc. Therefore, embodiments described as less ideal than other embodiments or prior art implementations relative to one or more features are not outside the scope of the present disclosure and may be ideal for specific applications.

[0050] The drawings are in simplified form and are not drawn to exact scale. Directional terms such as top, bottom, left, right, up, above, above, below, under, rear, and front may be used with respect to the drawings for the purpose of convenience and clarity of description only. These directional terms and similar directional terms should not be construed as limiting the scope of the present disclosure in any way.

[0051] Embodiments of the present disclosure are described herein. However, it should be understood that the disclosed embodiments are merely examples, and other embodiments in various alternative forms may be adopted. The drawings are not necessarily drawn to scale; certain features may be enlarged or minimized to show the details of a particular component. Therefore, the specific structural and functional details disclosed herein should not be interpreted as limiting, but only as a representative basis for teaching those skilled in the art to adopt the systems and methods of the present disclosure in different ways. As will be understood by those of ordinary skill in the art, the various features shown and described with reference to any of the accompanying drawings may be combined with the features shown in one or more other drawings to produce embodiments that are not explicitly shown or described. The combination of features shown provides representative embodiments of typical applications. However, various combinations and modifications of features consistent with the teachings of the present disclosure may be required for specific applications or implementations.

[0052] Embodiments of the present disclosure may be described herein in terms of functions and / or logic block components and various processing steps. It should be understood that such block components may be implemented by multiple hardware, software and / or firmware components configured to perform specified functions. For example, embodiments of the present disclosure may employ various integrated circuit components, such as memory elements, digital signal processing elements, logic elements, lookup tables, etc., which may perform multiple functions under the control of one or more microprocessors or other control devices. In addition, those skilled in the art will appreciate that embodiments of the present disclosure may be practiced in conjunction with a variety of systems, and the systems described herein are merely exemplary embodiments of the present disclosure.

[0053] For the sake of brevity, technologies related to signal processing, data fusion, signaling, control and other functional aspects of the system (as well as the various operating components of the system) may not be described in detail herein. In addition, the connecting lines shown in the various figures contained herein are intended to represent example functional relationships and / or physical couplings between the various elements. It should be noted that alternative or additional functional relationships or physical connections may exist in the embodiments of the present disclosure.

[0054] This description is merely illustrative in nature and is in no way intended to limit the present disclosure, its application or use. The broad teachings of the present disclosure can be implemented in many forms. Therefore, although the present disclosure includes specific examples, the true scope of the present disclosure should not be so limited, because other modifications will become apparent after studying the drawings, description and appended claims.

Claims

1. A method for predicting corridor occupancy by an actor, comprising: receiving input data, wherein the input data comprises map data and initial trajectory data, the map data comprising a map of an area around an actor, and the initial trajectory data comprising an initial trajectory of the actor; using reachability theory analysis and the input data to predict a plurality of occupancy sets of the actor over a predetermined future time period; Using the machine learning model and the input data, determining an occupancy corridor constraint for the actor; determining an occupancy corridor of the actor using the occupancy corridor constraints of the actor using the machine learning model and the occupancy set of the actor using reachability analysis; and Movement of a host vehicle is controlled based on the occupied corridor of the actor.

2. The method according to claim 1, further comprising: receiving perception data from a plurality of sensors of the host vehicle, wherein the perception data includes information about an area surrounding the actor; as well as The perception data is fused with the map data to generate scene data, wherein the scene data includes a nominal route path and a road centerline of a road on which the agent is moving.

3. The method according to claim 2, wherein: The initial trajectory data includes a longitudinal acceleration and a lateral acceleration of the actor, the reachability analysis uses the longitudinal acceleration and the lateral acceleration of the actor, and the occupancy set of the actor is a forward reachable set within the predetermined time period.

4. The method according to claim 3, wherein: Predicting the occupancy set of the actor further comprises: Determining that the forward reachable set of the agent does not intersect with other traffic agents within the predetermined time period; and It is determined that the forward reachable set of the actor is within allowed road locations.

5. The method according to claim 4, further comprising: using an inference engine to determine an output acceleration of the actor using the input data; The occupied corridor is determined using the output acceleration of the actor.

6. The method according to claim 5, wherein: The occupied corridor constraint condition of the actor includes: the maximum acceleration of the actor and the minimum acceleration of the actor.

7. The method according to claim 6, further comprising: A determination is made as to whether a trajectory of the host vehicle intersects the occupancy corridor of the actor.

8. The method according to claim 7, further comprising: Trajectories of the host vehicle that intersect the occupied corridor of the actor are removed.

9. The method according to claim 6, further comprising: A determination is made that a trajectory of the host vehicle does not intersect the occupancy corridor of the actor.

10. The method according to claim 9, wherein: Controlling movement of the host vehicle based on the occupied corridor of the actor includes moving along the trajectory of the host vehicle that does not intersect the occupied corridor of the actor.

Images