Vehicle control method, apparatus, medium, and device

Abstract

Embodiments of the present disclosure disclose a vehicle control method, device, medium and equipment, wherein the method comprises: determining first vehicle state information of a vehicle at a first time; determining first object state information of a dynamic object at the first time, and second object state information of a static object at the first time; determining first object predicted trajectory information of the dynamic object at a second time based on the first object state information; determining an initial trajectory tree of the vehicle at the second time based on the first object predicted trajectory information and the first vehicle state information; determining constraint information based on the first object predicted trajectory information and the second object state information; constructing a target function based on the constraint information; updating the initial trajectory tree based on the first vehicle state information and the target function to determine a target trajectory tree; and controlling the vehicle to travel based on the target trajectory tree. The embodiments of the present disclosure can improve the adaptability to roads with different curvatures and the robustness of trajectories in dynamic scenarios.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to intelligent driving technology, and in particular, to a vehicle control method, device, medium and equipment. BACKGROUND

[0002] Planning and control function is one of the core functions of intelligent driving, which involves planning a safe and collision-free ego vehicle trajectory for the ego vehicle to avoid, slow down, overtake and other actions of dynamic and static traffic participants. When planning the trajectory, the comfort of the driver and passengers should be considered on the basis of meeting the safety and collision-free requirements. In related technologies, the ego vehicle trajectory is usually planned based on the horizontal and vertical separation planning method of the Frenet coordinate system. However, the horizontal and vertical separation planning method based on the Frenet coordinate system needs to calibrate multiple sets of horizontal parameters for roads with different curvatures, and needs to calibrate parameters related to speed for different vehicle speeds. These parameters are closely related to the longitudinal direction, and the parameters are not easy to adjust, so the adaptability to different road curvatures is low, and the trajectory robustness in dynamic scenarios is low. SUMMARY

[0003] Embodiments of the present disclosure provide a vehicle control method, device, medium and equipment to improve the adaptability to roads with different curvatures and improve the robustness of the trajectory in dynamic scenarios.

[0004] In a first aspect, a vehicle control method is provided, comprising: determining first vehicle state information of a vehicle at a first time; determining first object state information of a dynamic object at the first time, and second object state information of a static object at the first time; determining first object predicted trajectory information of the dynamic object at a second time based on the first object state information, the first time being earlier than the second time; determining an initial trajectory tree of the vehicle at the second time based on the first object predicted trajectory information and the first vehicle state information; determining constraint information based on the first object predicted trajectory information and the second object state information; constructing an objective function based on the constraint information; updating the initial trajectory tree based on the first vehicle state information and the objective function to determine a target trajectory tree; and controlling the vehicle to travel based on the target trajectory tree.

[0005] In a second aspect, the present disclosure provides a vehicle control device, comprising: a first processing module configured to determine first vehicle state information of a vehicle at a first time; a second processing module configured to determine first object state information of a dynamic object at the first time, and second object state information of a static object at the first time; a third processing module configured to determine, based on the first object state information, first object predicted trajectory information of the dynamic object at a second time, the first time being earlier than the second time; a fourth processing module configured to determine, based on the first object predicted trajectory information and the first vehicle state information, an initial trajectory tree of the vehicle at the second time; a fifth processing module configured to determine, based on the first object predicted trajectory information and the second object state information, constraint information; a sixth processing module configured to construct, based on the constraint information, an objective function; a seventh processing module configured to perform iterative update on the initial trajectory tree based on the first vehicle state information and the objective function, to determine a target trajectory tree; and a control module configured to control the vehicle to travel based on the target trajectory tree.

[0006] In a third aspect, the present disclosure provides a computer readable storage medium, the storage medium storing a computer program, the computer program being executed by a processor to implement the vehicle control method according to any one of the above embodiments.

[0007] In a fourth aspect, the present disclosure provides an electronic device, the electronic device comprising: a processor; a memory configured to store executable instructions for the processor; and the processor configured to read the executable instructions from the memory and execute the instructions to implement the vehicle control method according to any one of the above embodiments.

[0008] In a fifth aspect, the present disclosure provides a computer program product, when instructions in the computer program product are executed by a processor, the vehicle control method according to any one of the above embodiments is performed.

[0009] Based on the vehicle control method, device, medium and equipment provided by the above embodiments of the present disclosure, by determining the first vehicle state information of the vehicle at the first time, determining the first object state information of the dynamic object at the first time and the second object state information of the static object at the first time, the first object prediction trajectory information of the dynamic object at the second time can be determined based on the first object state information, and then the initial trajectory tree of the vehicle at the second time can be determined based on the first object prediction trajectory information and the first vehicle state information; the constraint information is determined based on the first object prediction trajectory information and the second object state information; the target function is constructed based on the constraint information; the initial trajectory tree is updated based on the first vehicle state information and the target function, and the target trajectory tree is determined, so that the vehicle can be controlled to travel based on the target trajectory tree. Since the initial trajectory tree represents the initial trajectory of different trajectory branches of the vehicle under the influence of the predicted trajectory of the dynamic object, the constraint information represents the safety constraint of the vehicle drivable area by various dynamic objects and static objects around the vehicle, and the target function constructed based on the constraint information can guide the update of the initial trajectory tree, so that each trajectory branch in the target trajectory tree obtained by updating can meet the constraint conditions of the vehicle drivable area by various dynamic objects and static objects at the second time, ensuring the safety of each trajectory branch of the target trajectory tree, and the entire trajectory tree is updated, the influence of various dynamic and static objects around the vehicle on any trajectory branch is comprehensively considered, so that the state and control amount of any node in the target trajectory tree are more consistent with the actual road environment, the robustness of the vehicle trajectory is improved, and the safety and stability of the vehicle driving are ensured. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 is an exemplary application scenario of the vehicle control method provided by the present disclosure;

[0011] Figure 2 is a flowchart of the vehicle control method provided by an exemplary embodiment of the present disclosure;

[0012] Figure 3 is a schematic diagram of a trajectory tree provided by an exemplary embodiment of the present disclosure;

[0013] Figure 4 is a flowchart of the vehicle control method provided by another exemplary embodiment of the present disclosure;

[0014] Figure 5 is a flowchart of the vehicle control method provided by still another exemplary embodiment of the present disclosure;

[0015] Figure 6 is a visualization diagram of dynamic and static boundaries provided by an exemplary embodiment of the present disclosure;

[0016] Figure 7 is a flowchart of the vehicle control method provided by yet another exemplary embodiment of the present disclosure;

[0017] Figure 8 is a schematic diagram of a kinematic model provided by an example embodiment of the present disclosure;

[0018] Figure 9 is a flowchart of a vehicle control method provided by another example embodiment of the present disclosure;

[0019] Figure 10 is a multi-modal scene diagram provided by an example embodiment of the present disclosure;

[0020] Figure 11 is a structural diagram of a vehicle control device provided by an example embodiment of the present disclosure;

[0021] Figure 12 is a structural diagram of a vehicle control device provided by another example embodiment of the present disclosure;

[0022] Figure 13 is a structural diagram of an electronic device provided by an example embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] In order to explain the present disclosure, the example embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings, it is obvious that the described embodiments are only part of the embodiments of the present disclosure, but not all the embodiments, and it should be understood that the present disclosure is not limited by the example embodiments.

[0024] It should be noted that: unless otherwise specified, the relative arrangement, numerical expression and numerical value of the components and steps set forth in these embodiments do not limit the scope of the present disclosure.

[0025] SUMMARY

[0026] In the process of implementing the present disclosure, the inventors found that in the related art of vehicle control, the ego vehicle trajectory is usually planned based on the lateral-longitudinal separation planning mode of the Frenet coordinate system. However, the lateral-longitudinal separation planning mode based on the Frenet coordinate system needs to calibrate multiple sets of lateral parameters for roads with different curvatures, and needs to calibrate parameters related to speed for different vehicle speeds. These parameters are closely related to the longitudinal direction, and the parameters are not easy to adjust, so the adaptability to different road curvatures is low, and the robustness of the planned trajectory in dynamic scenes is low.

[0027] EXEMPLARY SUMMARY

[0028] Figure 1 is an example application scenario of the vehicle control method provided by the present disclosure. As shown in Figure 1As shown, during the driving of the vehicle (i.e., ego vehicle) 11, there can be objects such as dynamic objects (or dynamic obstacles), static objects, etc. around the ego vehicle 11 that affect the driving of the vehicle, and the dynamic objects can include, for example, other vehicles (e.g., dynamic object 12 in the figure), pedestrians, cyclists, animals, etc. around the ego vehicle 11. The static objects can include traffic cones, fences, other vehicles parked, lane lines, dashed lines, double yellow lines, guide lines, etc. map elements or traffic rule constraint elements. The surrounding environment can be perceived based on the sensors 13 provided on the ego vehicle 11, and the trajectory planning and control of the vehicle 11 can be implemented by using the vehicle control method provided in the embodiments of the present disclosure. Specifically, the first vehicle state information of the vehicle 11 at a first time can be determined, the first object state information of the dynamic object at the first time and the second object state information of the static object at the first time can be determined, the first object prediction trajectory information of the dynamic object at a second time can be determined based on the first object state information, and then the initial trajectory tree of the vehicle at the second time can be determined based on the first object prediction trajectory information and the first vehicle state information; the constraint information can be determined based on the first object prediction trajectory information and the second object state information; the objective function can be constructed based on the constraint information; the initial trajectory tree can be updated based on the first vehicle state information and the objective function to determine the target trajectory tree, so that the driving of the vehicle 11 can be controlled based on the target trajectory tree. Since the initial trajectory tree represents the initial trajectories of different trajectory branches of the vehicle 11 under the influence of the predicted trajectory of the dynamic object, as shown in the figure, the dynamic object 12 has at least one possible predicted trajectory, for example, the lane keeping trajectory L1 and the lane switching trajectory L2 in the figure, and according to the predicted trajectory of the dynamic object 12, a plurality of possible trajectory branches of the vehicle 11 are determined, for example, the lane keeping trajectory L3 and the detour trajectory L4 in the figure. The constraint information represents the safety constraints of the drivable area of the vehicle 11 around various dynamic objects and static objects, and the objective function constructed based on the constraint information can guide the update of the initial trajectory tree, so that each trajectory branch in the target trajectory tree obtained by updating can satisfy the constraint conditions of the drivable area of the vehicle 11 around various dynamic objects and static objects at the second time, ensuring the safety of each trajectory branch in the target trajectory tree, and the overall trajectory tree is updated, considering the influence of various dynamic and static objects around the vehicle on any trajectory branch in the trajectory tree, so that the state and control amount of the node (or trajectory node) in the target trajectory tree are more consistent with the actual road environment, improving the robustness of the vehicle trajectory in the dynamic scene, and ensuring the safety and stability of the driving of the vehicle.

[0029] Exemplary Method

[0030] Figure 2 is a flowchart of a vehicle control method provided by an example embodiment of the present disclosure. The vehicle control method provided in the embodiment of the present disclosure can be applied on an electronic device, specifically, for example, on a vehicle-mounted computing platform, such as Figure 2As shown, the method of the embodiments of the present disclosure can include the following steps:

[0031] Step 210, determining first vehicle state information of the vehicle at a first time.

[0032] The first time can be any time, for example, the first time can be the current time or any historical time. In the intelligent driving scenario, the first time is, for example, the current time during the driving of the vehicle. The first vehicle state information can include the kinematic information of the vehicle at the first time, and the kinematic information can include the position, speed, acceleration, yaw, yaw rate, etc. of the vehicle.

[0033] In some optional embodiments, the first vehicle state information of the vehicle at the first time can be collected based on the positioning sensors, speed sensors, angular velocity sensors, etc. related to the vehicle state on the vehicle.

[0034] Step 220, determining first object state information of a dynamic object at the first time, and second object state information of a static object at the first time.

[0035] The dynamic object is an obstacle in a moving state in the surrounding environment of the vehicle. The dynamic object can include other vehicles in the surrounding of the vehicle, pedestrians, cyclists, etc. The static object is an object in a stationary state in the surrounding environment of the vehicle, and the static object can include static obstacles such as traffic cones, fences, etc., and map elements or road traffic elements that constrain the behavior of the vehicle, such as lane lines, guide lines, etc.

[0036] In some optional embodiments, the first object state information of the dynamic object can include the kinematic information, size information, type information, etc. of the dynamic object at the first time. The kinematic information of the dynamic object can include the position, speed, acceleration, yaw, yaw rate, etc. of the dynamic object in the local coordinate system of the vehicle. The size information of the dynamic object can include the length, width, etc. of the dynamic object. The type information of the dynamic object refers to the type to which the dynamic object belongs, and the type can include other vehicles, pedestrians, cyclists, animals, etc. The local coordinate system of the vehicle refers to a coordinate system with a preset position of the vehicle as the origin, for example, the local coordinate system can be the ego coordinate system of the vehicle at the first time.

[0037] In some optional embodiments, the second object state information of the static object can include the state information of the static object at the first time, and the state information can include the position, orientation, size, type, etc. of the static object in the local coordinate system of the vehicle. The size of the static object can include the length, width, etc. of the static object. The type of the static object can include the above-mentioned traffic cones, fences, lane lines, guide lines, other vehicles parked in a stationary state, etc.

[0038] In some optional embodiments, the object in the surrounding environment of the vehicle can be perceived by a sensor arranged on the vehicle to perceive the environment, to obtain the first object state information of the dynamic object and the second object state information of the static object. For example, during the driving of the vehicle, the first object state information of the dynamic object and the second object state information of the static object can be determined based on the sensor data collected by the sensor at the current time through a perception algorithm model.

[0039] At step 230, the first object prediction trajectory information of the dynamic object at the second time is determined based on the first object state information, the first time being earlier than the second time.

[0040] The first time being earlier than the second time means that the second time is a future time relative to the first time. For example, the first time is the current time, and the second time is a future time. The first object prediction trajectory information of the dynamic object at the second time is the trajectory information of the dynamic object at the second time obtained by prediction. The first object prediction trajectory information can include one or more of the position, speed, acceleration, yaw, yaw rate, etc. of the dynamic object at the second time.

[0041] In some optional embodiments, the number of the second times is one or more, which can be represented as N, N being a positive integer. The number of the second times can be pre-set, or the number of the second times can be determined according to the length of time to be predicted and the time interval. For example, the trajectory information of the dynamic object at a future T seconds is predicted at the current time, and the time interval is t seconds, so the number of the second times is T / t. The first object prediction trajectory information can include the prediction state of the dynamic object at the N second times respectively. For any second time, the prediction state of the dynamic object can include one or more states, that is, the first object prediction trajectory information can include at least one prediction trajectory information of the dynamic object at the second time, see Figure 1 In the first object prediction trajectory information of the dynamic object 12, the first object prediction trajectory information includes the lane keeping trajectory L1 and the lane switching trajectory L2. Optionally, the first object prediction trajectory information includes at least one prediction trajectory of the dynamic object and the probability corresponding to the prediction trajectory, the probability representing the confidence of the dynamic object driving according to the prediction trajectory. The greater the probability, the greater the possibility of the dynamic object driving according to the prediction trajectory. The sum of the probabilities corresponding to the at least one prediction trajectory in the first object prediction trajectory information is 1. For example Figure 1 In the first object prediction trajectory information of the dynamic object 12, the probability corresponding to L1 is 0.3, and the probability corresponding to L2 is 0.7.

[0042] In some optional embodiments, the first object prediction trajectory information of the dynamic object at the second time can be predicted based on the first object state information through a pre-configured trajectory prediction algorithm model.

[0043] In some optional embodiments, the number of dynamic objects can be one or more, and the first object prediction trajectory information of any dynamic object can be determined.

[0044] In some optional embodiments, the first object prediction trajectory information of the dynamic object at the second time can be predicted based on the third object state information of the dynamic object at the third time and the first object state information of the dynamic object at the first time. The third time is earlier than the first time. The number of third times is one or more. For example, the first time is the current time, and the third time is a historical time. The first object prediction trajectory information of the dynamic object at the future time is predicted based on the third object state information of the dynamic object at the historical time and the first object state information of the dynamic object at the current time. The state information of the dynamic object at the plurality of times of the first time and the third time can effectively represent the motion trend of the dynamic object, which helps to improve the effectiveness and stability of the predicted first object prediction trajectory information.

[0045] In step 240, the initial trajectory tree of the vehicle at the second time is determined based on the first object prediction trajectory information and the first vehicle state information.

[0046] The initial trajectory tree of the vehicle at the second time represents the initial state quantity and the initial control quantity of at least one trajectory of the vehicle at the nodes (or trajectory points) of different second times. For example Figure 1 In the above formula, L3 and L4 represent two possible trajectories of the vehicle, and the initial trajectory tree can include the initial state quantity and the initial control quantity of the nodes corresponding to the N second times of the two trajectories L3 and L4, respectively.

[0047] In some optional embodiments, state variables represent the possible states of the vehicle at that node. State variables may include one or more of the following: position, velocity, acceleration, heading angle, wheel angle, wheel angular velocity, and distance traveled. Control variables represent the control parameters of the vehicle's actuators. Control variables may include wheel angular acceleration and longitudinal impact (or jerk). Position may include longitudinal position x and lateral position y. Velocity may be represented as v, acceleration as a, heading angle as θ, wheel angle as δ, wheel angular velocity as ω, distance traveled represents the length of the trajectory traveled by the vehicle, or arc length, and distance traveled may be represented as s. Wheel angular acceleration may be represented as α, longitudinal impact may be represented as jerk or J, state variables may be represented as X = [x, y, θ, δ, ω, s, v, a], and control variables may be represented as u = [α, jerk]. Initial state variables and initial control variables are the initial values ​​of the state variables and control variables determined for the nodes of the trajectory tree. Initial state variables may be represented as X0, and initial control variables may be represented as u0. Since the initial trajectory tree can include nodes of at least one trajectory at multiple second time points, the initial state of the node at the i-th second time point in the j-th trajectory can be expressed as: The initial control variable of the node at the second moment in the j-th trajectory can be expressed as: At least one trajectory may have shared nodes and its own branch nodes, see [link to relevant documentation]. Figure 1 In the L3 and L4 trajectories, after passing through a common trajectory, the two trajectories branch into two branches. The nodes on the common trajectory are shared nodes for both trajectories. Shared nodes can have the same state variables and / or control variables, or shared nodes can have state variables and control variables corresponding to different trajectories. The nodes on the two branches are the branch nodes of L3 and L4, respectively. The state variables of nodes at the same second time step on different branches are different.

[0048] In some optional examples, Figure 3 This is a schematic diagram of a trajectory tree provided in an exemplary embodiment of this disclosure. For example... Figure 3 As shown, the trajectory tree includes three types of trajectories. Circles represent nodes in the trajectory tree. The first time step is T0, the second time step includes T1, T2, ..., T... n+6 Time. n is a positive integer. P(0) represents the node at time T0, P(1)~P(n) represent the nodes shared by the three trajectories, n is a positive integer, P(i) represents the time T0. i The nodes at time points i = 1, 2, ..., n, P(n) is the bifurcation point, and P(n+1)1 to P(n+3)1 are the points T of the first type of trajectory. n+1 Time, T n+2 Time and T n+3The nodes corresponding to the time instants, P(n+1)2~P(n+3)2 are nodes shared by the second trajectory and the third trajectory, and correspond to T n+1 The time instants, T n+2 The time instants, T n+3 The time instants, T 21 P(n+3)2 is a bifurcation point, P(n+4) 21 ~P(n+6) n+4 The nodes corresponding to the time instants, P(n+1)2~P(n+3)2 are nodes shared by the second trajectory and the third trajectory, and correspond to T n+5 The time instants, T n+6 The time instants, T 22 The time instants, T 22 ~P(n+6) n+4 The nodes corresponding to the time instants, P(n+1)2~P(n+3)2 are nodes shared by the second trajectory and the third trajectory, and correspond to T n+5 The time instants, T n+6 The time instants, T Figure 3 The nodes in the initial trajectory tree correspond to initial state quantities and initial control quantities.

[0049] In some optional embodiments, the motion of the dynamic object relative to the vehicle can be determined based on the first object prediction trajectory information, and the nodes corresponding to at least one trajectory of the vehicle at the second time instant can be determined based on the motion of the dynamic object relative to the vehicle. The initial state quantities and the initial control quantities corresponding to the nodes can be determined based on the first vehicle state information and the preconfigured kinematic model of the vehicle, to obtain the initial trajectory tree.

[0050] In step 250, constraint information is determined based on the first object prediction trajectory information and the second object state information.

[0051] The constraint information is constraint information of the static object and the dynamic object around the vehicle on the drivable region of the vehicle, and can be used to constrain the trajectory of the vehicle at the second time instant, to avoid collision between the vehicle and the static object and the dynamic object, and to ensure the safety of the vehicle. The constraint information can include, but is not limited to, boundary constraint information of the drivable region of the vehicle, speed constraint information of the vehicle, acceleration constraint information, etc.

[0052] In some optional embodiments, the first object prediction trajectory information can represent the region occupied by the dynamic object at the second time instant, and the second object state information of the static object can represent the region occupied by the static object at the second time instant. Because the vehicle should not enter the regions occupied by the static object and the dynamic object, the constraint information can be determined based on the first object prediction trajectory information and the second object state information. Alternatively, the constraint information can be determined based on the first object prediction trajectory information, the size of the dynamic object, the second object state information, and the size of the static object.

[0053] In some optional embodiments, the constraint information can be determined based on the first object state information of the dynamic object at the first time, the first object predicted trajectory information, and the second object state information.

[0054] At step 260, a target function is constructed based on the constraint information.

[0055] The target function is a function of the state variable and the control variable corresponding to the node in the trajectory tree. In the trajectory optimization process, the target function is used to represent the cost of the state variable and the control variable of the node in the trajectory tree under the constraint information.

[0056] In some optional embodiments, the constraint information can include at least one constraint information corresponding to a second time. For any second time, the constraint information of the second time represents the constraint of the area occupied by the dynamic object and the static object on the state of the vehicle at the second time, such as the boundary constraint of the drivable area of the vehicle, the constraint of the vehicle speed and acceleration, etc. Based on the relationship between the state variable and the control variable corresponding to the node at the second time in the trajectory tree and the constraint information, at least one cost sub-function is constructed, and thus the target function is obtained.

[0057] At step 270, the initial trajectory tree is updated based on the first vehicle state information and the target function, and a target trajectory tree is determined.

[0058] The first vehicle state information represents the starting state (or initial state) of the trajectory of the vehicle at the second time. The initial trajectory tree represents the initial state variable and the initial control variable of the node of the vehicle at the second time. Based on the initial state variable and the initial control variable, the control variable of the node can be updated based on the target function, and then based on the first vehicle state information and the updated control variable of the node, the state variable of the node is updated through the kinematics model of the vehicle to obtain the updated state variable. It is determined whether the updated state variable and the control variable meet the iteration end condition through the target function. If the iteration end condition is met, the target trajectory tree can be obtained based on the updated state variable and the control variable of the node. If the iteration end condition is not met, the control variable and the state variable are continuously updated according to the above process until the iteration end condition is met, and the target trajectory tree is obtained.

[0059] In some optional embodiments, the initial trajectory tree can be updated based on a preconfigured optimization algorithm to obtain the target trajectory tree. The optimization algorithm includes but is not limited to an iterative linear quadratic regulator and its improved version.

[0060] At step 280, the vehicle is controlled to travel based on the target trajectory tree.

[0061] Wherein, after obtaining the target trajectory tree, the control parameter of the vehicle can be determined based on the target state quantity and / or the target control quantity of the nodes in the target trajectory tree, and the vehicle is controlled to run according to the control parameter. The control parameter can include but is not limited to wheel angular acceleration and longitudinal impact degree.

[0062] In some optional embodiments, the target trajectory tree includes at least one target trajectory updated by optimization, and for the case that the target trajectory tree includes multiple target trajectories, a target trajectory can be determined according to the probabilities corresponding to the multiple target trajectories respectively, and the control parameter of the vehicle is determined based on the state quantity and / or the controller of the nodes in the target trajectory to control the vehicle to run. The probabilities corresponding to the multiple target trajectories respectively can be determined according to the probabilities of the predicted trajectories of the dynamic objects. Alternatively, the probabilities corresponding to the multiple target trajectories respectively can be determined according to the probabilities of the predicted trajectories of the dynamic objects and the collision risk probabilities of the ego vehicle and the dynamic objects.

[0063] The vehicle control method provided by the embodiments can constrain the initial trajectory tree to represent the initial trajectories of different trajectory branches of the vehicle under the influence of the predicted trajectories of the dynamic objects, constrain the safety of the drivable area of the vehicle by various dynamic objects and static objects around the vehicle, and guide the update of the initial trajectory tree based on the target function constructed based on the constraint information, so that each trajectory branch in the target trajectory tree obtained by the update can meet the constraint conditions of the drivable area of the vehicle by various dynamic objects and static objects at the second time, ensure the safety of each trajectory branch of the target trajectory tree, and update the trajectory tree as a whole to comprehensively consider the influence of various dynamic and static objects around the vehicle on any trajectory branch, so that the state and control quantity of any node in the target trajectory tree are more consistent with the actual road environment, the robustness of the vehicle trajectory is improved, and the safety and stability of the vehicle running are ensured.

[0064] Figure 4 is a flowchart of a vehicle control method provided by another example embodiment of the disclosure.

[0065] In some optional embodiments, in the above Figure 2 based on the embodiments shown in Figure 4 based on the embodiments shown in

[0066] Step 2410, determining the predicted trajectory of the dynamic object and the probability corresponding to the predicted trajectory based on the first object predicted trajectory information.

[0067] Wherein, the first object predicted trajectory information includes at least one predicted trajectory of the dynamic object and the probability corresponding to the predicted trajectory, and the at least one predicted trajectory of the dynamic object and the probability corresponding to the predicted trajectory can be obtained from the first object predicted trajectory information.

[0068] At step 2420, based on the probabilities corresponding to the predicted trajectories, a first dynamic object satisfying a preset condition and a first predicted trajectory of the first dynamic object are determined.

[0069] The preset condition is a condition for determining a dynamic object from which a trajectory tree of the vehicle needs to be constructed. The preset condition can include a quantity condition and a probability condition of the predicted trajectories. The quantity condition refers to a condition that needs to be satisfied by the quantity of the predicted trajectories of the dynamic object, for example, the quantity of the predicted trajectories is greater than a quantity threshold. The probability condition refers to a condition that needs to be satisfied by the probabilities corresponding to the predicted trajectories of the dynamic object, for example, the probability corresponding to the predicted trajectory is greater than a probability threshold.

[0070] In some optional embodiments, the quantity of the dynamic objects is one or more. The first dynamic object satisfying the preset condition can be determined from the dynamic objects according to the matching of the quantity of the predicted trajectories of the dynamic objects and the probabilities corresponding to the predicted trajectories with the preset condition. The predicted trajectory of the first dynamic object is the first predicted trajectory. Any first dynamic object can be referred to as a multi-modal dynamic object if the quantity of the first predicted trajectory of the first dynamic object satisfies the quantity condition, for example, the quantity of the first predicted trajectory is greater than 1. Multi-modal refers to the dynamic object having multiple possible predicted trajectories.

[0071] At step 2430, based on the first vehicle state information and the first predicted trajectory of the first dynamic object, a branching time of a trajectory of the vehicle at the second time is determined.

[0072] The first predicted trajectory of the first dynamic object represents the state of the first dynamic object at the second time. Based on the first vehicle state information and the first predicted trajectory of the first dynamic object, the distance between the vehicle and the first dynamic object at the second time and / or the time to collision (TTC) can be determined. Based on the distance between the vehicle and the first dynamic object and / or the time to collision, the branching time (or branching point) of the trajectory of the vehicle at the second time is determined. For example, if the distance between the vehicle and the first dynamic object at a certain second time is less than a preset safety distance threshold, it indicates that the trajectory of the vehicle can branch at the second time. Alternatively, if the time to collision between the vehicle and the first dynamic object at a certain second time is less than a preset time threshold, it indicates that the trajectory of the vehicle can branch at the second time. The second time can be determined as the branching time.

[0073] In some optional embodiments, since the quantity of the first dynamic objects can be one or more, the branching time for any first dynamic object can be determined. Therefore, the quantity of the branching times can be one or more. As shown in FIG. 4B, the quantity of the branching times is two. Figure 3

[0074] At step 2440, an initial trajectory tree is determined based on the branching times.​

[0075] The bifurcation time determines the number of branches in the trajectory tree. Based on the bifurcation time, the nodes of various trajectories in the trajectory tree at the second time step can be determined, such as... Figure 3 The nodes of the three trajectories shown are then identified. The initial state variables and initial control variables corresponding to each node are then determined. Based on the nodes and their corresponding initial state variables and initial control variables, the initial trajectory tree is obtained.

[0076] In some optional embodiments, the initial control quantity corresponding to a node can be determined based on the vehicle's control quantity at the first moment, or based on the vehicle's control quantity at the first moment, the bifurcation time, and the types of different branches after the bifurcation. Branch types can include yielding (i.e., lane keeping), overtaking, etc., and different branch types characterize the possible driving directions of the vehicle. See also Figure 3 As shown, based on the vehicle's first vehicle state information at the first moment, the vehicle's state quantity and control quantity at the first moment can be determined, that is, the state quantity and control quantity corresponding to node P(0). For nodes P(0) to P(n), the initial control quantity can be the control quantity corresponding to P(0). After the bifurcation point P(n), according to the branch type of branch 1 and branch 2, the initial control quantity of P(n) is adjusted according to the control quantity adjustment amount corresponding to the branch type to obtain the initial control quantities corresponding to branch 1 and branch 2 respectively. For example, if P(0) to P(n) is going straight, branch 1 is overtaking, and branch 2 is slowing down and yielding, then according to the branch type of branch 1, the initial control quantity of P(n) is adjusted to the control quantity of overtaking and used as the initial control quantity of branch 1. According to the branch type of branch 2, the initial control quantity of P(n) is adjusted to the control quantity of slowing down and yielding and used as the initial control quantity of branch 2. The specific adjustment amount of the branch can be preset. Optionally, the adjustment amount can be determined according to the relative state of the dynamic object and the vehicle. After reaching the bifurcation point P(n+3)2, the initial control value of P(n+3)2 is adjusted according to the branch types of branches 21 and 22 to obtain the initial control values ​​corresponding to the nodes of branches 21 and 22, respectively. This is only an exemplary method for determining the initial control value of a node, and practical applications are not limited to this example.

[0077] In some optional embodiments, the initial state quantity corresponding to the node can be determined based on the state quantity of the vehicle at the first time, the initial control quantity of the node at the second time, and a kinematic model of the vehicle. Based on the kinematic model of the vehicle, a state transition function between adjacent two times can be determined. Based on the state quantity and the control quantity of the node P(0) corresponding to the first time, the initial state quantity corresponding to the node P(1) is obtained through the state transition function. Then, based on the initial state quantity and the initial control quantity of P(1), the initial state quantity of the node P(2) is obtained through the state transition function. In this way, the initial state quantity of the node of the branch 1 is obtained after the initial state quantity corresponding to P(n) is obtained. Based on the adjusted initial control quantity of the branch 1 and the initial state quantity of P(n), the initial state quantity of P(n+1)1 is determined. In this way, the initial state quantity of the node of the branch 1 is obtained. Based on the adjusted initial control quantity of the branch 2 and the initial state quantity of P(n), the initial state quantity of the node of the branch 2 is determined. In this way, the initial state quantity of the node of the branch 21 and the branch 22 is obtained. Here, only an exemplary way of determining the initial state quantity is shown, and the actual application is not limited to this exemplary way.

[0078] In the embodiments of the present disclosure, by determining the first dynamic object satisfying the preset condition from the dynamic objects and determining the first predicted trajectory of the first dynamic object, the effectiveness of the first dynamic object can be improved, and the trajectory branch of the ego vehicle is avoided to be constructed based on unnecessary dynamic objects and unnecessary predicted trajectories of the dynamic objects, thereby improving the effectiveness of the initial trajectory tree.

[0079] In some optional embodiments, the step 2420 of determining the first dynamic object satisfying the preset condition and the first predicted trajectory of the first dynamic object based on the probability corresponding to the predicted trajectory comprises:

[0080] determining a first candidate predicted trajectory from the predicted trajectory based on the probability corresponding to the predicted trajectory and a first probability threshold; determining a second candidate predicted trajectory from the first candidate predicted trajectory based on the first candidate predicted trajectory and a second probability threshold, through the preconfigured intelligent driver model; determining a dynamic object with a second candidate predicted trajectory quantity greater than a first preset quantity as the first dynamic object from the dynamic objects based on the second candidate predicted trajectory; and determining the first predicted trajectory of the first dynamic object based on the second candidate predicted trajectory of the first dynamic object.

[0081] The first probability threshold and the second probability threshold are preset probability values, which can be set according to experience, and the second probability threshold is greater than the first probability threshold. The first preset quantity can be set to 1 or other values.

[0082] In some optional embodiments, for any dynamic object, the probability corresponding to any predicted trajectory of the dynamic object can be compared with a first probability threshold, and if the probability corresponding to the predicted trajectory is greater than the first probability threshold, the predicted trajectory is taken as a first candidate predicted trajectory of the dynamic object. If the probability corresponding to the predicted trajectory is less than or equal to the first probability threshold, the predicted trajectory is filtered out or deleted or set as an invalid trajectory and does not participate in subsequent construction of a trajectory tree.

[0083] In some optional embodiments, for any dynamic object, after determining the first candidate predicted trajectory corresponding to the dynamic object, for any first candidate predicted trajectory, the probability corresponding to the first candidate predicted trajectory can be compared with a second probability threshold, and if the probability corresponding to the first candidate predicted trajectory is greater than the second probability threshold, the first candidate predicted trajectory can be taken as a second candidate predicted trajectory. If the probability corresponding to the first candidate predicted trajectory is less than or equal to the second probability threshold, it indicates that the probability of the first candidate predicted trajectory is low, and the first candidate predicted trajectory can be further determined whether it can be taken as a second candidate predicted trajectory by using a preconfigured intelligent driver model (IDM). The intelligent driver model is used to deduce the rationality of the predicted trajectory of the dynamic object, and according to the deducing result of the IDM, it is determined whether the first candidate predicted trajectory can be taken as a second candidate predicted trajectory, so as to filter out unreasonable first candidate predicted trajectories with low probability, for example, by using the IDM to determine whether the acceleration of the first candidate predicted trajectory is reasonable, and if the acceleration exceeds the physical limit of the dynamic object, it is determined that the predicted trajectory is unreasonable. The reasonable first candidate predicted trajectory is taken as a second candidate predicted trajectory.

[0084] In some optional embodiments, for any dynamic object, after determining the second candidate predicted trajectory of the dynamic object, it is further determined whether the number of the second candidate predicted trajectory of the dynamic object is greater than a first preset number based on the number of the second candidate predicted trajectory of the dynamic object, and if the number is greater than the first preset number, it indicates that the dynamic object has multiple valid predicted trajectories, the dynamic object is determined as a first dynamic object, and the second candidate predicted trajectory of the dynamic object is determined as a first predicted trajectory of the first dynamic object.

[0085] In some optional embodiments, the number of predicted trajectories of a dynamic object can be determined first, and then the first dynamic object and the first predicted trajectory of the first dynamic object can be determined from the multi-modal dynamic object based on the probability corresponding to the predicted trajectory of the multi-modal dynamic object, and the order of the specific screening is not limited.

[0086] In the embodiments of the present disclosure, the prediction trajectories with too low probabilities are filtered out by the first probability threshold, and then the prediction trajectories with lower probabilities are verified for validity (i.e., rationality) by deduction based on the IDM model, so as to further filter out invalid prediction trajectories and improve the validity of the prediction trajectories. In addition, based on the number of valid prediction trajectories of the dynamic object, the key multi-modal dynamic object is determined as the first dynamic object for constructing the initial trajectory tree, so as to effectively improve the validity of the initial trajectory tree.

[0087] Figure 5 is a flowchart of a vehicle control method provided by another exemplary embodiment of the present disclosure.

[0088] In some optional embodiments, on the basis of any of the above embodiments, as shown in Figure 5 determining, based on the first object prediction trajectory information and the second object state information, the constraint information, includes:

[0089] In step 2510, dynamic boundary constraint information of the vehicle is determined based on the first object prediction trajectory information.

[0090] The dynamic boundary constraint information is constraint information of the prediction trajectory of the dynamic object on the boundary of the drivable area of the vehicle. The dynamic boundary constraint information can include at least one boundary constraint information corresponding to each second time point.

[0091] In some optional embodiments, at least one prediction trajectory of the dynamic object at the second time point can be determined based on the first object prediction trajectory information, and each prediction trajectory includes a state of the dynamic object at at least one second time point, which includes the position, size, speed, yaw angle, etc. of the dynamic object. According to the position, yaw angle and size of the dynamic object, the area or contour occupied by the dynamic object at the second time point can be determined, and according to the area or contour of the dynamic object at any second time point, the boundary information of the drivable area of the vehicle at the second time point can be determined as the dynamic boundary constraint information of the vehicle.

[0092] In some optional embodiments, the dynamic boundary constraint information of the vehicle can be determined based on the first object state information at the first time point and the first object prediction trajectory information at the second time point, so that the dynamic boundary constraint information includes the boundary constraint information at the first time point.

[0093] In step 2520, static boundary constraint information of the vehicle is determined based on the second object state information.

[0094] The second object state information represents a position, a size, an orientation, etc. of the static object. Based on the second object state information, a region or a contour occupied by the static object can be determined. The contour includes a polygonal contour or a curve. For example, for a static obstacle with a three-dimensional shape, the contour is represented as a polygonal contour. For a lane line, the contour is represented as a curve. The region or the contour occupied by the static object constrains a drivable region of the vehicle. Therefore, static boundary constraint information of the vehicle is determined based on the region or the contour occupied by the static object.

[0095] In step 2530, constraint information is determined based on the dynamic boundary constraint information and the static boundary constraint information.

[0096] The dynamic boundary constraint information represents a constraint on a behavior of the vehicle caused by a trajectory of the dynamic object. The static boundary constraint information represents a constraint on the behavior of the vehicle caused by the static object. The dynamic object and the static object represent various objects around the vehicle that affect the behavior of the vehicle. Therefore, the constraint information can be determined based on the dynamic boundary constraint information and the static boundary constraint information, and is used to constrain a trajectory of the vehicle at the second time.

[0097] In embodiments of the present disclosure, the dynamic boundary constraint information of the vehicle is determined based on the predicted trajectory of the dynamic object, and the static boundary constraint information of the vehicle is determined based on the state information of the static object. Effective constraint references are provided for updating of the trajectory tree, and the effectiveness and safety of trajectories in the target trajectory tree obtained by updating are ensured.

[0098] In some optional embodiments, the determination of the static boundary constraint information of the vehicle based on the second object state information in step 2520 includes:

[0099] First contour information of the static object is determined based on the second object state information. The static boundary constraint information is determined based on the first contour information.

[0100] The first contour information of the static object is information used to describe a shape of the static object in space. For example, contour information in a local coordinate system of the vehicle at the first time. The second object state information includes a position, a size, an orientation, etc. of the static object. Based on the position, the size, and the orientation, etc. of the static object, the first contour information of the static object can be determined. Different types of static objects can be respectively provided with corresponding contour information representation manners. For example, for a static object occupying a certain area, such as a traffic cone, a static parked vehicle, etc., the first contour information can be represented as a polygon in a local coordinate system, for example, as an ordered point set on the polygon. For a linear static object, such as a lane line, a road edge, etc., the first contour information can be represented as a curve describing the static object. Alternatively, the first contour information can be represented as a curve coefficient or an ordered point set on the curve. The specific representation manner of the first contour information is not limited.

[0101] In some optional embodiments, the static object can include static objects in different directions of the vehicle, such as static objects on the left side of the vehicle, static objects on the right side of the vehicle, and the like, and the static boundary constraint information of the vehicle can be determined according to the relative positional relationship between the static object and the vehicle. For example, for the static object on the left side of the vehicle, the left static boundary constraint information of the vehicle can be determined, and for the static object on the right side of the vehicle, the right static boundary constraint information can be determined.

[0102] In some optional embodiments, the static boundary constraint information can include static hard edge constraint information and static risk edge constraint information. The hard edge refers to an impassable driving area boundary that is physically or clearly defined by rules, such as a road edge, a separation belt, a static parked vehicle, a lane line, and the like. The risk edge refers to a potential dangerous area boundary that is dynamically generated based on real-time environment perception and risk assessment, which can include a virtual line, a double yellow line, a guide line, and a boundary determined based on a hard edge and a safety distance threshold, and the like, without limitation.

[0103] In the embodiments of the present disclosure, the static boundary constraint information of the vehicle is determined by the contour information of the static object, so that the planned trajectory of the vehicle satisfies the related constraints of the static object, and collision between the vehicle and the static object is avoided.

[0104] In some optional embodiments, based on the first contour information, the static boundary constraint information is determined, including:

[0105] Based on the first contour information, distance gradient field information corresponding to the static boundary is determined; and based on the distance gradient field information, the static boundary constraint information is determined.

[0106] The distance gradient field (Euclidean Signed Distance Field, abbreviated as ESDF, also known as Euclidean Signed Distance Field) information is a data structure used to describe the spatial distance information of a spatial point (or voxel) and an obstacle (corresponding to the static boundary or static object in the present disclosure) in the environment. Its principle is to assign a signed distance value and gradient information to any point in space. The distance value represents the Euclidean distance from the point to the nearest obstacle, and the sign distinguishes whether the point is inside or outside the obstacle. The sign includes positive, zero, and negative. The distance value with a positive sign is referred to as positive distance, the distance value with a zero sign is referred to as zero distance, and the distance with a negative sign is referred to as negative distance. The positive distance indicates that the point is in free space, i.e., outside the obstacle, the zero distance indicates that the point is on the surface of the obstacle, and the negative distance indicates that the point is inside the obstacle. The gradient information points to the direction away from the obstacle, and is used for obstacle avoidance thrust calculation in trajectory tree optimization.

[0107] In some optional embodiments, distance gradient field information of a static boundary corresponding to any static object can be established based on first contour information of the static object. Static boundary constraint information can be determined based on distance gradient field information of one or more static objects in the surrounding environment of the vehicle. Alternatively, a static boundary line on the left and / or right side of the vehicle can be determined based on first contour information of one or more static objects in the surrounding environment of the vehicle, and distance gradient field information corresponding to the static boundary line can be established. The static boundary line can include a static hard boundary line and / or a static risk boundary line. The static boundary constraint information can be represented as where left represents the left side of the vehicle, right represents the right side of the vehicle, hard edge represents a hard boundary (or safety boundary), risk edge represents a risk boundary, and represent the left hard boundary and the right hard boundary, respectively, and represent the left risk boundary and the right risk boundary, respectively, and any edge can include a point list {(x0, y0), (x1, y1), …, (x m , y m )} composed of a plurality of discrete points, and the points in the point list have corresponding distance values and gradient information in the distance gradient field. In actual applications, the static boundary constraint information can include one or more of the left hard boundary, the right hard boundary, the left risk boundary, and the right risk boundary. That is, any edge in the above formula can include an effective point list or be an invalid boundary, which is determined according to the actual environment.

[0108] In embodiments of the present disclosure, by establishing distance gradient field information corresponding to the static boundary, the relative position relationship between any point in space and the obstacle can be effectively represented, and the static boundary constraint information is determined based on the distance gradient field information, so that in the trajectory tree updating process, the distance and gradient of the position of the node in the trajectory tree to the obstacle can be determined, which is used as a cost sub-function in the objective function to guide the obstacle avoidance thrust calculation of the node, and the position of the node is ensured to be in the free space, so as to avoid the position of the node falling on the surface of the obstacle or inside the obstacle, thereby ensuring the effectiveness and safety of the trajectory.

[0109] In some optional embodiments, distance gradient field information corresponding to the static boundary can be determined based on the first contour information, including:

[0110] first boundary information is determined based on the first contour information; second boundary information is determined based on the first boundary information and a safety distance threshold corresponding to the static object; and the distance gradient field information is determined based on the first boundary information and the second boundary information.

[0111] The first boundary information is information of a safety boundary determined based on the first contour information, and the safety boundary can refer to the hard boundary described above. The second boundary information is information of a risk boundary obtained by increasing or decreasing a safety distance threshold based on the first boundary information, and the risk boundary can refer to the content described above.

[0112] In some optional embodiments, for any side of the vehicle, if there is a physically delimited insurmountable static object such as a curb, a static stopped vehicle, or the like, the distance between the second boundary information and the static object is greater than the distance between the first boundary information and the static object, that is, the lateral distance between the second boundary information and the vehicle is less than the lateral distance between the first boundary information and the vehicle, that is, the safety distance threshold relative to the static object is increased based on the first boundary information to obtain the second boundary information. If there is no physically delimited insurmountable static object, the lateral distance between the second boundary information and the vehicle can be greater than, equal to, or less than the lateral distance between the first boundary information and the vehicle.

[0113] In some optional embodiments, the second boundary information can be determined based on the first boundary information and different safety distance thresholds corresponding to different static objects. Then, the distance gradient field information corresponding to the first boundary information and the distance gradient field information corresponding to the second boundary information are determined based on the first boundary information and the second boundary information.

[0114] In some optional examples, Figure 6 is a schematic diagram of a dynamic-static boundary provided by an example embodiment of the present disclosure. As shown in Figure 6 , the right side of the vehicle 11 includes a plurality of static objects such as a curb A, a roadblock B, and a roadblock C in the figure. According to the relative positional relationship between the plurality of static objects and the vehicle 11, the first boundary information is determined, which can be visualized as the first boundary E1 in the figure. The second boundary information determined based on the first boundary information and the safety distance threshold corresponding to the static object can be visualized as the second boundary E2 in the figure. In the road section on any side of the vehicle where there is a physically delimited insurmountable static object (such as the roadblock B and the roadblock C), the distance between the second boundary E2 and the roadblock is greater than the distance between the first boundary E1 and the roadblock, ensuring the safety distance between the vehicle and the roadblock, as shown in Figure 6 , in the left half of E1 and E2, E2 is away from the roadblock B and the roadblock C relative to E1. In the road section on any side of the vehicle where there is no physically delimited insurmountable static object, since there is no roadblock to affect the vehicle driving, the lateral distance between the second boundary E2 and the vehicle can be greater than the lateral distance of the first boundary E1, as shown in Figure 6 , in the right half of E1 and E2, there is no roadblock, and E2 is on the outside of E1. Here, it is only an example boundary, and in actual applications, it is not limited to Figure 6 the boundary shown in the figure, which can be set according to the safety requirements of vehicle driving.

[0115] In some optional embodiments, the safety distance threshold values corresponding to different types of static objects can be the same or different.

[0116] In some optional embodiments, the safety distance threshold value of any type of static object can be a preset value, and the safety distance threshold value corresponding to a static object can be set according to the characteristics of the different types of static objects. For example, for static objects such as road edges and road barriers that can be collided by vehicles crossing, the safety distance threshold value can be larger, and for static objects such as virtual and real lines on the ground, the safety distance threshold value can be smaller, which is not limited in particular.

[0117] In some optional embodiments, the safety distance threshold value of a static object can be determined according to the safety coefficient of the static object, the speed of the vehicle, and the maximum risk threshold value. For example, the safety distance threshold value can be represented as follows:

[0118]

[0119] wherein k0 represents the safety coefficient, which can be a preset value. v represents the speed of the vehicle. buffer max represents the maximum risk threshold value, which can be a preset value, and min() represents the minimum value function. Since the faster the vehicle speed, the higher the risk of contact with the hard boundary, and the slower the vehicle speed, the lower the risk, the safety distance threshold value is determined in combination with the speed of the vehicle and the maximum risk threshold value, so that a smaller safety distance threshold value can be used when the vehicle speed is slower, and a larger safety distance threshold value can be used when the vehicle speed is faster, and the maximum risk threshold value is used to constrain the upper limit of the safety distance threshold value, to ensure the effectiveness of the safety distance threshold value.

[0120] In the embodiments of the present disclosure, the hard boundary is determined based on the contour information of the static object, and then the risk boundary is determined based on the hard boundary and the safety distance threshold value corresponding to the static object. The distance gradient field information is determined based on the two types of boundary information, which is used to determine the static boundary constraint information. In the process of updating the trajectory tree, the node state quantity is constrained to meet the static boundary constraint information, to avoid the vehicle from colliding or contacting with the static object, and to improve the safety of the trajectory.

[0121] In some optional embodiments, the step 2510 of determining the dynamic boundary constraint information of the vehicle based on the first object predicted trajectory information comprises:

[0122] determining a second predicted trajectory of a second dynamic object satisfying a preset object condition at a second time based on the first object predicted trajectory information; determining the lateral constraint information and / or the longitudinal constraint information corresponding to the vehicle at the second time based on the second predicted trajectory; and determining the dynamic boundary constraint information based on the lateral constraint information and / or the longitudinal constraint information.

[0123] The preset object condition is used to determine a dynamic object (which can be referred to as a potential interference obstacle or a potential interference object) interfering or overlapping with the vehicle trajectory and / or a CIPV (Closest In-Path Vehicle) object, and the CIPV object refers to a dynamic object closest to the ego vehicle in the driving path of the ego vehicle, such as Figure 6 the dynamic object D2 in front of the vehicle 11. The potential interference obstacle is such as the dynamic object D1 on the left lane of the vehicle 11. The lateral constraint information is constraint information for the lateral movement of the vehicle, and the longitudinal constraint information is constraint information for the longitudinal movement of the vehicle. Figure 6

[0124] In some optional embodiments, the vehicle trajectory information of the vehicle at the second time can be predicted based on the first vehicle state information by a preconfigured trajectory prediction algorithm, the interference between the vehicle trajectory and the predicted trajectory of the dynamic object can be determined based on the predicted vehicle trajectory information and the first object predicted trajectory information, and the second dynamic object interfering with the vehicle trajectory and the second predicted trajectory of the second dynamic object at the second time can be determined from the dynamic object based on the interference.

[0125] In some optional embodiments, the direction and distance of the dynamic object relative to the vehicle can be determined based on the first object state information of the dynamic object, the closest CIPV object on the driving path in front of the vehicle can be determined based on the direction and distance, and the second dynamic object can be determined as the CIPV object, and the predicted trajectory of the second dynamic object can be determined as the second predicted trajectory.

[0126] In some optional embodiments, the number of the second dynamic objects can be one or more. The second predicted trajectory of any second dynamic object can be the most probable predicted trajectory of the second dynamic object.

[0127] In some optional embodiments, after the second predicted trajectory of the second dynamic object at the second time is determined, the corresponding lateral constraint information and / or longitudinal constraint information of the vehicle at the second time can be determined according to the relative relationship between the second predicted trajectory of the second dynamic object and the vehicle. For the second dynamic object on the left lane or the right lane of the vehicle, if the predicted trajectory of the second dynamic object interferes with the vehicle trajectory, it indicates that the second dynamic object will cut into the current lane where the vehicle is located, and the second predicted trajectory of the second dynamic object has certain constraint on the lateral movement of the vehicle. Therefore, the corresponding lateral constraint information of the vehicle at the second time can be determined based on the second predicted trajectory of the second dynamic object. For the CIPV object in front of the vehicle, it has certain constraint on the longitudinal movement of the vehicle, and therefore the longitudinal constraint information of the vehicle can be determined based on the second predicted trajectory of the second dynamic object of the CIPV in front of the vehicle. The lateral constraint information at any second time can include a dynamic constraint point column corresponding to the second time, and can be represented as ​Where i represents the i-th second time moment, i = 0, 1, ..., N, and N represents the number of second time moments, with the 0th second time moment being the first time moment. These represent the lateral constraint information on the left side of the vehicle and the lateral constraint information on the right side of the vehicle, respectively.

[0128] In some optional examples, such as Figure 6 As shown, the second dynamic object includes dynamic object D1 and dynamic object D2. Dynamic object D1 imposes certain constraints on the lateral movement of the vehicle. Therefore, the lateral constraint information of the vehicle can be determined based on the second predicted trajectory L5 of D1. Since D1 is in the left lane of the vehicle, its lateral constraint information corresponds to the aforementioned left-side lateral constraint information. Refer to the dashed boundary E3 along L5 in the diagram. Dynamic object D2 is in front of vehicle 11. The distance, velocity, and acceleration of dynamic object D2 relative to vehicle 11 at any second moment affect the longitudinal motion of vehicle 11. Based on the second predicted trajectory of dynamic object D2, longitudinal constraint information can be determined. In a real-world scenario, the second dynamic object can include one or both of potential interference objects and CIPV objects, or it may not exist. If no second dynamic object meets the preset object conditions, the determination of lateral and longitudinal constraint information can be avoided. If a second dynamic object exists, one or both of lateral and longitudinal constraint information can be determined based on the type of the second dynamic object, and then dynamic boundary constraint information can be determined based on the lateral and / or longitudinal constraint information.

[0129] In the embodiments of this disclosure, by determining the potential interference object and / or CIPV object of the vehicle and the corresponding second predicted trajectory, the lateral constraint information and / or longitudinal constraint information of the vehicle at the second moment are determined. The constraint information is used to update the trajectory tree, so that the state variables of the nodes in the trajectory tree satisfy the constraint conditions of the dynamic object on the lateral and longitudinal motion of the vehicle, effectively avoiding collisions between the vehicle and the dynamic object and ensuring the safety of vehicle driving.

[0130] In some optional embodiments, based on the second predicted trajectory, determining the lateral constraint information and / or longitudinal constraint information of the vehicle at the second time moment includes:

[0131] determining second contour information of the second dynamic object at the second time based on the second predicted trajectory; determining a boundary point set corresponding to the second time based on the second contour information; the boundary point set being a point set of the vehicle in a local coordinate system at the first time; determining lateral constraint information based on the boundary point set; and / or, determining a longitudinal travel distance from the vehicle to the second dynamic object based on the second contour information; determining a first longitudinal safety boundary of the vehicle at the second time based on the longitudinal travel distance; determining a second longitudinal safety boundary based on the first longitudinal safety boundary and a preset distance threshold; and determining the longitudinal constraint information based on the first longitudinal safety boundary and the second longitudinal safety boundary.

[0132] wherein the second contour information of the second dynamic object at the second time represents an area occupied by the second dynamic object at the second time, such as Figure 6 wherein a contour of D1 at any second time moves along L5 to obtain the second contour information of D1 at the second time. Alternatively, the second contour information of D1 at the second time is determined based on the position, size and heading angle of D1 at the second time. The boundary point set corresponding to the second time can be determined according to the relative position relationship between the second dynamic object and the vehicle, for example Figure 6 wherein E3, the boundary point set corresponding to the second time, is determined based on the second contour information of D1 at the second time and the preset distance threshold. The local coordinate system of the vehicle at the first time is a self-vehicle coordinate system with the center of the rear axle of the vehicle at the first time as the origin, such as Figure 6 xoy in FIG. 8. After determining the boundary point set, the lateral constraint information of the vehicle can be obtained based on the boundary point set. The longitudinal travel distance from the vehicle to the second dynamic object refers to the arc length of the travel trajectory of the vehicle to the point of contact with the second dynamic object, or the trajectory distance. The longitudinal travel distance includes at least one longitudinal travel distance corresponding to the second time, which can be represented as S i . Alternatively, the longitudinal travel distance can be determined as the first longitudinal safety boundary, or the first longitudinal safety boundary can be determined based on the longitudinal travel distance in combination with the longitudinal safety distance threshold, such as Figure 6 wherein the difference between the longitudinal travel distance from the vehicle 11 to D2 and the longitudinal safety distance threshold is determined as the first longitudinal safety boundary of the vehicle, i.e., the longitudinal travel distance from the vehicle to D2 at any second time needs to be greater than the longitudinal safety distance threshold, otherwise there is a risk of collision with D2.

[0133] In some optional embodiments, the second longitudinal safety boundary is a comfort distance boundary that takes into account the driving comfort of users in the vehicle. It can be further enhanced by adding a preset distance threshold based on the first longitudinal safety boundary to constrain the longitudinal travel distance from the vehicle to the second dynamic object to have a larger trajectory distance. After determining the second longitudinal safety boundary, the longitudinal constraint information of the vehicle can be determined based on the first and second longitudinal safety boundaries. The longitudinal constraint information may include the first and second longitudinal safety boundaries. Optionally, based on the first and second longitudinal safety boundaries, one or more boundaries of the vehicle's speed, acceleration, jerk, etc., can be further determined, specifically according to actual needs.

[0134] In some optional embodiments, the longitudinal constraint information can be represented as constraint information in the Frenet coordinate system, specifically as an ST plot in the Frenet coordinate system, i.e., a plot showing the longitudinal safety boundary changing over time. See [link to relevant documentation]. Figure 6 As shown, based on the second predicted trajectory of D2, D2 occupies a quadrilateral region in the ST diagram. Taking D2 as a vehicle, the upper edge of the quadrilateral region corresponds to the front edge of D2 at different times, and the lower edge corresponds to the rear edge of D2 at different times. The first and second longitudinal safety boundaries are E4 and E5 in the figure. The Frenet coordinate system is a dynamic coordinate system with a reference line as the baseline. The vertical axis is the S-axis, along the tangent direction of the reference line, representing the cumulative arc length of the vehicle traveling along the road, i.e., the distance from the starting point. The horizontal axis is along the normal direction of the reference line, representing the lateral distance of the vehicle deviating from the reference line. The specific Frenet coordinate system is not described in detail.

[0135] In some optional embodiments, distance gradient field information can be established based on the boundary point set at the second time step, as part of the dynamic boundary constraint information, to determine the distance and gradient information of any point in space relative to the second dynamic object corresponding to the boundary point set, and can participate in the construction of the cost sub-function in the objective function. Alternatively, distance gradient field information about the second dynamic object can be established based on the second contour information of the second dynamic object, and participate in the construction of the relevant cost sub-function.

[0136] In some alternative embodiments, distance gradient field information about the second dynamic object can be established based on the first and second longitudinal safety boundaries, and used as part of the dynamic boundary constraint information to construct the relevant cost sub-function. Alternatively, distance gradient field information about the second dynamic object can be established based on the second contour information of the second dynamic object to participate in the construction of the relevant cost sub-function.

[0137] In the embodiments of the present disclosure, since the second dynamic object includes a dynamic object interfering with the vehicle trajectory and / or a nearest dynamic object in front of the vehicle, the lateral movement and longitudinal movement of the vehicle are constrained to a certain extent, and therefore, the lateral constraint information and / or the longitudinal constraint information of the vehicle are determined based on the second profile information of the second dynamic object at the second time, so that the state quantity of the node in the trajectory tree satisfies the lateral and longitudinal constraints brought by the dynamic object, and the collision between the vehicle and the dynamic object is avoided, and the effectiveness and safety of the trajectory in the trajectory tree are improved.

[0138] In some optional embodiments, the longitudinal constraint information is determined based on the first longitudinal safety boundary and the second longitudinal safety boundary, including:

[0139] At least one of the upper limit value of the speed, the upper limit value of the acceleration, the lower limit value of the acceleration, the upper limit value of the acceleration change rate and the lower limit value of the acceleration change rate of the vehicle is determined based on the first longitudinal safety boundary and the second longitudinal safety boundary, and the longitudinal constraint information is determined based on the first longitudinal safety boundary and the second longitudinal safety boundary and at least one of the upper limit value of the speed, the upper limit value of the acceleration, the lower limit value of the acceleration, the upper limit value of the acceleration change rate and the lower limit value of the acceleration change rate.

[0140] The upper limit value of the speed represents the maximum speed at which the vehicle can travel under the constraint of the first longitudinal safety boundary and the second longitudinal safety boundary, the upper limit value of the acceleration and the lower limit value of the acceleration represent the acceleration boundary value at which the vehicle can travel, and the upper limit value of the acceleration change rate and the lower limit value of the acceleration change rate represent the boundary value of the acceleration change rate at which the vehicle can travel. The boundary values of the speed, the acceleration and the acceleration change rate of the vehicle can be determined based on the first vehicle state information of the vehicle at the first time, the first longitudinal safety boundary and the second longitudinal safety boundary of the second dynamic object at any second time, the speed and the acceleration of the second dynamic object at any second time, and the collision time of the vehicle and the second dynamic object being less than a safety time threshold, the dynamics limit of the vehicle, etc. For example, the first longitudinal safety boundary and the second longitudinal safety boundary at any second time represent the range of the longitudinal travel distance of the vehicle at any second time, and based on the range of the longitudinal travel distance at the second time, the speed and the acceleration of the second dynamic object, and the conditions that the collision time of the vehicle and the second dynamic object is less than a safety time threshold and the dynamics limit of the vehicle, at least one of the upper limit value of the speed, the upper limit value of the acceleration, the lower limit value of the acceleration, the upper limit value of the acceleration change rate and the lower limit value of the acceleration change rate of the vehicle can be determined to constrain the vehicle state of the vehicle at any second time to satisfy the conditions that the collision time is less than the safety time threshold and the dynamics limit of the vehicle, etc. The longitudinal constraint information is determined based on the first longitudinal safety boundary and the second longitudinal safety boundary and at least one of the upper limit value of the speed, the upper limit value of the acceleration, the lower limit value of the acceleration, the upper limit value of the acceleration change rate and the lower limit value of the acceleration change rate.

[0141] In the embodiments of the present disclosure, by determining the boundary values of the vehicle with respect to speed, acceleration, and acceleration rate of change, the longitudinal constraint of the vehicle is further improved, thereby further improving the effectiveness and safety of the updated target trajectory tree.

[0142] Figure 7 is a flowchart of a vehicle control method provided by another exemplary embodiment of the present disclosure.

[0143] In some optional embodiments, on the basis of any of the above embodiments, as shown in Figure 7 Based on the constraint information, constructing a target function in step 260 includes:

[0144] In step 2610, the state variable and the control variable corresponding to the node in the initial trajectory tree are determined.

[0145] The initial value of the state variable is the initial state variable corresponding to the node in the initial trajectory tree, and the initial value of the control variable is the initial control variable corresponding to the node.

[0146] The target function is a function of the state variable and the control variable of the node, and the state variable and the control variable need to be updated constantly in the trajectory tree updating process. Therefore, the state variable and the control variable can be used as variables to construct the target function. Any node has a corresponding state variable and control variable. The state variable starts from the initial state variable, and the control variable starts from the initial control variable. In the trajectory tree updating and optimization process, the values of the state variable and the control variable are constantly updated to minimize or converge the value of the target function. For example, the state variable and the control variable are represented as X = [x, y, θ, δ, ω, s, v, a] and u = [α, jerk], respectively. The meanings of the symbols are described in the foregoing embodiments. The nodes at different times have respective variable values.

[0147] In step 2620, based on the constraint information, the state variable, and the control variable, at least one of the following cost sub-functions with respect to the state variable and the control variable is determined: centripetal acceleration cost sub-function, centripetal acceleration rate of change cost sub-function, dynamic-static boundary collision time cost sub-function, dynamic lateral constraint cost sub-function, and dynamic longitudinal constraint cost sub-function.

[0148] wherein the centripetal acceleration cost sub-function is a cost sub-function with respect to the centripetal acceleration corresponding to the node. The centripetal acceleration is an acceleration component of the vehicle along the curve, which is directed to the center of curvature of the trajectory, and reflects the speed direction change speed. The centripetal acceleration rate of change cost sub-function is a cost sub-function with respect to the centripetal acceleration rate of change corresponding to the node. The centripetal acceleration rate of change characterizes the speed of the centripetal acceleration change. The dynamic-static boundary collision time cost sub-function is a cost sub-function of the collision time of the node under the dynamic-static boundary constraint. The dynamic lateral constraint cost sub-function is a cost sub-function of the node violating the dynamic lateral constraint. The dynamic longitudinal constraint cost sub-function is a cost sub-function of the node violating the longitudinal constraint.

[0149] In some optional embodiments, for any node, the centripetal acceleration corresponding to the node can be determined based on the wheel steering angle δ corresponding to the node, the speed v of the vehicle, and the wheelbase L of the vehicle, and is represented as follows:

[0150]

[0151] wherein, represents the centripetal acceleration function, X i and u i respectively represent the state variable and the control variable of the node corresponding to the i-th second time in any trajectory in the trajectory tree, i = 0, 1, …, N, N represents the number of second times, the 0-th second time is the first time, v i represents the component of the state variable representing the vehicle speed, δ i represents the component of the state variable representing the wheel steering angle, and the wheelbase L of the vehicle represents the distance between the front axle and the rear axle of the vehicle. In practical applications, different symbols can be used to represent the state variable and the control variable for the nodes on different trajectories and at different times in the trajectory tree, such as X ji (or X ji ) and u ji (or u ji ) are respectively used to represent the state variable and the control variable of the node at the i-th second time in the j-th trajectory.

[0152] In some optional embodiments, the centripetal acceleration cost sub-function can be determined based on the centripetal acceleration and the centripetal acceleration weight. For example, based on the centripetal acceleration in the above formula (2), the centripetal acceleration cost sub-function corresponding to any node can be represented as:

[0153]

[0154] wherein, represents the centripetal acceleration cost sub-function, represents the centripetal acceleration weight, which can be a preset value or dynamically determined according to actual conditions.

[0155] In some optional embodiments, the centripetal acceleration rate of change cost sub-function can be determined based on the centripetal acceleration function and the centripetal acceleration rate of change weight, and can be expressed as follows:

[0156]

[0157] in, The cost subfunction represents the rate of change of centripetal acceleration. Represents the centripetal acceleration function The first derivative, This represents the weight of the rate of change of centripetal acceleration, which can be a preset value.

[0158] In some optional embodiments, a distance query function and a gradient calculation function can be determined based on the distance gradient field information in the static boundary constraint information and the dynamic boundary constraint information. The distance query function is used to query the perpendicular distance from the node to the static and dynamic boundaries, and the gradient calculation function is used to calculate the gradient information of the node. Then, based on the perpendicular distance function and the gradient calculation function, the dynamic and static boundary collision time cost sub-function is determined, as follows:

[0159] f bd (X i ,u i =DistanceQuery(X) i ,u i ,bd line ) Formula (5)

[0160]

[0161] In formulas (5) to (7), DistanceQuery(,,) represents the distance query function, bd line Represents a dynamic or static boundary, including at least one of static and dynamic boundaries. DistanceQuery(X) i ,u i ,bd line ) indicates a query for [X i ,u i To the dynamic and static boundary bd line The distance value, i.e., the distance value under the distance gradient field, is represented by the query result as f. bd (X i ,u i GradQuery(,,) is the gradient calculation function. i ,u i ,bd line ) indicates the calculation of [X i ,u iThe gradient information of ] is calculated and expressed as follows: This represents the time cost sub-function for collisions between dynamic and static boundaries. This represents the cost weight, which can be a preset value or dynamically determined based on the actual scenario. (kTTC and kSafe) dis These represent the lateral safety time domain (i.e., lateral collision time) and the minimum lateral safety distance, respectively. proj The vertical velocity pointing towards the boundary can be expressed as follows:

[0162]

[0163] in, and They represent The two components in the gradient information are the longitudinal gradient and the transverse gradient, where θ represents X. i The heading angle in the diagram, v represents the X-axis. i The speed in.

[0164] In some optional embodiments, the violation amount of node state variables and control variables relative to the lateral constraint information can be determined based on the lateral constraint information in the dynamic boundary constraint information. Based on the violation amount and its weight, the dynamic lateral constraint cost sub-function can be determined. It is expressed as follows:

[0165]

[0166] In formula (9), This represents the i-th second time step (i.e., T). i The dynamic lateral constraint cost sub-function at time (M), where M represents the number of dynamic boundary lines on either side of the vehicle in the lateral constraint information. Represents the node's [X] i ,u i The violation amount relative to the k-th dynamic boundary on the left. Represents the node's [X] i ,u i The violation amount relative to the k-th dynamic boundary on the right side, the violation amount refers to [X i ,u i Values ​​exceeding the boundary. For example, taking the left boundary as an example, the vehicle's trajectory point should be to the right of the left boundary. If [X] i ,u i The corresponding trajectory point falls to the left of the left boundary, indicating that [X] i ,u i The value exceeding the left boundary is considered a violation. It should be noted that in practical applications, at any given time, there can be one or both left and right boundaries. If only the left boundary exists, then... may be represented as 0, such as Figure 6 E3 is the left dynamic boundary of a node in the trajectory tree, and the right side is constrained by a static boundary, and there is no right dynamic boundary. and represent the left side violation weight and the right side violation weight, respectively.

[0167] In some optional embodiments, a dynamic longitudinal constraint cost sub-function can be determined based on the longitudinal constraint information of the dynamic boundary constraint information in the constraint information. Specifically, the dynamic longitudinal constraint cost sub-function can be determined based on the violation of the node's [X i ,u i ] relative to the dynamic longitudinal boundary. may be represented as follows:

[0168]

[0169] In formula (10), represents the dynamic lateral constraint cost sub-function at the i-th second time (i.e., T i time), ω ud and ω ld represent the cost weights of the upper and lower boundaries, respectively, are the violation amounts of the node's [X i ,u i ] exceeding the longitudinal driving distance s, speed v, acceleration a, and jerk upper boundary, are the violation amounts of the node's [X i ,u i ] exceeding the longitudinal driving distance s, speed v, acceleration a, and jerk lower boundary.

[0170] In some optional embodiments, the initial trajectory tree includes K trajectories, and any trajectory can include nodes at the first time and N second times. For any node, at least one of the above cost sub-functions corresponding to the node can be determined based on the constraint information.

[0171] Step 2630, constructing a target function based on the cost sub-function.

[0172] wherein any cost sub-function can be used as a term in the target function to obtain the target function.

[0173] In some optional examples, for any trajectory in the trajectory tree, the target sub-function can be represented as follows:

[0174]

[0175] In formula (11), l j (X i ,ui ) represents the trajectory T in the j-th type. i The cost subfunctions corresponding to the nodes at time points N represent the number of second time points, and time point T0 represents the first time point. Other symbols can be found in the preceding content. Optionally, the cost subfunction at time point T0 may not be included in the objective subfunction. The ellipsis “…” in formula (11) indicates that, based on the above cost subfunctions, other cost subfunctions can be extended according to the actual application scenario, and are not limited to the cost subfunctions in the formula.

[0176] The objective function can be determined based on the objective sub-functions corresponding to the K (K is a positive integer) trajectories included in the trajectory tree, and can be expressed as follows:

[0177] In the embodiments of this disclosure, at least one cost sub-function is determined based on constraint information regarding the state variables and control variables of nodes in the trajectory tree. Then, an objective function is constructed based on the cost sub-function, so that the objective function can characterize the overall cost of nodes in the trajectory tree under the constraint information. Using the objective function to optimize the initial trajectory tree can guide the minimization of the overall cost of the nodes in the trajectory tree, ensuring that the target state of the nodes in the trajectory tree can meet the environmental constraints in the driving scenario and improve vehicle driving safety.

[0178] In some optional embodiments, based on any of the above embodiments, step 270, which updates the initial trajectory tree based on the first vehicle state information and the objective function, and determines the target trajectory tree, includes:

[0179] Based on the first vehicle state information, the objective function, and the pre-configured state transition rules, the initial trajectory tree is iteratively updated through an iterative linear quadratic regulator to obtain the target trajectory tree.

[0180] Among them, the state transition rule (or state transition function) is the rule for the vehicle's state transition from one time moment to another. The Iterative Linear Quadratic Regulator (iLQR) is an optimal control algorithm based on iterative optimization. It gradually approaches the global optimum by repeatedly linearizing the nonlinear system and applying a linear quadratic regulator (LQR) to solve for the local optimal control law.

[0181] In some optional embodiments, during any iteration update process, the state variables of the nodes in the trajectory can be generated by forward simulation of the current control variable (or the current control strategy). Based on the objective function, the control variable is optimized by reverse recursion, and then the state variables are updated by forward propagation to realize the iterative update of the trajectory tree until the iteration termination condition is met, and the target trajectory tree is obtained.

[0182] In some alternative embodiments,Figure 8 This is a schematic diagram of a kinematic model provided in an exemplary embodiment of this disclosure. For example... Figure 8 As shown, taking the vehicle's local coordinate system xoy at the first moment as an example, the vehicle at the nth second moment (i.e., T) n The state variable at time ( ) is represented as X n =[x n ,y n ,θ n ,δ n ,ω n ,s n ,v n ,a n ], n = 0, 1, ..., N-1, where N represents the number of second time points, then the (n+1)th second time point (i.e., T) n+1 The state variable X at time ( ) n+1 =[x n+1 ,y n+1 ,θ n+1 ,δ n+1 ,ω n+1 ,s n+1 ,v n+1 ,a n+1 ] can be represented as follows:

[0183] x n+1 =x n +v n *cosθ n *d t

[0184] y n+1 =y n +v n *sinθ n *d t

[0185] θ n+1 =θ n +tanθ n / L*d t

[0186]

[0187] s n+1 =s n +v n *d t

[0188] v n+1 =v n +a n *d t

[0189] a n+1 =an jerk n *d t

[0190] In formula (12), taking a kinematic model with constant jerk as an example, α n and jerk n are the wheel angular acceleration and longitudinal jerk in the control quantity at time T n , d t represents the time interval between time T n and time T n+1 , or the time step, and L represents the wheelbase. Formula (12) shows the state transition function from time T n to time T n+1 . The state quantity and control quantity at time T0 can be determined based on the state information of the vehicle at the first time.

[0191] It should be noted that the kinematic model in the embodiments of the present disclosure is only taken as an example of formula (12), and the kinematic model in actual application is not limited to the above example. For example, the kinematic model can also be a uniform acceleration model or other models.

[0192] In the embodiments of the present disclosure, the node state of the trajectory tree is iteratively updated and optimized by the iterative linear quadratic regulator, the complex nonlinear problem is converted into a local linear quadratic optimization problem for solving, strong nonlinear scenes such as high-speed turning of the vehicle can be processed, the effectiveness and accuracy of the trajectory in the complex traffic scene are improved, and high real-time performance is achieved.

[0193] Figure 9 is a flowchart of a vehicle control method provided by another exemplary embodiment of the present disclosure.

[0194] In some optional embodiments, in step 270, based on the first vehicle state information, the target function, and the preconfigured state transition rule, the initial trajectory tree is iteratively updated by the iterative linear quadratic regulator to obtain a target trajectory tree, including:

[0195] In step 2710, based on the first vehicle state information, an initial state of the vehicle is determined.

[0196] In which, the first time is taken as the initial time of the trajectory, that is, time T0, the state information corresponding to the state quantity is obtained from the first vehicle state information to obtain the initial state of the vehicle. For example, the initial state can include [x0, y0, θ0, δ0, ω0, s0, v0, a0].

[0197] In some optional embodiments, the initial trajectory tree includes the node corresponding to the first moment, and the initial state of the vehicle can be obtained from the initial trajectory tree. That is, the state quantity of the node corresponding to the first moment in the initial trajectory tree is the initial state of the vehicle.

[0198] Step 2720: Based on the initial trajectory tree, determine the current state variable and current control variable corresponding to the node.

[0199] In the initial iteration process, the initial state variables of the nodes in the initial trajectory tree are used as the current state variables, and the initial control variables of the nodes in the initial trajectory tree are used as the current control variables.

[0200] Step 2730: Based on the current state variable, current control variable, and objective function corresponding to the node, update the current control variable to obtain the updated first control variable.

[0201] In some optional embodiments, the current state quantity corresponding to a node in the trajectory tree represents the possible current state of that node. Based on the objective function, the control correction quantity can be solved, and the current control quantity is corrected based on the control correction quantity to obtain the updated first control quantity.

[0202] In some alternative embodiments, the objective function can be determined based on at least one cost sub-function and state cost function described above. The state cost function of any node is a function that penalizes the deviation of the node's state variables, representing the distance from that node to the endpoint (i.e., T). N The remaining cost of a node at time T, the aforementioned cost sub-function is used to determine the instantaneous cost of a node, and can be based on T. N The cost subfunction of the node at time T N The state cost function corresponding to the state variable at time step 1 is used to inversely determine the state-action cost function. Based on the state-action cost function, the objective function is determined. The state cost function corresponding to the state variable at the endpoint can be expressed as V. N (X N )=l f (X N The defined objective function can be expressed as follows:

[0203]

[0204] In formula (13), k is the parent node of j, which can correspond to Figure 3 The node at the first moment, j represents the j-th branch, that is, the j-th trajectory in the trajectory tree, f(X i ,u i ) represents the state transition function, V j (f(X i ,u i )) represents the optimal residual cost, which is the cost at the next time step (T). i+1The minimum cumulative cost from the initial time to the end time. ω j The state-action cost function is represented as:

[0205] Q i =∑ j ω j (l j (X i ,u i )+V j (f(X i ,u i )) Formula (14)

[0206] In Formula (14), Q i represents the state-action cost at time T i , j represents the jth branch, and if there are multiple branches, the state-action cost is obtained by weighted summation of the instantaneous cost and the residual cost of the multiple branches.

[0207] In some optional embodiments, based on the state-action cost function, the control quantity is updated forward from the initial time to the initial time T N , the cost function is linearized and quadratic approximated at each time, the feedback gain and the feedforward gain are determined, the control quantity is updated forward based on the feedforward gain and the feedback gain, and the updated first control quantity corresponding to any node is obtained. The specific operations of determining the feedback gain and the feedforward gain and updating the control quantity can adopt any implementable operation of iLQR, for example, for any time, the Taylor expansion to the second order is performed on Q i , the Hessian matrix and the gradient are determined, the optimal control correction quantity is solved by setting the partial derivative of the control quantity to 0, and then the current control quantity is corrected based on the optimal control correction quantity to obtain the corrected first control quantity, and the specific operations are not repeated here. i

[0208] Step 2740, based on the initial state, the first control quantity and the preconfigured state transition function, the current state quantity is updated to obtain the updated first state quantity.

[0209] Wherein, after obtaining the updated first control quantity, the initial state is forward propagated based on the first control quantity corresponding to the node and the state transition function, and the current state quantity of the node is updated to obtain the updated first state quantity, that is, under the control of the updated first control quantity, the initial state is state transitioned according to the state transition function obtained from the kinematic model to obtain the updated first state quantity corresponding to the node, and the state transition function is shown in Formula (12).

[0210] ​At step 2750, in response to the iteration end condition not being met, the first state quantity is taken as the current state quantity, the first control quantity is taken as the current control quantity, and the step of updating the current control quantity based on the current state quantity and the current control quantity corresponding to the node and the target function and the steps following the step are iteratively performed to obtain an updated first control quantity.

[0211] The iteration end condition can include a state change quantity convergence, a cost decrease rate convergence, or the like. If any of the conditions is met, it is determined that the iteration end condition is met. If the iteration end condition is not met, the first state quantity and the first control quantity are determined as the current state quantity and the current control quantity, respectively, and the iteration update is continued according to the steps 2730 to 2740 until the iteration end condition is met.

[0212] At step 2760, in response to the iteration end condition being met, the first state quantity is taken as the target state quantity corresponding to the node, the first control quantity is taken as the target control quantity corresponding to the node, and a target trajectory tree is obtained based on the target state quantity and the target control quantity corresponding to the node.

[0213] If the iteration end condition is met, the iteration process is ended, and the first state quantity is determined as the target state quantity and the first control quantity is determined as the target control quantity. Through the iteration update, the target state quantity and the target control quantity corresponding to any node in the trajectory tree can be obtained, and the target trajectory tree can be obtained based on the target state quantity and the target control quantity corresponding to the node.

[0214] In some optional embodiments, the target trajectory tree is obtained based on the target state quantity and the target control quantity corresponding to each node in the trajectory tree.

[0215] In the embodiments of the present disclosure, the iLQR algorithm is used to perform multiple back propagation and forward propagation based on the target function, to update and optimize the control quantity and the state quantity until the overall optimal trajectory is obtained, the target trajectory tree is obtained, and the effectiveness and safety of the target trajectory tree are improved.

[0216] In some optional embodiments, Figure 10 is a multi-modal scene schematic diagram provided by an example embodiment of the present disclosure. As shown in Figure 10As shown, the dynamic object D1 has two predicted trajectories (or intention trajectories), namely Cutin and Lanekeep. If the Cutin intention of D1 is not considered, the planning trajectory of vehicle 11 is as shown in G1. Since the Cutin trajectory of D1 is not considered, the vehicle 11 is not constrained in the longitudinal direction, and thus the longitudinal movement of the vehicle 11 is fast. If only the Cutin intention of D1 is considered, the constraint of the vehicle 11 in the longitudinal direction is stronger, so that the planning trajectory of the vehicle is as shown in G2. The longitudinal movement of the vehicle 11 is greatly affected. If the multimodal intention of D1 is considered, the planning trajectory is as shown in G3. The multimodal intention of D1 is comprehensively considered, so that the influence on the longitudinal movement of the vehicle 11 is smaller than that in G2. It is ensured that the vehicle 11 can safely travel regardless of which predicted trajectory D1 travels in the future.

[0217] The vehicle control method provided by the embodiments of the present disclosure determines the spatiotemporal joint constraint, that is, the constraint on the trajectory planning of the vehicle in time and space, and solves the optimal target trajectory tree in combination with the iLQR algorithm framework, thereby effectively improving the reachability, comfort and safety of the trajectory in the target trajectory tree. Specifically, the trajectory planning is realized in the local coordinate system of the vehicle, and the local coordinate system is the Cartesian coordinate system, which can solve the problem of insufficient motion position constraint of the ego vehicle in the frenet coordinate system and improve the safety. In the Cartesian coordinate system, the constraint information can be coupled horizontally and vertically, thereby solving the problem of insufficient reachability of the trajectory in the scene requiring horizontal and vertical coordination caused by the horizontal and vertical decoupling of related technologies, and improving the reachability of the trajectory. Secondly, by using the spatiotemporal joint trajectory tree, the multimodal intention of the key obstacles around the vehicle is comprehensively considered, so as to avoid the situation that the ego vehicle reacts excessively due to the ambiguous intention of the obstacles at a long distance, thereby causing the discontinuity of the decision planning, improving the comfort of the vehicle driving, and improving the inter-frame stability of the ego vehicle trajectory in the scene such as detouring and overtaking cancellation, for example, avoiding the situation that the planning trajectory of the previous frame is a detouring trajectory, and the planning trajectory of the next frame becomes a non-detouring trajectory, thereby causing the situation that the actuator cannot respond smoothly.

[0218] The above-mentioned embodiments of the present disclosure can be implemented alone or in any combination without conflict, and can be set according to actual needs. The present disclosure is not limited.

[0219] Any vehicle control method provided by the embodiments of the present disclosure can be executed by any appropriate electronic device with data processing capability, including but not limited to terminal devices and servers and the like. Alternatively, any vehicle control method provided by the embodiments of the present disclosure can be executed by a processor, such as a processor executing any vehicle control method mentioned in the embodiments of the present disclosure by calling corresponding instructions stored in a memory. The following will not be described again.

[0220] Exemplary Apparatus

[0221] Figure 11 This is a schematic diagram of a vehicle control device provided in an exemplary embodiment of the present disclosure. The device in this embodiment can be used to implement corresponding method embodiments of the present disclosure, such as… Figure 11 The apparatus shown may include:

[0222] The first processing module 51 is used to determine the first vehicle status information at a first moment. The system also includes the first processing module 51, the second processing module 52, the third processing module 53, the fourth processing module 54, the fifth processing module 55, the sixth processing module 56, the seventh processing module 57, and the control module 58.

[0223] The second processing module 52 is used to determine the first object state information of the dynamic object at the first moment and the second object state information of the static object at the first moment.

[0224] The third processing module 53 is used to determine the predicted trajectory information of the dynamic object at the second time based on the state information of the first object, wherein the first time is earlier than the second time.

[0225] The fourth processing module 54 is used to determine the initial trajectory tree of the vehicle at the second time point based on the predicted trajectory information of the first object and the state information of the first vehicle.

[0226] The fifth processing module 55 is used to determine constraint information based on the predicted trajectory information of the first object and the state information of the second object.

[0227] The sixth processing module 56 is used to construct the objective function based on the constraint information.

[0228] The seventh processing module 57 is used to iteratively update the initial trajectory tree based on the first vehicle state information and the objective function, and determine the target trajectory tree.

[0229] Control module 58 is used to control vehicle movement based on the target trajectory tree.

[0230] Figure 12 This is a schematic diagram of the structure of a vehicle control device provided in another exemplary embodiment of this disclosure.

[0231] In some alternative embodiments, in the above... Figure 11 Based on the illustrated embodiments, as Figure 12 As shown, the fourth processing module 54 includes: a first determining unit 541, a second determining unit 542, a third determining unit 543, and a fourth determining unit 544.

[0232] The first determining unit 541 is used to determine the predicted trajectory of the dynamic object and the probability corresponding to the predicted trajectory based on the predicted trajectory information of the first object.

[0233] The second determining unit 542 is used to determine the first dynamic object and the first predicted trajectory of the first dynamic object that meet the preset conditions based on the probability corresponding to the predicted trajectory.

[0234] The third determining unit 543 is used to determine the bifurcation time of the vehicle's trajectory at the second time based on the first vehicle state information and the first predicted trajectory of the first dynamic object.

[0235] The fourth determining unit 544 is used to determine the initial trajectory tree based on the bifurcation time.

[0236] In some optional embodiments, the second determining unit 542 is specifically used for:

[0237] Based on the probability corresponding to the predicted trajectory and a first probability threshold, a first candidate predicted trajectory is determined from the predicted trajectories; based on the first candidate predicted trajectory and a second probability threshold, a second candidate predicted trajectory is determined from the first candidate predicted trajectories using a pre-configured intelligent driver model; based on the second candidate predicted trajectory, dynamic objects with a number of second candidate predicted trajectories greater than a first preset number are determined as first dynamic objects; based on the second candidate predicted trajectories of the first dynamic objects, a first predicted trajectory of the first dynamic object is determined.

[0238] In some alternative embodiments, based on any of the above embodiments, such as Figure 12 As shown, the fifth processing module 55 includes: a first processing unit 551, a second processing unit 552, and a third processing unit 553.

[0239] The first processing unit 551 is used to determine the dynamic boundary constraint information of the vehicle based on the predicted trajectory information of the first object.

[0240] The second processing unit 552 is used to determine the static boundary constraint information of the vehicle based on the second object state information.

[0241] The third processing unit 553 is used to determine constraint information based on dynamic boundary constraint information and static boundary constraint information.

[0242] In some optional embodiments, the second processing unit 552 is specifically used for:

[0243] Based on the state information of the second object, the first contour information of the static object is determined; based on the first contour information, the static boundary constraint information is determined.

[0244] In some optional embodiments, the second processing unit 552 is specifically used for:

[0245] Based on the first contour information, the distance gradient field information corresponding to the static boundary is determined; based on the distance gradient field information, the static boundary constraint information is determined.

[0246] In some optional embodiments, the second processing unit 552 is specifically configured to:

[0247] determine first boundary information based on the first contour information; determine second boundary information based on the first boundary information and a safety distance threshold corresponding to the static object; and determine distance gradient field information based on the first boundary information and the second boundary information.

[0248] In some optional embodiments, the first processing unit 551 is specifically configured to:

[0249] determine second predicted trajectory of the second dynamic object at the second time based on the first object predicted trajectory information, determine lateral constraint information and / or longitudinal constraint information corresponding to the vehicle at the second time based on the second predicted trajectory, and determine dynamic boundary constraint information based on the lateral constraint information and / or the longitudinal constraint information.

[0250] In some optional embodiments, the first processing unit 551 is specifically configured to:

[0251] determine second contour information of the second dynamic object at the second time based on the second predicted trajectory, determine a boundary point set corresponding to the second time based on the second contour information, the boundary point set being a point set in a local coordinate system of the vehicle at the first time, determine the lateral constraint information based on the boundary point set, and / or determine a longitudinal travel distance from the vehicle to the second dynamic object based on the second contour information, determine a first longitudinal safety boundary of the vehicle at the second time based on the longitudinal travel distance, determine a second longitudinal safety boundary based on the first longitudinal safety boundary and a preset distance threshold, and determine the longitudinal constraint information based on the first longitudinal safety boundary and the second longitudinal safety boundary.

[0252] In some optional embodiments, the first processing unit 551 is specifically configured to:

[0253] determine at least one of a speed upper limit value, an acceleration upper limit value, an acceleration lower limit value, an acceleration change rate upper limit value, and an acceleration change rate lower limit value of the vehicle based on the first longitudinal safety boundary and the second longitudinal safety boundary, and determine the longitudinal constraint information based on the first longitudinal safety boundary and the second longitudinal safety boundary, and the at least one of the speed upper limit value, the acceleration upper limit value, the acceleration lower limit value, the acceleration change rate upper limit value, and the acceleration change rate lower limit value.

[0254] In some optional embodiments, on the basis of any of the above-mentioned embodiments, as shown in Figure 12 the sixth processing module 56 includes a fifth determination unit 561, a sixth determination unit 562, and a seventh determination unit 563.

[0255] The fifth determining unit 561 is configured to determine state variable and control variable corresponding to the node in the initial trajectory tree; the initial value of the state variable is the initial state variable corresponding to the node in the initial trajectory tree, and the initial value of the control variable is the initial control variable corresponding to the node.

[0256] The sixth determining unit 562 is configured to determine at least one of the following cost sub-functions of the state variable and the control variable based on the constraint information, the state variable and the control variable: centripetal acceleration cost sub-function, centripetal acceleration rate of change cost sub-function, dynamic-static boundary collision time cost sub-function, dynamic lateral constraint cost sub-function, and dynamic longitudinal constraint cost sub-function.

[0257] The seventh determining unit 563 is configured to construct a target function based on the cost sub-function.

[0258] In some optional embodiments, the seventh processing module 57 includes a fourth processing unit 571 configured to perform iterative update on the initial trajectory tree based on the first vehicle state information, the target function and the preconfigured state transition rule by an iterative linear quadratic regulator, to obtain the target trajectory tree.

[0259] In some optional embodiments, the fourth processing unit 571 is specifically configured to:

[0260] determine an initial state of the vehicle based on the first vehicle state information; determine a current state variable and a current control variable corresponding to the node based on the initial trajectory tree; update the current control variable based on the current state variable, the current control variable and the target function corresponding to the node, to obtain an updated first control variable; update the current state variable based on the initial state, the first control variable and the preconfigured state transition function, to obtain an updated first state variable; in response to the iteration end condition not being met, perform the steps of updating the current control variable based on the current state variable, the current control variable and the target function corresponding to the node, to obtain an updated first control variable, and the steps subsequent to the steps, iteratively, with the first state variable as the current state variable and the first control variable as the current control variable; and in response to the iteration end condition being met, obtain the target trajectory tree based on the target state variable and the target control variable corresponding to the node, with the first state variable as the target state variable corresponding to the node and the first control variable as the target control variable corresponding to the node.

[0261] The above-mentioned embodiments of the present disclosure can be implemented independently or in any combination without conflict, and can be set according to actual needs. The present disclosure is not limited.

[0262] The beneficial technical effects of the exemplary embodiments of the device can be referred to the corresponding beneficial technical effects of the exemplary method described above, which will not be repeated here.

[0263] Exemplary Electronic Device

[0264] Figure 13 FIG. 1 is a structural diagram of an electronic device according to an embodiment of the disclosure.

[0265] The processor 91 can be a central processing unit (CPU) or other form of processing unit that has data processing capability and / or instruction executing capability, and can control other components in the electronic device 90 to perform desired functions.

[0266] The memory 92 can include one or more computer program products that can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory, for example, can include random access memory (RAM), cache memory, and / or the like. The non-volatile memory, for example, can include read-only memory (ROM), hard disk, flash memory, and / or the like. One or more computer program instructions can be stored on the computer-readable storage media, and the processor 91 can execute the one or more computer program instructions to implement the methods of the various embodiments of the disclosure and / or other desired functions.

[0267] In one example, the electronic device 90 can further include an input device 93 and an output device 94, which are interconnected through a bus system and / or other form of connection mechanism (not shown).

[0268] The input device 93 can further include, for example, a touch screen, a microphone, various sensors, and / or the like. The sensors, for example, can include an image sensor (e.g., a camera, a video camera, and / or the like), a laser radar, a millimeter wave radar, an ultrasonic radar, a positioning sensor, a pressure sensor, an air quality sensor, a temperature sensor, and / or the like. The image sensor, the laser radar, the millimeter wave radar, the ultrasonic radar, and / or the like can be used for perception of a surrounding environment, i.e., detecting dynamic and static objects in the surrounding environment. The dynamic and static objects, for example, can include static objects such as lane lines, road edges, arrows, signboards, trees, buildings, and / or the like, and dynamic objects such as surrounding vehicles, pedestrians, cyclists, and / or the like. The positioning sensor is used to implement positioning of a movable device (e.g., a self-driving vehicle, a robot, and / or the like) in which the electronic device is located. The positioning sensor, for example, can include an inertial measurement unit (IMU), a global positioning system (GPS), and / or the like. The pressure sensor can be used to detect seat pressure. The temperature sensor can be used to detect temperature in a vehicle cabin. The air quality sensor can be used to detect air quality in the vehicle cabin.

[0269] The output device 94 can externally output various information, which can include, for example, a display, a speaker, a communication network and a remote output device connected thereto, and the like.

[0270] Of course, in order to simplify, Figure 13 Only some of the components of the electronic device 90 related to the present disclosure are shown in the middle, and components such as a bus, an input / output interface, and the like are omitted. In addition to this, the electronic device 90 can further include any other appropriate components according to a specific application.

[0271] Exemplary Computer Program Product and Computer-Readable Storage Medium

[0272] In addition to the above-mentioned method and device, embodiments of the present disclosure can also provide a computer program product including computer program instructions, which, when executed by a processor, cause the processor to perform the steps in the methods of various embodiments of the present disclosure described in the "Exemplary Methods" section above.

[0273] The computer program product can be written in any combination of one or more programming languages, including an object-oriented programming language such as Java, C++, and the like, and conventional procedural programming languages, such as the "C" programming language, or the like. The program code can execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server.

[0274] In addition, embodiments of the present disclosure can also be a computer readable storage medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to perform the steps in the methods of various embodiments of the present disclosure described in the "Exemplary Methods" section above.

[0275] The computer readable storage medium can take any combination of one or more of the following: a readable signal medium or a readable storage medium. The readable storage medium, for example, but not limited to, includes an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples (non-exhaustive list) of readable storage medium include: an electrical connection having one or more wires, a portable disc, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0276] The basic principles of the present disclosure are described above with reference to specific embodiments, but the advantages, benefits and effects mentioned in the present disclosure are only examples and are not considered to be mandatory for each embodiment of the present disclosure. In addition, the specific details of the above disclosure are only for the purpose of illustration and understanding, and are not considered to limit the present disclosure to the above specific details. It is necessary to implement the present disclosure.

[0277] Those skilled in the art can make various modifications and variations to the present disclosure without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is also intended to include these modifications and variations.

Claims

1. A vehicle control method, comprising: Determine the vehicle's initial vehicle status information at the first moment; Determine the first object state information of the dynamic object at the first moment, and the second object state information of the static object at the first moment; Based on the first object state information, the first object prediction trajectory information of the dynamic object at a second time moment is determined, wherein the first time moment is earlier than the second time moment; Based on the predicted trajectory information of the first object and the state information of the first vehicle, the initial trajectory tree of the vehicle at the second time moment is determined; Based on the predicted trajectory information of the first object and the state information of the second object, the constraint information is determined; Based on the constraint information, construct the objective function; Based on the first vehicle state information and the objective function, the initial trajectory tree is updated to determine the target trajectory tree; The vehicle's movement is controlled based on the target trajectory tree.

2. The method according to claim 1, wherein, The step of determining the initial trajectory tree of the vehicle at the second time moment based on the predicted trajectory information of the first object and the state information of the first vehicle includes: Based on the predicted trajectory information of the first object, the predicted trajectory of the dynamic object and the probability corresponding to the predicted trajectory are determined; Based on the probability corresponding to the predicted trajectory, a first dynamic object that meets the preset conditions and a first predicted trajectory of the first dynamic object are determined. Based on the first vehicle status information and the first predicted trajectory of the first dynamic object, the bifurcation time of the vehicle's trajectory at the second time moment is determined. The initial trajectory tree is determined based on the bifurcation time.

3. The method according to claim 2, wherein, The step of determining a first dynamic object and a first predicted trajectory of the first dynamic object that satisfy preset conditions based on the probability corresponding to the predicted trajectory includes: Based on the probability corresponding to the predicted trajectory and the first probability threshold, a first candidate predicted trajectory is determined from the predicted trajectory; Based on the first candidate predicted trajectory and the second probability threshold, a second candidate predicted trajectory is determined from the first candidate predicted trajectory using a pre-configured intelligent driver model. Based on the second candidate predicted trajectory, dynamic objects with a number of second candidate predicted trajectories greater than a first preset number are determined from the dynamic objects and used as the first dynamic object. Based on the second candidate predicted trajectory of the first dynamic object, the first predicted trajectory of the first dynamic object is determined.

4. The method according to claim 1, wherein, The step of determining constraint information based on the predicted trajectory information of the first object and the state information of the second object includes: Based on the predicted trajectory information of the first object, the dynamic boundary constraint information of the vehicle is determined; Based on the second object state information, the static boundary constraint information of the vehicle is determined; The constraint information is determined based on the dynamic boundary constraint information and the static boundary constraint information.

5. The method according to claim 4, wherein, Determining the static boundary constraint information of the vehicle based on the second object state information includes: Based on the second object state information, the first contour information of the static object is determined; Based on the first contour information, the static boundary constraint information is determined.

6. The method according to claim 5, wherein, The step of determining the static boundary constraint information based on the first contour information includes: Based on the first contour information, determine the distance gradient field information corresponding to the static boundary; Based on the distance gradient field information, the static boundary constraint information is determined.

7. The method according to claim 6, wherein, The step of determining the distance gradient field information corresponding to the static boundary based on the first contour information includes: Based on the first contour information, the first boundary information is determined; Based on the first boundary information and the safety distance threshold corresponding to the static object, the second boundary information is determined; Based on the first boundary information and the second boundary information, the distance gradient field information is determined.

8. The method according to claim 4, wherein, Based on the predicted trajectory information of the first object, the dynamic boundary constraint information of the vehicle is determined, including: Based on the predicted trajectory information of the first object, the second predicted trajectory of the second dynamic object that meets the preset object conditions is determined at the second time. Based on the second predicted trajectory, determine the lateral constraint information and / or longitudinal constraint information of the vehicle at the second moment; The dynamic boundary constraint information is determined based on the lateral constraint information and / or the longitudinal constraint information.

9. The method according to claim 8, wherein, The step of determining the lateral constraint information and / or longitudinal constraint information of the vehicle at the second time moment based on the second predicted trajectory includes: Based on the second predicted trajectory, determine the second contour information of the second dynamic object at the second moment; Based on the second contour information, the boundary point set corresponding to the second time moment is determined; the boundary point set is the point set of the vehicle in the local coordinate system at the first time moment. Based on the set of boundary points, determine the lateral constraint information; and / or, Based on the second contour information, the longitudinal travel distance from the vehicle to the second dynamic object is determined; Based on the longitudinal travel distance, the first longitudinal safety boundary of the vehicle at the second moment is determined; Based on the first longitudinal security boundary and the preset distance threshold, a second longitudinal security boundary is determined; The longitudinal constraint information is determined based on the first longitudinal safety boundary and the second longitudinal safety boundary.

10. The method according to claim 9, wherein, The determination of the longitudinal constraint information based on the first longitudinal safety boundary and the second longitudinal safety boundary includes: Based on the first longitudinal safety boundary and the second longitudinal safety boundary, at least one of the following is determined: the upper limit of the vehicle's speed, the upper limit of its acceleration, the lower limit of its acceleration, the upper limit of its rate of change of acceleration, and the lower limit of its rate of change of acceleration. The longitudinal constraint information is determined based on the first longitudinal safety boundary and the second longitudinal safety boundary, as well as at least one of the upper limit value of velocity, the upper limit value of acceleration, the lower limit value of acceleration, the upper limit value of the rate of change of acceleration, and the lower limit value of the rate of change of acceleration.

11. The method according to any one of claims 1-10, wherein, The construction of the objective function based on the constraint information includes: Determine the state variables and control variables corresponding to the nodes in the initial trajectory tree; the initial value of the state variables is the initial state variable corresponding to the node in the initial trajectory tree, and the initial value of the control variables is the initial control variable corresponding to the node; Based on the constraint information, the state variables, and the control variables, at least one of the following cost sub-functions for the state variables and the control variables is determined: centripetal acceleration cost sub-function, centripetal acceleration rate of change cost sub-function, dynamic and static boundary collision time cost sub-function, dynamic lateral constraint cost sub-function, and dynamic longitudinal constraint cost sub-function. Based on the cost sub-function, the objective function is constructed.

12. The method according to any one of claims 1-10, wherein, The step of updating the initial trajectory tree based on the first vehicle state information and the objective function to determine the target trajectory tree includes: Based on the first vehicle state information, the objective function, and the pre-configured state transition rules, the initial trajectory tree is iteratively updated using an iterative linear quadratic regulator to obtain the target trajectory tree.

13. The method according to claim 12, wherein, The step of iteratively updating the initial trajectory tree based on the first vehicle state information, the objective function, and the pre-configured state transition rules, through an iterative linear quadratic regulator, to obtain the target trajectory tree includes: Based on the first vehicle status information, the initial state of the vehicle is determined; Based on the initial trajectory tree, determine the current state variable and current control variable corresponding to the node; Based on the current state quantity, the current control quantity, and the objective function corresponding to the node, the current control quantity is updated to obtain the updated first control quantity; Based on the initial state, the first control variable, and the pre-configured state transition function, the current state variable is updated to obtain the updated first state variable; In response to the failure to meet the iteration termination condition, the first state quantity is taken as the current state quantity, the first control quantity is taken as the current control quantity, and the steps of updating the current control quantity based on the current state quantity, the current control quantity and the objective function corresponding to the node, and obtaining the updated first control quantity, and the subsequent steps are executed iteratively. In response to the fulfillment of the iteration termination condition, the first state variable is taken as the target state variable corresponding to the node, and the first control variable is taken as the target control variable corresponding to the node. Based on the target state variable and the target control variable corresponding to the node, the target trajectory tree is obtained.

14. A vehicle control device, comprising: The first processing module is used to determine the first vehicle status information at the first moment. The second processing module is used to determine the first object state information of the dynamic object at the first moment and the second object state information of the static object at the first moment. The third processing module is used to determine the first object prediction trajectory information of the dynamic object at a second time based on the first object state information, wherein the first time is earlier than the second time. The fourth processing module is used to determine the initial trajectory tree of the vehicle at the second time point based on the predicted trajectory information of the first object and the state information of the first vehicle. The fifth processing module is used to determine constraint information based on the predicted trajectory information of the first object and the state information of the second object; The sixth processing module is used to construct an objective function based on the constraint information; The seventh processing module is used to update the initial trajectory tree based on the first vehicle state information and the objective function, and determine the target trajectory tree; The control module is used to control the vehicle's movement based on the target trajectory tree.

15. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in any one of claims 1-13.

16. An electronic device, the electronic device comprising: processor; Memory used to store the processor's executable instructions; The processor is configured to read the executable instructions from the memory and execute the instructions to implement the method described in any one of claims 1-13.

17. A computer program product comprising instructions that, when executed, implement the method of any one of claims 1-13.

Images