Methods, devices, terminals, and storage media for determining vehicle trajectory

Abstract

This application provides a method, apparatus, terminal, and storage medium for determining a vehicle trajectory, belonging to the field of vehicle technology. The method includes: acquiring state data of a target vehicle; creating an initial behavior decision tree based on the state data; pruning the initial behavior decision tree based on at least one of a first heuristic rule and a second heuristic rule to obtain a target behavior decision tree, wherein the first heuristic rule indicates that the number of behavior changes of the target vehicle within the future time period does not exceed a preset number, and the second heuristic rule indicates that the target vehicle meets lane-changing conditions; for each behavior sequence in the target behavior decision tree, fitting the time and vehicle position within the future time period based on multiple behaviors in the behavior sequence and the state data to obtain the driving trajectory corresponding to the behavior sequence. The above technical solution can quickly determine the driving trajectory of a target vehicle.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a method, device, terminal and storage medium for determining driving trajectory. Background Technology

[0002] With the continuous development of vehicle technology, intelligent vehicles capable of autonomous driving have emerged. During autonomous driving, how to plan reasonable driving trajectories for intelligent vehicles to ensure their safe operation is a key research focus in this field.

[0003] Currently, there has been some research on trajectory planning methods for intelligent vehicles. For example, patent application number CN202010763228.1, entitled "A Global Path Planning Method for Intelligent Vehicles Based on Improved A* Algorithm," describes a method that divides a specific outdoor area into grids and uses the center of each grid as a control point to determine the driving trajectory of the intelligent vehicle based on the improved A* algorithm.

[0004] However, the above technical solutions require multiple iterations and a lot of computing power to determine the driving trajectory of the intelligent vehicle, which is inefficient and cannot determine the driving trajectory of the intelligent vehicle in real time based on its current state. Summary of the Invention

[0005] This application provides a method, apparatus, terminal, and storage medium for determining vehicle trajectory, which can quickly determine the trajectory of a target vehicle. The technical solution is as follows:

[0006] On the one hand, a method for determining a vehicle trajectory is provided, the method comprising:

[0007] Acquire the status data of the target vehicle, wherein the status data is used to represent the status of the target vehicle at the current moment;

[0008] Based on the state data, an initial behavior decision tree is created. The root node in the initial behavior decision tree represents the behavior performed by the target vehicle at the current moment. Other nodes in the initial behavior decision tree, excluding the root node, represent the behaviors that the target vehicle can perform in a future time period. The behaviors that the target vehicle can perform include three categories: changing lanes to the left, maintaining the original lane, and changing lanes to the right.

[0009] Based on at least one of the first heuristic rule and the second heuristic rule, the initial behavior decision tree is pruned to obtain the target behavior decision tree. The first heuristic rule is used to indicate that the number of behavior changes of the target vehicle within the future time period does not exceed a preset number, and the second heuristic rule is used to indicate that the target vehicle meets the lane-changing conditions.

[0010] For each behavior sequence in the target behavior decision tree, based on multiple behaviors in the behavior sequence and the state data, the time and vehicle position within the future time period are fitted to obtain the driving trajectory corresponding to the behavior sequence.

[0011] On the other hand, a device for determining a vehicle trajectory is provided, the device comprising:

[0012] The acquisition module is used to acquire the status data of the target vehicle, wherein the status data is used to represent the status of the target vehicle at the current moment;

[0013] A creation module is used to create an initial behavior decision tree based on the state data. The root node in the initial behavior decision tree represents the behavior performed by the target vehicle at the current moment. Other nodes in the initial behavior decision tree, excluding the root node, represent the behaviors that the target vehicle can perform in a future time period. The behaviors that the target vehicle can perform include three categories: changing lanes to the left, maintaining the original lane, and changing lanes to the right.

[0014] The first processing module is used to prune the initial behavior decision tree based on at least one of a first heuristic rule and a second heuristic rule to obtain a target behavior decision tree. The first heuristic rule is used to indicate that the number of behavior changes of the target vehicle in the future time period does not exceed a preset number, and the second heuristic rule is used to indicate that the target vehicle meets the lane-changing conditions.

[0015] The second processing module is used to fit the time and vehicle position in the future time period for each behavior sequence in the target behavior decision tree based on multiple behaviors in the behavior sequence and the state data, so as to obtain the driving trajectory corresponding to the behavior sequence.

[0016] In some embodiments, the creation module is configured to:

[0017] Based on the longitudinal position, lateral position, first speed, second speed, yaw angle, and yaw rate of the target vehicle in the state data, the behavior performed by the target vehicle at the current moment is determined, where the first speed is the speed along the vehicle body direction and the second speed is the speed perpendicular to the vehicle body direction.

[0018] Based on the behavior of the target vehicle at the current moment, create the root node of the initial behavior decision tree;

[0019] Based on the behavior set, other nodes in the initial behavior decision tree, excluding the root node, are created on the basis of the root node. The behavior set includes three types of behaviors that the target vehicle can perform.

[0020] In the initial behavior decision tree, each node other than its child nodes has three child nodes. The three child nodes of each node are used to represent changing lanes to the left, keeping the original lane, and changing lanes to the right, respectively.

[0021] In some embodiments, the first processing module is configured to:

[0022] Based on the first heuristic rule, multiple first nodes are determined from the initial behavior decision tree. The number of target node pairs between each first node and the root node is equal to the preset number of times. There is a parent-child relationship between the two nodes in the target node pair, and the behaviors represented by the two nodes are different.

[0023] For any one of the plurality of first nodes, remove the second node corresponding to the first node and all nodes after the second node from the initial behavior decision tree. The second node is a child node of the first node, and the behavior represented by the second node is different from the behavior represented by the first node.

[0024] In some embodiments, the first processing module is configured to:

[0025] Based on the second heuristic rule, the first warning distance, first lane change gap, second warning distance, and second lane change gap of the target vehicle at the current moment are obtained. The first warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when it changes lanes to the left. The first lane change gap is the distance between the target vehicle and surrounding vehicles when it changes lanes to the left. The second warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when it changes lanes to the right. The second lane change gap is the distance between the target vehicle and surrounding vehicles when it changes lanes to the right.

[0026] Based on the first warning distance, the first lane change gap, the second warning distance, and the second lane change gap, the node corresponding to the target behavior sequence is removed from the initial behavior decision tree to obtain the target behavior decision tree. The target behavior sequence includes the target lane change behavior, and the warning distance corresponding to the target lane change behavior is less than the corresponding lane change gap.

[0027] In some embodiments, the first processing module is configured to:

[0028] Based on the first heuristic rule, the initial behavior decision tree is pruned for the first time to obtain the first intermediate behavior decision tree;

[0029] Based on the second heuristic rule, the intermediate behavior decision tree is pruned a second time to obtain the target behavior decision tree.

[0030] In some embodiments, the apparatus further includes:

[0031] The third processing module is used to determine multiple third nodes from the initial behavior decision tree, wherein the behavior represented by each third node is the opposite of the behavior represented by the parent node of the third node; and to remove the multiple third nodes and all nodes after the third nodes from the initial behavior decision tree to obtain a second intermediate behavior decision tree.

[0032] The first processing module is used to prune the second intermediate behavior decision tree based on at least one of the first heuristic rule and the second heuristic rule to obtain the target behavior decision tree.

[0033] In some embodiments, the second processing module includes:

[0034] The simulation unit is used to perform dynamic simulation on the target vehicle for each behavior sequence in the target behavior decision tree, based on multiple behaviors in the behavior sequence and the state data, to determine the destination position of the target vehicle after the future time period.

[0035] The processing unit is configured to perform polynomial fitting on the target vehicle's lateral starting position and lateral ending position at the current moment with the time in the future time period to obtain the lateral trajectory of the target vehicle, which represents the change of the target vehicle's lateral position in the future time period.

[0036] The determining unit is used to determine the acceleration at multiple times within the future time period using a longitudinal car-following model, and to determine the longitudinal trajectory of the target vehicle based on the acceleration at the multiple times and the longitudinal starting position of the target vehicle at the current time. The longitudinal trajectory is used to represent the change in the longitudinal position of the target vehicle within the future time period.

[0037] In some embodiments, the simulation unit is configured to:

[0038] For each behavior sequence in the target behavior decision tree, based on the behavior to be executed at multiple times in the behavior sequence and the state data, the target vehicle is subjected to dynamic simulation to obtain the state data of the target vehicle at multiple times. The state data corresponding to each time time includes the longitudinal position of the target vehicle, the road adhesion coefficient and the vehicle speed at that time.

[0039] Cluster the state data at the multiple time points to obtain multiple sets of state data;

[0040] For any one set of state data in the multiple sets of state data, the average of the multiple state data in the set of state data is calculated to obtain the center state data corresponding to the set of state data.

[0041] Based on the central state data, a dynamic simulation is performed on the target vehicle to obtain the destination position of the target vehicle after the future time period.

[0042] On the other hand, a computer device is provided, the computer device including a processor and a memory, the memory being used to store at least one computer program, the at least one computer program being loaded and executed by the processor to implement the method for determining the driving trajectory in the embodiments of this application.

[0043] On the other hand, a computer-readable storage medium is provided, wherein at least one computer program is stored in the computer-readable storage medium, the at least one computer program being loaded and executed by a processor to implement the method for determining the driving trajectory as described in the embodiments of this application.

[0044] On the other hand, a computer program product is provided, including a computer program stored in a computer-readable storage medium, a processor of a computer device reading the computer program from the computer-readable storage medium, and the processor executing the computer program, causing the computer device to perform the method for determining the driving trajectory provided in the above-described aspects or various alternative implementations of the above-described aspects.

[0045] This application provides a method for determining a driving trajectory. It can generate an initial behavior decision tree in real time based on the current behavior of the target vehicle. Furthermore, it can design first heuristic rules for behavior change restrictions and primary vehicles based on human driving habits to perform initial pruning on the initial behavior decision tree. Secondly, it can also establish a human-like lane-change warning model considering vehicle and driver heterogeneity to perform secondary pruning of the behavior decision tree from a safety perspective. Finally, based on the state of the vehicle behavior within each decision cycle in the target behavior decision tree, it generates a driving trajectory corresponding to each behavior sequence in the target behavior decision tree, providing it to the target vehicle for selection. This reduces the complexity of the decision-making algorithm, improves the efficiency of driving trajectory determination, and also helps ensure the robustness of the subsequently determined driving trajectory, making it more practical. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram illustrating the implementation environment of a method for determining a driving trajectory according to an embodiment of this application;

[0048] Figure 2 This is a flowchart of a method for determining a driving trajectory according to an embodiment of this application;

[0049] Figure 3 This is a schematic diagram of an initial behavior decision tree provided according to an embodiment of this application;

[0050] Figure 4 This is a schematic diagram of an initial behavior decision tree after pruning, according to an embodiment of this application.

[0051] Figure 5 This is a block diagram of a vehicle trajectory determination device according to an embodiment of this application;

[0052] Figure 6 This is a schematic diagram of the structure of a vehicle controller for a target vehicle according to an embodiment of this application;

[0053] Figure 7 This is a schematic diagram of the structure of a server according to an embodiment of this application. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0055] In this application, the terms "first," "second," etc., are used to distinguish identical or similar items with essentially the same function. It should be understood that there is no logical or temporal dependency between "first," "second," and "nth," nor are there any restrictions on quantity or execution order.

[0056] In this application, the term "at least one" means one or more, and "multiple" means two or more.

[0057] It should be noted that all information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in this application have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the status data of the target vehicle and the status data of surrounding vehicles involved in this application were obtained with full authorization.

[0058] The method for determining driving trajectory provided in this application can be executed by a computer device. In some embodiments, the computer device is a terminal or a server. The following section first uses a computer device as a server as an example to describe the implementation environment of the method for determining driving trajectory provided in this application. Figure 1 This is a schematic diagram illustrating the implementation environment of a method for determining a vehicle trajectory according to an embodiment of this application. See also... Figure 1 The implementation environment includes a smart car 101 and a server 102. The smart car 101 and the server 102 can be connected directly or indirectly via wired or wireless communication, which is not limited herein.

[0059] In some embodiments, the intelligent vehicle 101 may be a new energy vehicle, such as a pure electric vehicle, a plug-in hybrid electric vehicle, or a fuel cell electric vehicle; this application embodiment does not impose any limitations on this. The intelligent vehicle 101 is equipped with a vehicle controller, an on-board charger, a power battery, and a battery management system for managing the power battery. The vehicle controller can communicate with the server 102 via a CAN (Controller Area Network) bus. Accordingly, the vehicle controller of the intelligent vehicle 101 can obtain its own driving trajectory from the server 102 via the network and control the intelligent vehicle 101's driving based on this trajectory.

[0060] In some embodiments, server 102 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), big data, and artificial intelligence platforms. Server 102 is used to provide backend services to support intelligent vehicles. In some embodiments, server 102 undertakes the main computing work, and terminal 101 undertakes the secondary computing work; or, server 102 undertakes the secondary computing work, and terminal 101 undertakes the main computing work; or, server 102 and terminal 101 collaborate on computing using a distributed computing architecture.

[0061] Figure 2 This is a flowchart of a method for determining a driving trajectory according to an embodiment of this application. See also... Figure 2 In this embodiment, the method for determining the vehicle trajectory is illustrated using the server as an example. The method includes the following steps:

[0062] 201. The server obtains the status data of the target vehicle. The status data is used to represent the status of the target vehicle at the current moment.

[0063] In this embodiment, the target vehicle is an intelligent vehicle that supports autonomous driving technology. During the autonomous driving process, the target vehicle can collect its own state data through millimeter-wave radar, lidar, cameras, IMU (Inertial Measurement Unit), navigation system, and vehicle speed sensor, and upload the state data to the server.

[0064] The target vehicle's state data may include its current position, speed, acceleration, yaw angle, and yaw rate, etc., and this embodiment does not impose limitations on these aspects. The target vehicle's position includes longitudinal and lateral positions. Longitudinal position refers to the target vehicle's location along the road direction. Lateral position refers to the target vehicle's location perpendicular to the road direction. The target vehicle's speed includes longitudinal and lateral speeds. Longitudinal speed refers to the speed along the vehicle's body direction. Lateral speed refers to the speed perpendicular to the vehicle's body direction. The target vehicle's acceleration includes longitudinal and lateral acceleration. Longitudinal acceleration refers to the acceleration along the vehicle's body direction. Lateral acceleration refers to the acceleration perpendicular to the vehicle's body direction. The target vehicle's yaw angle refers to the angle of change in the target vehicle's body posture during driving. The target vehicle's yaw rate refers to the rate of change of the target vehicle's body posture, i.e., the rate of change of the yaw angle.

[0065] 202. Based on the state data, the server creates an initial behavior decision tree. The root node in the initial behavior decision tree represents the behavior performed by the target vehicle at the current moment. Other nodes in the initial behavior decision tree, excluding the root node, represent the behaviors that the target vehicle can perform in the future time period. The behaviors that the target vehicle can perform include three categories: changing lanes to the left, keeping the original lane, and changing lanes to the right.

[0066] In this embodiment, the server determines the target vehicle's current action based on the target vehicle's longitudinal position, lateral position, first speed, second speed, yaw angle, and yaw rate from the state data. The first speed is the speed along the vehicle's direction. The second speed is the speed perpendicular to the vehicle's direction. The target vehicle's current action can be any one of changing lanes to the left, maintaining the original lane, or changing lanes to the right; this embodiment does not impose any restrictions on this.

[0067] Then, based on the target vehicle's current actions, the server creates the root node of the initial behavior decision tree. Based on the behavior set, the server creates all other nodes in the initial behavior decision tree, excluding the root node, from the root node. The behavior set includes three types of actions the target vehicle can perform: changing lanes to the left, maintaining the current lane, and changing lanes to the right. Correspondingly, each node in the initial behavior decision tree, excluding leaf nodes, has three child nodes, each representing a lane change to the left, maintaining the current lane, or changing lanes to the right, respectively.

[0068] The initial behavior decision tree represents the target vehicle's behavior within a future time period. The depth of the initial behavior decision tree represents the number of times a behavior is determined within that future time period. That is, the future time period is a decision cycle, and each decision cycle contains multiple decision moments for determining a behavior. The server can discretize the decision cycle into multiple decision moments based on the number of decisions. For any given decision moment, the server determines the target vehicle's behavior once. In other words, for any given decision moment, the server can create a new layer of nodes based on the set of behaviors, thereby constructing the behavior decision tree. Correspondingly, each layer in the initial behavior decision tree corresponds one-to-one with a decision moment, and the nodes in each layer represent the possible behaviors of the target vehicle at the corresponding decision moment.

[0069] The solution provided in this application determines the current behavior of the target vehicle based on the current state data of the target vehicle. Based on the current behavior, an initial behavior decision tree is constructed based on the behavior set. Since the behavior set contains all types of behaviors that the target vehicle can perform, the constructed initial behavior decision tree contains all behavior changes of the target vehicle in the future time period, ensuring that the subsequent determined driving trajectory is determined based on all behavior changes, that is, ensuring the diversity and accuracy of the driving trajectory.

[0070] For example, the current status data of the target vehicle is Where, x k Let y be the longitudinal position of the target vehicle. k v represents the lateral position of the target vehicle. uk Let v be the velocity of the target vehicle along its body direction. vk The speed of the target vehicle perpendicular to the vehicle body direction. Let r be the yaw angle of the target vehicle. kLet be the yaw rate of the target vehicle. The set of behaviors during the target vehicle's behavior decision-making process is A = {LLC, LK, RLC}, where LLC represents changing lanes to the left, LK represents staying in the original lane, and RLC represents changing lanes to the right. The server constructs an initial behavior decision tree for the target vehicle based on its current state data and the set of behaviors. Assuming the target vehicle's current behavior is to stay in the original lane, then staying in the original lane will be used as the initial behavior for creating the initial behavior decision tree. During the construction of the initial behavior decision tree, the server can use discrete time intervals T... S The decision cycle T P The decision is discretized into multiple decision times, from t0 to t4. For each decision time, the server creates a new layer of nodes based on the latest layer of nodes already created. See also... Figure 3 , Figure 3 This is a schematic diagram of an initial behavior decision tree provided according to an embodiment of this application. Figure 3 In the diagram, L, K, and R represent LLC (left lane change), LK (stay in lane), and RLC (right lane change), respectively. Figure 3 As can be seen, each layer in the initial behavior decision tree corresponds one-to-one with the decision time, and the nodes in each layer are used to represent the possible behaviors of the target vehicle at the corresponding decision time.

[0071] 203. The server prunes the initial behavior decision tree based on at least one of the first heuristic rule and the second heuristic rule to obtain the target behavior decision tree. The first heuristic rule is used to indicate that the number of behavior changes of the target vehicle in the future time period does not exceed a preset number, and the second heuristic rule is used to indicate that the target vehicle meets the lane-changing conditions.

[0072] In this embodiment, considering human driving habits and vehicle stability, human drivers do not frequently change vehicle behavior in a short period. Therefore, a first heuristic rule can be set to constrain the number of behavior changes of the target vehicle within a future time period to not exceed a preset number. Considering that lane-changing requires certain conditions to be met, a second heuristic rule can be set to constrain the lane-changing behaviors that can be achieved within a future time period. Then, the server can prune the initial behavior decision tree based on at least one of the first and second heuristic rules to obtain the target behavior decision tree.

[0073] In other words, the server can prune the initial behavior decision tree based on a first heuristic rule to remove behaviors that have changed more than a preset number of times. This embodiment does not limit the preset number of times; for example, the preset number can be positively correlated with a future time period (i.e., the decision cycle). Alternatively, the server can prune the initial behavior decision tree based on a second heuristic rule to remove lane-changing behaviors that cannot be implemented in the initial behavior decision tree. Or, the server can also prune the initial behavior decision tree based on both the first and second heuristic rules to remove behaviors that have changed more than a preset number of times and lane-changing behaviors that cannot be implemented. The following sections describe these various pruning methods.

[0074] The first pruning method involves the server pruning the initial behavior decision tree based on a first heuristic rule. The pruning process includes: the server determining multiple first nodes from the initial behavior decision tree based on the first heuristic rule. The number of target node pairs between each first node and the root node is equal to a preset number. A parent-child relationship exists between the two nodes in a target node pair, and the behaviors represented by the two nodes are different. That is, each target node pair represents one change in the target vehicle's behavior. The number of target node pairs in each link (behavior sequence) of the initial behavior decision tree represents the number of behavior changes when the target vehicle travels along that link. In other words, the server traverses the initial behavior decision tree from the root node, searching for the first node where the number of behavior changes reaches the preset number.

[0075] Then, for any one of the multiple first nodes, the server removes the corresponding second node and all subsequent nodes from the initial behavior decision tree. The second node is a child node of the first node. The behavior represented by the second node differs from the behavior represented by the first node. In other words, the server removes all nodes in the initial behavior decision tree that are after the first node and represent behaviors different from those of the first node. In other words, the server removes the node whose behavior has changed more than a preset number of times and all subsequent nodes from the initial behavior decision tree.

[0076] The embodiments of this application provide a solution that prunes the initial behavior decision tree based on a first heuristic rule, so that the number of behavior changes of the target vehicle in the future time period does not exceed a preset number. This not only directly reduces the number of driving trajectories to be determined later, effectively reducing the complexity of determining driving trajectories, thereby obtaining driving trajectories faster and improving the efficiency of determining driving trajectories, but also helps to ensure the stability of the target vehicle's driving in the future time period, that is, helps to ensure the robustness of the subsequently determined driving trajectories, and has stronger practicality.

[0077] The second pruning method involves the server pruning the initial behavior decision tree based on a second heuristic rule. The pruning process includes: the server obtaining the target vehicle's first warning distance, first lane-changing gap, second warning distance, and second lane-changing gap at the current moment, based on the second heuristic rule. The first warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when changing lanes to the left. The first lane-changing gap is the distance between the target vehicle and surrounding vehicles when changing lanes to the left. The second warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when changing lanes to the right. The server then removes the nodes corresponding to the target behavior sequence from the initial behavior decision tree based on these values, obtaining the target behavior decision tree. The target behavior sequence includes the target lane-changing behavior. The warning distance corresponding to the target lane-changing behavior is less than the corresponding lane-changing gap. If the warning distance is less than the corresponding lane-changing gap, the target vehicle cannot change lanes; that is, the target lane-changing behavior is one that the target vehicle cannot perform.

[0078] In other words, when a target vehicle changes lanes to the left, the system checks if there is a safe lane change gap in the lane immediately to the left of its current lane. If the distance between the vehicle in front and behind in the lane immediately to the left of the target vehicle is greater than the warning distance, the target vehicle can safely change lanes to the left. When a target vehicle changes lanes to the right, the system checks if there is a safe lane change gap in the lane immediately to the right of its current lane. If the distance between the vehicle in front and behind in the lane immediately to the right of the target vehicle is greater than the warning distance, the target vehicle can safely change lanes to the right.

[0079] The first and second warning distances mentioned above can both be determined using the following formula.

[0080] Formula 1:

[0081]

[0082] Among them, S wd This indicates the overall warning distance corresponding to any lane-changing behavior, which can be either the first warning distance or the second warning distance; This indicates the warning distance between the target vehicle and the vehicle in front. The vehicle in front refers to the vehicle in front of the target vehicle in the lane that the target vehicle is about to change to. For example, when changing lanes to the left, the vehicle in front is the vehicle in the lane adjacent to the left of the target vehicle's current lane that is in front of the target vehicle. This indicates the warning distance between the target vehicle and the vehicle behind it. The vehicle in front refers to the vehicle behind the target vehicle in the lane that the target vehicle is about to change to. For example, when changing lanes to the left, the vehicle behind is the vehicle in the lane adjacent to the left of the target vehicle's current lane.

[0083] The above It can be determined by the following formula 2.

[0084] Formula 2:

[0085]

[0086] Among them, v k τk represents the velocity of the target vehicle at time k; τ1 represents the reaction time of the target vehicle's delayed deceleration; t′1 represents the braking time of the target vehicle with a larger deceleration; a max,brake t″1 represents the maximum deceleration of the target vehicle; t″1 represents the duration of the target vehicle's braking at a comfortable deceleration; a comfort,brake Indicates the target vehicle's comfortable braking deceleration; This represents the speed of the vehicle ahead of the target vehicle at time k.

[0087] The above It can be determined by the following formula three.

[0088] Formula 3:

[0089]

[0090] in, τk represents the speed of the vehicle following the target vehicle at time k; τ2 represents the reaction time of the delayed deceleration of the vehicle following the target vehicle; t′2 represents the braking time of the vehicle following the target vehicle at its maximum deceleration. t″2 represents the maximum deceleration of the following vehicle; t″2 represents the duration of the following vehicle's braking at a comfortable deceleration. Indicates the comfort braking deceleration of the following vehicle; v k This represents the speed of the target vehicle at time k.

[0091] The values ​​τ1 and τ2 mentioned above reflect the perception time of different vehicles for impending danger. Larger values ​​for τ1 and τ2 indicate a slower response to dangerous situations; smaller values ​​indicate a more sensitive response. comfort,brake and This demonstrates the vehicle's adaptability to different drivers. Specifically, different drivers apply varying degrees of braking intensity, resulting in different comfortable braking decelerations. In other words, the aforementioned warning distance calculation method considers the impact of different vehicles and drivers, essentially functioning as a human-like lane-change warning model.

[0092] Then, the aforementioned human-like early warning model filters all behavior sequences in the initial behavior decision tree that involve left and right lane changes. It checks whether the target vehicle has a safe lane-changing gap in the target lane when changing lanes, thus filtering the behavior sequences and eliminating those without a safe lane-changing gap to ensure lane-changing safety. For example, if the behavior sequence is a left lane change, the human-like early warning model checks whether there is a safe lane-changing gap in the adjacent left lane; if not, the left lane-changing behavior sequence is deleted. If the behavior sequence is a right lane change, it checks whether there is a safe lane-changing gap in the adjacent right lane; if not, the right lane-changing behavior sequence is deleted. In other words, the server removes behavior sequences that do not meet safety requirements. Specifically, the warning distance and lane-changing gap corresponding to the lane-changing behavior must meet the conditions expressed in Formula 4 below to be retained in the behavior decision tree.

[0093] Formula 4:

[0094]

[0095] in, This indicates the longitudinal position of the vehicle preceding it in the adjacent lane of the target vehicle at time k; f This represents the length of the vehicle preceding it in the adjacent lane to the target vehicle's current lane; x k This represents the longitudinal position of the target vehicle at the current time k; l represents the length of the target vehicle. Indicates the warning distance between the target vehicle and the vehicle in front; This indicates the longitudinal position of the vehicle following it in the adjacent lane of the target vehicle at time k; r This indicates the length of the vehicle following in the adjacent lane to the target vehicle's current lane; Indicates the warning distance between the target vehicle and the vehicle behind it; S wd Indicates the overall warning distance.

[0096] As can be seen from Formula 4 above, when the lane-changing clearance between the target vehicle and the aforementioned preceding vehicle is greater than or equal to... The lane-changing clearance between the target vehicle and the aforementioned following vehicle is greater than or equal to The distance between the aforementioned preceding vehicle and the aforementioned following vehicle (i.e., the overall lane-changing clearance) is greater than or equal to S. wd Only if the conditions in Formula 4 are met can it be determined that the target vehicle has a safe lane-changing gap. Then, the server retains the behavior sequence with the corresponding lane-changing behavior. Otherwise, if the conditions in Formula 4 are not met, the target vehicle does not have a safe lane-changing gap, and the server deletes the behavior sequence with the corresponding lane-changing behavior.

[0097] The solution provided in this application prunes the initial behavior decision tree based on a second heuristic rule, taking into account vehicle heterogeneity and driver heterogeneity. It proposes human-like parameters related to vehicle delay response deceleration time and vehicle comfort, improving the adaptability of the target vehicle to different vehicles and passengers. This makes the generation of subsequent driving trajectories more human-like, and can also select safe driving behavior sequences for driving trajectory calculation, ensuring the safety of the target vehicle's driving. Furthermore, pruning can directly reduce the number of driving trajectories to be determined, effectively reducing the complexity of determining driving trajectories, thereby obtaining driving trajectories faster and improving the efficiency of driving trajectory determination.

[0098] The third pruning method involves the server pruning the initial behavior decision tree based on both the first and second heuristic rules. Accordingly, the pruning process includes: the server performing a first pruning based on the first heuristic rule to obtain a first intermediate behavior decision tree; then, the server performing a second pruning based on the second heuristic rule to obtain the target behavior decision tree. In other words, the server can use the first pruning method to initially prune the initial behavior decision tree to obtain a first intermediate behavior decision tree, and then use the second pruning method to further prune the initially pruned first intermediate behavior decision tree to obtain the target behavior decision tree. This will not be elaborated further here.

[0099] In some embodiments, considering that after a target vehicle changes lanes in a certain direction, it usually travels along the lane of the vehicle that changed lanes or continues to change lanes in the same direction, it is unlikely to change lanes in the opposite direction again in a short period of time, i.e., it is unlikely to perform the opposite behavior in a short period of time. Accordingly, in addition to adopting the above-described pruning method, the server can also remove nodes that perform the opposite behavior in a short period of time from the initial behavior decision tree. Accordingly, the server determines multiple third nodes from the initial behavior decision tree. The behavior represented by each third node is the opposite of the behavior represented by the parent node of the third node. Then, the server removes multiple third nodes and all nodes after the third nodes from the initial behavior decision tree to obtain a second intermediate behavior decision tree. Then, the server can further prune the second intermediate behavior decision tree based on at least one of the first heuristic rule and the second heuristic rule to obtain the target behavior decision tree. Alternatively, the server can first use at least one of the first heuristic rule and the second heuristic rule to perform preliminary pruning on the initial behavior decision tree, and then remove nodes that perform the opposite behavior in a short period of time from the intermediate behavior decision tree obtained from the preliminary pruning. The pruning order is not limited in the embodiments of this application.

[0100] The solution provided in this application removes lane-changing behaviors in the behavior decision tree that are opposite to the previous lane-changing behavior. This makes the behavior changes of the target vehicle in the future time period conform to human habits. This not only directly reduces the number of driving trajectories to be determined, effectively reducing the complexity of determining driving trajectories and thus obtaining driving trajectories faster and improving the efficiency of determining driving trajectories, but also helps to ensure the stability of the target vehicle's driving in the future time period, that is, helps to ensure the robustness of the subsequently determined driving trajectory, and has stronger practicality.

[0101] For example, the first heuristic rule has a preset number of occurrences of 1, meaning the target vehicle can only change its behavior once within a future time period. The server then adjusts its behavior based on this preset number of occurrences. Figure 3 The initial behavior decision tree shown is pruned to obtain... Figure 4 The target behavior decision tree is shown. Figure 3 and Figure 4 As can be seen, the server removes all unreasonable behavior sequences from the initial behavior decision tree (including decision sequences that simultaneously involve left and right lane changes or two or more vehicle behavior changes). See also Figure 4 In (a), if the target vehicle's initial behavior is to remain in lane K, then it can change its behavior once within the entire future time period (decision cycle) to either change lanes to the left (L) or to the right (R). See also Figure 4 In (b), if the target vehicle's initial behavior is to change lanes to the right (R), then within the entire future time period (decision cycle), it is only allowed to change its behavior once to maintain its original lane (K), and it is not allowed to change lanes to the right and then to the left within a short period of time. That is, while ensuring that the number of behavior changes does not exceed the preset number, it is also possible to remove lane-changing behaviors that are in the opposite direction to the previous lane-changing behavior.

[0102] 204. For each behavior sequence in the target behavior decision tree, the server fits the time and vehicle position in the future time period based on multiple behavior and state data in the behavior sequence to obtain the driving trajectory corresponding to the behavior sequence.

[0103] In this embodiment, the behavior sequence in the target behavior decision tree includes the behaviors represented by each node on the link from the root node to any leaf node, representing multiple behaviors that the target vehicle will sequentially execute within a future time period. The number of behavior sequences in the target behavior decision tree is equal to the number of leaf nodes. For any behavior sequence in the target behavior decision tree, the server can fit the time and vehicle position within a future time period based on the multiple behaviors in the behavior sequence and the current state data of the target vehicle, to determine the position of the target vehicle at each time within the future time period, that is, to obtain the driving trajectory corresponding to driving according to the behaviors in the behavior sequence.

[0104] In determining the driving trajectory corresponding to each behavior sequence, for each behavior sequence in the target behavior decision tree, the server performs dynamic simulation of the target vehicle based on multiple behaviors and state data in the behavior sequence to determine the target vehicle's endpoint position after a future time period. Then, based on the target vehicle's lateral starting position and lateral ending position at the current moment, the server performs polynomial fitting with the time points within the future time period to obtain the target vehicle's lateral trajectory. The lateral trajectory represents the change in the target vehicle's lateral position over the future time period. The server uses a longitudinal car-following model to determine the acceleration at multiple time points within the future time period. Based on the acceleration at multiple time points and the target vehicle's longitudinal starting position at the current moment, the server determines the target vehicle's longitudinal trajectory, which represents the change in the target vehicle's longitudinal position over the future time period.

[0105] In other words, the server can establish an adaptive lateral k-th degree polynomial mathematical model based on the target vehicle's current state data, and then apply this model to the target vehicle's decisions in each decision cycle T. P The server calculates the target vehicle's lateral trajectory within a decision period by performing a polynomial fit on the starting position and its constraints, corresponding to the local trajectory of the target vehicle's behavior within the future timeframe. Furthermore, the server establishes a longitudinal car-following model to obtain a discrete acceleration sequence and uses kinematic formulas to derive the target vehicle's longitudinal trajectory within a decision period. Then, the server uses the target vehicle's state after one decision period as the current state and repeats the above operations four times to generate the target vehicle's lateral and longitudinal trajectories within five decision periods of the pruned behavior decision tree.

[0106] Specifically, in determining the destination position of the target vehicle after a future time period, the server performs dynamic simulation on the target vehicle for each behavior sequence in the target behavior decision tree, based on the behaviors and state data to be executed at multiple time points in the behavior sequence, to obtain the target vehicle's state data at multiple time points. The state data corresponding to each time point includes the target vehicle's longitudinal position, road surface adhesion coefficient, and vehicle speed. Then, the server clusters the state data from multiple time points to obtain multiple sets of state data. For any set of state data, the server averages the multiple state data within the set to obtain the central state data corresponding to the set. Then, based on the central state data, the server performs dynamic simulation on the target vehicle to obtain the target vehicle's destination position after a future time period.

[0107] For example, the current status data of the target vehicle is Where, x k y represents the longitudinal position of the target vehicle. k The lateral position of the target vehicle; v represents the velocity of the target vehicle along its body direction, i.e., its longitudinal velocity. vk The velocity of the target vehicle perpendicular to the vehicle body direction, i.e., the lateral velocity; This refers to the acceleration of the target vehicle along the vehicle body direction, i.e., the longitudinal acceleration. This refers to the acceleration of the target vehicle perpendicular to its body direction, i.e., the lateral acceleration. The target vehicle experiences acceleration within a decision period T. P The subsequent state is Among them, the target vehicle goes through a decision cycle T P The subsequent horizontal position Therefore, the lateral position of the target vehicle within this decision period, and the target vehicle in one decision period T P Lateral velocity after time and lateral acceleration All are 0.

[0108] The server can calculate the longitudinal position x of the target vehicle at time k by performing dynamic simulation on the target vehicle. k Road surface adhesion coefficient β k and the speed v of the target vehicle k Then, the server is able to create a dataset D containing all the data points, where each data point is composed of m... k =(x k ,β k ,v j The dataset D is represented by a three-dimensional vector. The server can then normalize the dataset D to ensure that all features in dataset D have similar ranges.

[0109] Then, the server can cluster the data points in dataset D into j clusters. Each data point m k Belongs to a cluster C i Where i = 1, 2, ..., j, for each cluster C i The server calculates the average of its eigenvectors as the center point of the cluster. The center point can be represented as: in These represent the average values ​​of lateral position, road surface adhesion coefficient, and vehicle speed, respectively. The clustering algorithm can be k-means clustering, and this application does not limit this approach.

[0110] For each cluster C i The server can be based on the cluster's center point A dynamic simulation is performed on the target vehicle to obtain its final position after a future time period. Then, the server can determine the target vehicle's final position based on its current lateral starting position and decision period T. PThe lateral endpoint position of the target vehicle is then used as a constraint for the adaptive lateral k-th degree polynomial (Formula 5) to determine the fitting parameter a. η This allows us to obtain the lateral trajectory of the target vehicle within a decision-making cycle.

[0111] Formula 5:

[0112]

[0113] Where y(k) represents the lateral position of the target vehicle at time k, and a η k represents the fitting parameters. i This represents the i-th discrete time point (i.e., the i-th decision moment) within a decision cycle.

[0114] The solution provided in this application uses an adaptive lateral k-th degree polynomial mathematical model to fit the lateral candidate trajectory of the vehicle, thereby improving the adaptability of the generated lateral candidate trajectory to different road surfaces.

[0115] For the specific method of calculating the acceleration of the target vehicle using the above longitudinal following model, please refer to Formula 6 below.

[0116] Formula Six:

[0117]

[0118] Among them, a k denoted as , where is the optimal acceleration of the target vehicle at time k; 'a' represents the maximum deceleration of the target vehicle; 'b' represents the comfortable deceleration of the target vehicle; 'v' represents the optimal acceleration of the target vehicle at time k; k This represents the speed of the target vehicle at time k; Δs represents the speed of the vehicle in front of the target vehicle in its current lane at time k; k This represents the relative distance between the target vehicle and the vehicle in front at time k; v l Indicates the road speed limit; δ represents the acceleration exponent factor; s0 represents the minimum distance between vehicles when stationary; T h This indicates the expected headway.

[0119] Then, based on the acceleration at multiple moments calculated in the above manner and the longitudinal starting position of the target vehicle at the current moment, the server determines the longitudinal trajectory of the target vehicle, as shown in Formula 7 below.

[0120] Formula 7:

[0121]

[0122] Where x(k) represents the longitudinal position of the target vehicle at time k; x k Indicates the decision period T P The longitudinal starting position; T represents the longitudinal velocity of the target vehicle at time k; S Indicates the decision period T P Discrete time intervals within; a k This represents the optimal acceleration of the target vehicle at time k.

[0123] In this embodiment, there are multiple behavior sequences in the target behavior decision tree. For each behavior sequence, the server can calculate the corresponding driving trajectory (including lateral and longitudinal trajectories) using the method described above. Accordingly, for the target behavior decision tree, the server can determine multiple driving trajectories for the target vehicle as candidate trajectories, so that the target vehicle can subsequently determine a driving trajectory from the multiple driving trajectories to drive on. This embodiment does not limit the method of trajectory selection.

[0124] This application provides a method for determining a driving trajectory. It can generate an initial behavior decision tree in real time based on the current behavior of the target vehicle. Furthermore, it can design first heuristic rules for behavior change restrictions and primary vehicles based on human driving habits to perform initial pruning on the initial behavior decision tree. Secondly, it can also establish a human-like lane-change warning model considering vehicle and driver heterogeneity to perform secondary pruning of the behavior decision tree from a safety perspective. Finally, based on the state of the vehicle behavior within each decision cycle in the target behavior decision tree, it generates a driving trajectory corresponding to each behavior sequence in the target behavior decision tree, providing it to the target vehicle for selection. This reduces the complexity of the decision-making algorithm, improves the efficiency of driving trajectory determination, and also helps ensure the robustness of the subsequently determined driving trajectory, making it more practical.

[0125] Figure 5 This is a block diagram of a vehicle trajectory determination device according to an embodiment of this application. This vehicle trajectory determination device is used to perform the steps of the above-described vehicle trajectory determination method, see [link to relevant documentation]. Figure 5 The device includes:

[0126] The acquisition module 501 is used to acquire the status data of the target vehicle, which represents the status of the target vehicle at the current moment.

[0127] Module 502 is used to create an initial behavior decision tree based on state data. The root node in the initial behavior decision tree represents the behavior performed by the target vehicle at the current moment. Other nodes in the initial behavior decision tree, excluding the root node, represent the behaviors that the target vehicle can perform in the future time period. The behaviors that the target vehicle can perform include three categories: changing lanes to the left, keeping the original lane, and changing lanes to the right.

[0128] The first processing module 503 is used to prune the initial behavior decision tree based on at least one of the first heuristic rule and the second heuristic rule to obtain the target behavior decision tree. The first heuristic rule is used to indicate that the number of behavior changes of the target vehicle in the future time period does not exceed a preset number, and the second heuristic rule is used to indicate that the target vehicle meets the lane-changing conditions.

[0129] The second processing module 504 is used to fit the time and vehicle position in the future time period for each behavior sequence in the target behavior decision tree based on multiple behavior and state data in the behavior sequence, so as to obtain the driving trajectory corresponding to the behavior sequence.

[0130] In some embodiments, module 502 is created for:

[0131] Based on the target vehicle's longitudinal position, lateral position, first speed, second speed, yaw angle, and yaw rate in the state data, determine the target vehicle's behavior at the current moment. The first speed is the speed along the vehicle body direction, and the second speed is the speed perpendicular to the vehicle body direction.

[0132] Based on the behavior of the target vehicle at the current moment, create the root node of the initial behavior decision tree;

[0133] Based on the behavior set, other nodes in the initial behavior decision tree are created on the basis of the root node. The behavior set includes three types of behaviors that the target vehicle can perform.

[0134] In the initial behavior decision tree, each node other than its child nodes has three child nodes. The three child nodes of each node are used to represent changing lanes to the left, keeping the original lane, and changing lanes to the right, respectively.

[0135] In some embodiments, the first processing module 503 is configured to:

[0136] Based on the first heuristic rule, multiple first nodes are determined from the initial behavior decision tree. The number of target node pairs between each first node and the root node is equal to the preset number of times. There is a parent-child relationship between the two nodes in the target node pair, and the behaviors represented by the two nodes are different.

[0137] For any one of the multiple first nodes, remove the second node corresponding to the first node and all nodes after the second node from the initial behavior decision tree. The second node is a child node of the first node, and the behavior represented by the second node is different from the behavior represented by the first node.

[0138] In some embodiments, the first processing module 503 is configured to:

[0139] Based on the second heuristic rule, the first warning distance, first lane change gap, second warning distance, and second lane change gap of the target vehicle at the current moment are obtained. The first warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when it changes lanes to the left. The first lane change gap is the distance between the target vehicle and surrounding vehicles when it changes lanes to the left. The second warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when it changes lanes to the right. The second lane change gap is the distance between the target vehicle and surrounding vehicles when it changes lanes to the right.

[0140] Based on the first warning distance, the first lane change gap, the second warning distance, and the second lane change gap, the nodes corresponding to the target behavior sequence are removed from the initial behavior decision tree to obtain the target behavior decision tree. The target behavior sequence includes the target lane change behavior, and the warning distance corresponding to the target lane change behavior is less than the corresponding lane change gap.

[0141] In some embodiments, the first processing module 503 is configured to:

[0142] Based on the first heuristic rule, the initial behavior decision tree is pruned for the first time to obtain the first intermediate behavior decision tree;

[0143] Based on the second heuristic rule, the intermediate behavior decision tree is pruned a second time to obtain the target behavior decision tree.

[0144] In some embodiments, the apparatus further includes:

[0145] The third processing module is used to determine multiple third nodes from the initial behavior decision tree, each third node representing the opposite behavior to that represented by its parent node; and to remove multiple third nodes and all nodes after the third nodes from the initial behavior decision tree to obtain the second intermediate behavior decision tree.

[0146] The first processing module 503 is used to prune the second intermediate behavior decision tree based on at least one of the first heuristic rule and the second heuristic rule to obtain the target behavior decision tree.

[0147] In some embodiments, the second processing module 504 includes:

[0148] The simulation unit is used to perform dynamic simulation of the target vehicle for each behavior sequence in the target behavior decision tree, based on multiple behavior and state data in the behavior sequence, and to determine the destination position of the target vehicle after a future time period.

[0149] The processing unit is used to perform polynomial fitting between the lateral starting position and the lateral ending position of the target vehicle at the current time and the time in the future time period to obtain the lateral trajectory of the target vehicle. The lateral trajectory is used to represent the change of the lateral position of the target vehicle in the future time period.

[0150] The determination unit is used to determine the acceleration at multiple points in the future time period using a longitudinal car-following model. Based on the acceleration at multiple points in the future time period and the longitudinal starting position of the target vehicle at the current time, the longitudinal trajectory of the target vehicle is determined. The longitudinal trajectory is used to represent the change of the longitudinal position of the target vehicle in the future time period.

[0151] In some embodiments, the simulation unit is configured to:

[0152] For each behavior sequence in the target behavior decision tree, based on the behavior and state data to be executed at multiple time points in the behavior sequence, a dynamic simulation is performed on the target vehicle to obtain the state data of the target vehicle at multiple time points. The state data corresponding to each time point includes the longitudinal position of the target vehicle, the road adhesion coefficient, and the vehicle speed at that time point.

[0153] Clustering the state data at multiple time points yields multiple sets of state data;

[0154] For any set of state data in multiple sets of state data, the average of multiple state data within the set of state data is calculated to obtain the central state data corresponding to the set of state data.

[0155] Based on the central state data, a dynamic simulation is performed on the target vehicle to obtain the final position of the target vehicle after a future time period.

[0156] This application provides a device for determining driving trajectories. It can generate an initial behavior decision tree in real time based on the current behavior of a target vehicle. Furthermore, it can design first heuristic rules for behavior change restrictions and primary vehicles based on human driving habits to perform initial pruning on the initial behavior decision tree. Secondly, it can also establish a human-like lane-change warning model considering vehicle and driver heterogeneity to perform secondary pruning of the behavior decision tree from a safety perspective. Finally, based on the state of vehicle behavior within each decision cycle in the target behavior decision tree, it generates driving trajectories corresponding to each behavior sequence in the target behavior decision tree, providing options for the target vehicle. This reduces the complexity of the decision-making algorithm, improves the efficiency of driving trajectory determination, and also helps ensure the robustness of the subsequently determined driving trajectory, making it more practical.

[0157] It should be noted that the above embodiments of the vehicle trajectory determination device, when calculating the vehicle trajectory of a target vehicle in a future time period, only illustrate the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the vehicle trajectory determination device and the vehicle trajectory determination method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, which will not be repeated here.

[0158] In the embodiments of this application, the computer device can be configured as a terminal (e.g., the target vehicle itself) or a server. When the computer device is configured as a terminal, the terminal can act as the execution subject to implement the technical solution provided in the embodiments of this application. When the computer device is configured as a server, the server can act as the execution subject to implement the technical solution provided in the embodiments of this application. Alternatively, the technical solution provided in this application can be implemented through the interaction between the terminal and the server. The embodiments of this application do not limit this.

[0159] Figure 6 This is a schematic diagram of the structure of a vehicle controller for a target vehicle according to an embodiment of this application.

[0160] Typically, the vehicle controller 600 includes: a main control module 601, a CAN interface 602, a hard-wired input interface 603, and a hard-wired output interface 604. The main control module 601 is connected to the CAN interface 602, the hard-wired input interface 603, and the hard-wired output interface 604, respectively.

[0161] The main control module 601 typically includes a processor and memory. The processor may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor can be implemented using at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array). The processor may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content required to be displayed on the vehicle's screen. In some embodiments, the processor may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning. The memory may include one or more computer-readable storage media, which may be non-transitory. The memory may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, a non-transitory computer-readable storage medium in the memory is used to store at least one computer program, which is executed by a processor to implement the method for determining a vehicle trajectory provided in the method embodiments of this application.

[0162] The CAN interface 602 may include a powertrain CAN interface, a motor CAN interface, and a diagnostic CAN interface. The powertrain CAN interface is used to communicate with the vehicle's powertrain module, the motor CAN interface is used to communicate with the vehicle's motor controller, and the diagnostic CAN interface is used to communicate with diagnostic equipment.

[0163] The hard-wired input interface 603 is used to receive hard-wired control signals. The hard-wired output interface 604 is used to send control commands to the vehicle's electronic control components, causing the vehicle's electronic control components to perform corresponding actions. The vehicle's electronic control components include a power management system, a motor controller, an on-board charger, and a body control system.

[0164] The main control module 601 can communicate with the vehicle's powertrain module, motor controller, and diagnostic equipment via the CAN interface 602, and generate control commands based on the hard-wired control signals received by the hard-wired input interface 603, so as to send the control commands to the vehicle's electronic control components via the hard-wired output interface 604.

[0165] Those skilled in the art will understand that Figure 6 The structure shown does not constitute a limitation on the vehicle controller 600, and may include more or fewer components than shown, or combine certain components, or use different component arrangements.

[0166] Figure 7 This is a schematic diagram of a server structure according to an embodiment of this application. The server 700 can vary considerably due to different configurations or performance. It may include one or more Central Processing Units (CPUs) 701 and one or more memories 702. The memory 702 stores at least one computer program, which is loaded and executed by the processor 701 to implement the vehicle trajectory determination method provided in the above-described method embodiments. Of course, the server may also have wired or wireless network interfaces, a keyboard, and input / output interfaces for input and output. The server may also include other components for implementing device functions, which will not be elaborated here.

[0167] This application also provides a computer-readable storage medium storing at least one computer program. This computer program is loaded and executed by a processor of a computer device to implement the operations performed by the computer device in the vehicle trajectory determination method of the above embodiments. For example, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device, etc.

[0168] In some embodiments, the computer program involved in the present application embodiments may be deployed and executed on a computer device, or executed on multiple computer devices located in one location, or executed on multiple computer devices distributed in multiple locations and interconnected through a communication network. Multiple computer devices distributed in multiple locations and interconnected through a communication network may constitute a blockchain system.

[0169] This application also provides a computer program product, including a computer program stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium and executes the computer program, causing the computer device to perform the vehicle trajectory determination method provided in the various optional implementations described above.

[0170] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0171] The above description is merely an optional embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for determining a vehicle trajectory, characterized in that, The method includes: Acquire the status data of the target vehicle, wherein the status data is used to represent the status of the target vehicle at the current moment; Based on the state data, an initial behavior decision tree is created. The root node in the initial behavior decision tree represents the behavior performed by the target vehicle at the current moment. Other nodes in the initial behavior decision tree, excluding the root node, represent the behaviors that the target vehicle can perform in a future time period. The behaviors that the target vehicle can perform include three categories: changing lanes to the left, maintaining the original lane, and changing lanes to the right. Based on at least one of the first heuristic rule and the second heuristic rule, the initial behavior decision tree is pruned to obtain the target behavior decision tree. The first heuristic rule is used to indicate that the number of behavior changes of the target vehicle within the future time period does not exceed a preset number, and the second heuristic rule is used to indicate that the target vehicle meets the lane-changing conditions. For each behavior sequence in the target behavior decision tree, based on multiple behaviors in the behavior sequence and the state data, the time and vehicle position within the future time period are fitted to obtain the driving trajectory corresponding to the behavior sequence.

2. The method for determining the driving trajectory according to claim 1, characterized in that, The step of creating an initial behavior decision tree based on the state data includes: Based on the longitudinal position, lateral position, first speed, second speed, yaw angle, and yaw rate of the target vehicle in the state data, the behavior performed by the target vehicle at the current moment is determined, where the first speed is the speed along the vehicle body direction and the second speed is the speed perpendicular to the vehicle body direction. Based on the behavior of the target vehicle at the current moment, create the root node of the initial behavior decision tree; Based on the behavior set, other nodes in the initial behavior decision tree, excluding the root node, are created on the basis of the root node. The behavior set includes three types of behaviors that the target vehicle can perform. In the initial behavior decision tree, each node other than its child nodes has three child nodes. The three child nodes of each node are used to represent changing lanes to the left, keeping the original lane, and changing lanes to the right, respectively.

3. The method for determining the driving trajectory according to claim 1, characterized in that, The step of pruning the initial behavior decision tree based on at least one of the first and second heuristic rules to obtain the target behavior decision tree includes: Based on the first heuristic rule, multiple first nodes are determined from the initial behavior decision tree. The number of target node pairs between each first node and the root node is equal to the preset number of times. There is a parent-child relationship between the two nodes in the target node pair, and the behaviors represented by the two nodes are different. For any one of the plurality of first nodes, remove the second node corresponding to the first node and all nodes after the second node from the initial behavior decision tree. The second node is a child node of the first node, and the behavior represented by the second node is different from the behavior represented by the first node.

4. The method for determining the driving trajectory according to claim 1, characterized in that, The step of pruning the initial behavior decision tree based on at least one of the first and second heuristic rules to obtain the target behavior decision tree includes: Based on the second heuristic rule, the first warning distance, first lane change gap, second warning distance, and second lane change gap of the target vehicle at the current moment are obtained. The first warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when it changes lanes to the left. The first lane change gap is the distance between the target vehicle and surrounding vehicles when it changes lanes to the left. The second warning distance is the minimum safe distance between the target vehicle and surrounding vehicles when it changes lanes to the right. The second lane change gap is the distance between the target vehicle and surrounding vehicles when it changes lanes to the right. Based on the first warning distance, the first lane change gap, the second warning distance, and the second lane change gap, the node corresponding to the target behavior sequence is removed from the initial behavior decision tree to obtain the target behavior decision tree. The target behavior sequence includes the target lane change behavior, and the warning distance corresponding to the target lane change behavior is less than the corresponding lane change gap.

5. The method for determining the driving trajectory according to claim 1, characterized in that, The step of pruning the initial behavior decision tree based on at least one of the first and second heuristic rules to obtain the target behavior decision tree includes: Based on the first heuristic rule, the initial behavior decision tree is pruned for the first time to obtain the first intermediate behavior decision tree; Based on the second heuristic rule, the intermediate behavior decision tree is pruned a second time to obtain the target behavior decision tree.

6. The method for determining the driving trajectory according to claim 1, characterized in that, The method further includes: From the initial behavior decision tree, multiple third nodes are determined, and the behavior represented by each third node is the opposite of the behavior represented by its parent node; From the initial behavior decision tree, remove the plurality of third nodes and all nodes after the third nodes to obtain the second intermediate behavior decision tree; The step of pruning the initial behavior decision tree based on at least one of the first and second heuristic rules to obtain the target behavior decision tree includes: Based on at least one of the first and second heuristic rules, the second intermediate behavior decision tree is pruned to obtain the target behavior decision tree.

7. The method for determining the driving trajectory according to claim 1, characterized in that, For each behavior sequence in the target behavior decision tree, based on multiple behaviors in the behavior sequence and the state data, the time and vehicle position within the future time period are fitted to obtain the driving trajectory corresponding to the behavior sequence, including: For each behavior sequence in the target behavior decision tree, based on multiple behaviors in the behavior sequence and the state data, a dynamic simulation is performed on the target vehicle to determine the destination position of the target vehicle after the future time period. Based on the target vehicle's lateral starting position at the current moment and the lateral ending position within the starting position, a polynomial fit is performed with the time within the future time period to obtain the lateral trajectory of the target vehicle. The lateral trajectory is used to represent the change of the target vehicle's lateral position within the future time period. A longitudinal car-following model is used to determine the acceleration at multiple points in time within the future time period. Based on the acceleration at these multiple points in time and the longitudinal starting position of the target vehicle at the current time, the longitudinal trajectory of the target vehicle is determined. The longitudinal trajectory is used to represent the change in the longitudinal position of the target vehicle within the future time period.

8. The method for determining the driving trajectory according to claim 7, characterized in that, For each behavior sequence in the target behavior decision tree, based on multiple behaviors in the behavior sequence and the state data, a dynamic simulation is performed on the target vehicle to determine the destination position of the target vehicle after the future time period, including: For each behavior sequence in the target behavior decision tree, based on the behavior to be executed at multiple times in the behavior sequence and the state data, the target vehicle is subjected to dynamic simulation to obtain the state data of the target vehicle at multiple times. The state data corresponding to each time time includes the longitudinal position of the target vehicle, the road adhesion coefficient and the vehicle speed at that time. Cluster the state data at the multiple time points to obtain multiple sets of state data; For any one set of state data in the multiple sets of state data, the average of the multiple state data in the set of state data is calculated to obtain the center state data corresponding to the set of state data. Based on the central state data, a dynamic simulation is performed on the target vehicle to obtain the destination position of the target vehicle after the future time period.

9. A device for determining a vehicle trajectory, characterized in that, The device includes: The acquisition module is used to acquire the status data of the target vehicle, wherein the status data is used to represent the status of the target vehicle at the current moment; A creation module is used to create an initial behavior decision tree based on the state data. The root node in the initial behavior decision tree represents the behavior performed by the target vehicle at the current moment. Other nodes in the initial behavior decision tree, excluding the root node, represent the behaviors that the target vehicle can perform in a future time period. The behaviors that the target vehicle can perform include three categories: changing lanes to the left, maintaining the original lane, and changing lanes to the right. The first processing module is used to prune the initial behavior decision tree based on at least one of a first heuristic rule and a second heuristic rule to obtain a target behavior decision tree. The first heuristic rule is used to indicate that the number of behavior changes of the target vehicle in the future time period does not exceed a preset number, and the second heuristic rule is used to indicate that the target vehicle meets the lane-changing conditions. The second processing module is used to fit the time and vehicle position in the future time period for each behavior sequence in the target behavior decision tree based on multiple behaviors in the behavior sequence and the state data, so as to obtain the driving trajectory corresponding to the behavior sequence.

10. A computer device, characterized in that, The computer device includes a processor and a memory, the memory being used to store at least one computer program, the at least one computer program being loaded by the processor and executed as the method for determining the driving trajectory according to any one of claims 1 to 8.

Images