A vehicle trajectory planning method and device, and a storage medium

Abstract

The application relates to a vehicle trajectory planning method and device and a storage medium, which comprises the following steps: searching a reference line based on a three-lane model, predicting future trajectory points of a vehicle and obstacles around the vehicle based on the reference line, performing space-time discretization on a current position point of the vehicle and the plurality of predicted trajectory points, binding corresponding points of each time layer, connecting the points bound by any adjacent two time layers to obtain a plurality of connection edges, connecting adjacent two time layers to form a plurality of connection edges, that is, the vehicle has a plurality of possible driving trajectories in a certain time period, further, calculating the cost of each connection edge according to a preset cost function, and performing trajectory planning according to the cost of each connection edge, a preset constraint condition and the predicted trajectory points of the obstacles bound by the plurality of time layers to obtain a vehicle planning trajectory which satisfies the lane boundary constraint and has the minimum cost, so that vehicle trajectory planning in a dynamic obstacle avoidance scene is realized.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle trajectory planning, and in particular to a vehicle trajectory planning method, a device, and a storage medium thereof. Background Art

[0002] Vehicle trajectory planning involves determining a vehicle's trajectory through a trajectory planning algorithm, given road network and environmental information, to achieve a specific goal and meet various constraints. The design of the vehicle trajectory planning algorithm should take into account road conditions, vehicle dynamics, safety, efficiency, and other requirements to provide a reliable, efficient, and safe trajectory. With the advancement of autonomous driving technology, vehicle trajectory planning faces greater challenges, including complex traffic environments, multi-vehicle coordination, and dynamic obstacle handling. Therefore, different scenarios and requirements may require different vehicle trajectory planning algorithms to achieve optimal results. Summary of the Invention

[0003] The purpose of this application is to propose a vehicle trajectory planning method and device, and a computer-readable storage medium to realize vehicle trajectory planning in dynamic obstacle avoidance scenarios.

[0004] To achieve the above objectives, an embodiment of the present application provides a vehicle trajectory planning method, the method comprising:

[0005] Searching for reference lines based on a three-lane model; the reference lines include the center lines of the lane and the two lanes to the left, or the center lines of the lane and the two lanes to the right, or the center lines of the lane and the left and right adjacent lanes;

[0006] generating a plurality of predicted trajectory points of the vehicle and obstacles around the vehicle within a preset distance range based on the reference line;

[0007] Binding the current position of the vehicle and the multiple predicted trajectory points to multiple preset time layers, with a certain time interval between two adjacent time layers; wherein the first time layer is bound to the current position of the vehicle, and the other time layers are bound to the predicted trajectory points of the vehicle and the obstacles;

[0008] Connect the current position points or predicted trajectory points of the vehicle bound to any two adjacent time layers to obtain multiple connection edges, and connect any two points to form a connection edge;

[0009] A preset cost function calculates the cost of each connecting edge, and based on the cost of each connecting edge, preset constraints, and the predicted trajectory points of obstacles bound to the multiple time layers, trajectory planning is performed to obtain a vehicle planning trajectory that satisfies the lane boundary constraints and has the lowest cost.

[0010] Preferably, the predicted trajectory points include horizontal predicted trajectory points and vertical predicted trajectory points;

[0011] The lateral predicted trajectory point of the host vehicle is a future trajectory point after the lateral movement of the host vehicle is predicted, and the longitudinal predicted trajectory point of the host vehicle is a future trajectory point after the longitudinal movement of the host vehicle is predicted.

[0012] The lateral predicted trajectory point of the obstacle is a future trajectory point after the lateral movement of the obstacle is predicted, and the longitudinal predicted trajectory point of the obstacle is a future trajectory point after the longitudinal movement of the obstacle is predicted.

[0013] Preferably, the cost of each connection edge includes at least one of an acceleration cost, an acceleration change rate cost, a reference line distance cost, a collision cost, and a lateral boundary cost.

[0014] The acceleration cost is a cost corresponding to the acceleration required for the host vehicle to move from one point of the connection edge to another point.

[0015] The acceleration change rate cost is a cost corresponding to the acceleration change rate required for the host vehicle to move from one point of the connection edge to another point.

[0016] The reference line distance cost is a cost corresponding to the distance from the other point to the reference line of the lane in which the host vehicle is located when the host vehicle moves from one point of the connection edge to the other point.

[0017] The collision cost is a cost corresponding to whether a collision occurs between the host vehicle and the obstacle during the movement of the host vehicle from one point of the connection edge to another point.

[0018] The lateral boundary cost is a cost corresponding to the distance between the host vehicle and the left and right lane boundaries during the movement of the host vehicle from one point of the connection edge to another point.

[0019] Preferably, the constraint condition includes a lane boundary constraint and an obstacle linear boundary constraint.

[0020] The lane boundary constraint means that the host vehicle can cross a dashed line or press a dashed line, but cannot cross a solid line or press a solid line.

[0021] The constraint condition includes an obstacle linear boundary constraint, and the obstacle linear boundary constraint means that the obstacle boundary is represented by a straight line, one side of the straight line is regarded as a collision, the other side of the straight line is regarded as safe and no collision, and the host vehicle needs to travel on the other side of the straight line.

[0022] Preferably, the multiple time layers include multiple time layers of a first time period and multiple time layers of a second time period, and any two adjacent time layers of the multiple time layers of the first time period are separated by t1 time, and any two adjacent time layers of the multiple time layers of the second time period are separated by t2 time; wherein, t2 time and t1 time are both preset values, and t2 time is less than t1 time.

[0023] The present application also provides a vehicle trajectory planning device, the device comprising:

[0024] A reference line acquisition module is used to search for reference lines based on the three-lane model; the reference lines include the center lines of the lane and the two lanes to the left, the center lines of the lane and the two lanes to the right, or the center lines of the lane and the left and right adjacent lanes;

[0025] A trajectory point prediction module, configured to generate a plurality of predicted trajectory points of the vehicle and obstacles around the vehicle within a preset distance range based on the reference line;

[0026] A spatiotemporal discretization module is used to bind the current position of the vehicle and the multiple predicted trajectory points to multiple preset time layers, with a certain time interval between two adjacent time layers; wherein the first time layer is bound to the current position of the vehicle, and the other time layers are bound to the predicted trajectory points of the vehicle and obstacles;

[0027] The connection edge generation module is used to connect the current position points or predicted trajectory points of the vehicle bound to any two adjacent time layers to obtain multiple connection edges. Any two points are connected to form a connection edge;

[0028] The optimization solution module is used to calculate the cost of each connecting edge using a preset cost function, and perform trajectory planning based on the cost of each connecting edge, preset constraints, and the predicted trajectory points of obstacles bound to the multiple time layers to obtain a vehicle planning trajectory that satisfies the lane boundary constraints and minimizes the cost.

[0029] Preferably, the predicted trajectory points include horizontal predicted trajectory points and vertical predicted trajectory points;

[0030] The predicted lateral trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves laterally, and the predicted longitudinal trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves longitudinally.

[0031] The predicted lateral trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the lateral movement, and the predicted longitudinal trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the longitudinal movement.

[0032] Preferably, the cost of each connecting edge includes at least one of an acceleration cost, an acceleration change rate cost, a reference line distance cost, a collision cost, and a lateral boundary cost;

[0033] The acceleration cost is the cost corresponding to the acceleration required for the vehicle to move from one point on the connecting edge to another point;

[0034] The acceleration change rate cost is the cost corresponding to the acceleration change rate required for the vehicle to move from one point on the connecting edge to another point;

[0035] The reference line distance cost is the cost corresponding to the distance from one point on the connecting edge to another point on the lane where the vehicle is located;

[0036] The collision cost is the cost corresponding to whether the vehicle collides with an obstacle when moving from one point on the connecting edge to another point;

[0037] The lateral boundary cost is the cost corresponding to the distance between the vehicle and the left and right lane boundaries when the vehicle moves from one point on the connecting edge to another point.

[0038] Preferably, the constraints include lane boundary constraints and obstacle linear boundary constraints;

[0039] The lane boundary constraint means that the vehicle can cross or press the dashed line, but cannot cross or press the solid line;

[0040] The constraint conditions include obstacle linear boundary constraints, which refer to representing the obstacle boundary with a straight line. A collision is considered on one side of the straight line, and a safe collision-free state is considered on the other side of the straight line. The vehicle must travel on the other side of the straight line.

[0041] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the vehicle trajectory planning method as described above is implemented.

[0042] The vehicle trajectory planning method and device, and computer-readable storage medium provided in the embodiments of the present application have the following beneficial effects:

[0043] The embodiment of the present application searches a reference line based on a three-lane model, predicts future trajectory points of a host vehicle and obstacles around the host vehicle based on the reference line, performs spatio-temporal discretization on a current position point of the host vehicle and the plurality of predicted trajectory points, binds corresponding points of each time layer, and connects points bound by any adjacent two time layers to obtain a plurality of connection edges. The connection edge can be understood as a trajectory line segment that the host vehicle can travel. The connection edges are formed between adjacent two time layers, that is, the host vehicle has a plurality of possible trajectory lines in a certain time period. Further, a cost of each connection edge is calculated according to a preset cost function, and a vehicle planning trajectory that satisfies a lane boundary constraint and has a minimum cost is obtained according to the cost of each connection edge, a preset constraint condition, and predicted trajectory points of obstacles bound by the plurality of time layers, so as to realize vehicle trajectory planning in a dynamic obstacle avoidance scenario.

[0044] Details and advantages of the embodiment of the present application that are not described in detail are described in detail in the specific implementation. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0046] Figure 1 A flowchart of a vehicle trajectory planning method in an embodiment of the present application.

[0047] Figure 2 A schematic diagram of trajectory point connection between adjacent time layers in an embodiment of the present application.

[0048] Figure 3 A schematic diagram of a drivable area under a lane boundary constraint in an embodiment of the present application.

[0049] Figure 4 A structural diagram of a vehicle trajectory planning device in an embodiment of the present application. DETAILED DESCRIPTION

[0050] The detailed description of the drawings is intended to illustrate the current embodiments of the present application, but is not intended to represent the only form that the present application can be implemented. It should be understood that the same or equivalent functions can be completed by different embodiments intended to be included in the spirit and scope of the present application.

[0051] Reference Figure 1 One embodiment of the present application provides a vehicle trajectory planning method, comprising the following steps:

[0052] Step S10: searching for reference lines based on the three-lane model; the reference lines include the center lines of the lane and the two lanes on the left, or the center lines of the lane and the two lanes on the right, or the center lines of the lane and the left and right adjacent lanes.

[0053] Specifically, in the three-lane model, the road is divided into three parallel lanes, each lane allows vehicles to travel in the same direction. Vehicles can change lanes between different lanes and choose the appropriate lane according to traffic conditions; when the vehicle is in the rightmost lane, it can search for the center line of its lane and the two lanes on the left; when the vehicle is in the leftmost lane, it can search for the center line of its lane and the two lanes on the right; when the vehicle is in the middle lane, it can search for the center line of its lane and the left and right adjacent lanes.

[0054] Step S20 : ​​generating a plurality of predicted trajectory points of the vehicle and obstacles around the vehicle within a preset distance range based on the reference line.

[0055] Specifically, obstacles around the vehicle can be detected by the vehicle's sensing unit, which includes but is not limited to a vehicle camera, ultrasonic radar, lidar, etc. The preset distance range is the upper limit of the trajectory point prediction. If the range is exceeded, inaccurate predictions are likely to occur. Based on this, it can also be understood that the trajectory planning of this embodiment is carried out in sections, and the planned trajectory is continuously adjusted during the vehicle's driving to achieve trajectory planning in a dynamic obstacle avoidance scenario. If the parking point is within the preset distance range, the end point of the output planned trajectory is the parking point. The parking point can be the destination of the vehicle or a temporary parking location, such as a vehicle ahead parking or a red light at an intersection. It should be noted that the multiple predicted trajectory points of the vehicle and the multiple predicted trajectory points of the obstacle are evenly distributed, and the interval between two adjacent trajectory points is the same, for example, 3.5 meters. The corresponding preset distance range can be within 133 meters in the longitudinal direction in front of the vehicle. 133 meters is calculated based on a maximum speed of 80 km / h for 6 seconds and needs to be as divisible as possible by 3.5.

[0056] In step S30, the current position point of the vehicle and the multiple predicted trajectory points are respectively bound to multiple preset time layers, with a certain time interval between two adjacent time layers; among them, the first time layer is bound to the current position point of the vehicle, and the other time layers are bound to the predicted trajectory points of the vehicle and the obstacles.

[0057] Specifically, this embodiment discretizes the current position point of the vehicle and the multiple predicted trajectory points in time and space, and binds each time layer to the corresponding point. For example, the time layer is set at a time interval of 0.2 seconds. The first time layer of 0 seconds is bound to the current position point of the vehicle, the second time layer of 0.2 seconds is bound to the trajectory point that the vehicle and the obstacle may reach after 0.2 seconds, and the third time layer of 0.4 seconds is bound to the trajectory point that the vehicle and the obstacle may reach after 0.4 seconds. And so on, all time layers are bound to the corresponding trajectory points.

[0058] In step S40 , the current position points or predicted trajectory points of the vehicle bound to any two adjacent time layers are connected to obtain a plurality of connection edges, and any two points are connected to form a connection edge.

[0059] Specifically, if Figure 2 As shown, this embodiment connects the points bound by any two adjacent time layers to obtain multiple connecting edges. The connecting edges can be understood as the trajectory segments that the vehicle may travel. The connection between two adjacent time layers forms many connecting edges, which means that the vehicle has many possible driving trajectories in a certain time period.

[0060] In step S50, a preset cost function is used to calculate the cost of each connecting edge, and trajectory planning is performed based on the cost of each connecting edge, the preset constraints, and the predicted trajectory points of the obstacles bound to the multiple time layers to obtain a vehicle planning trajectory that satisfies the lane boundary constraints and has the minimum cost.

[0061] Specifically, the constraints can be set according to actual conditions, such as constraints on road traffic rules (such as stop signs, traffic lights, lane regulations), and constraints on the dynamic performance of the vehicle (such as maximum acceleration, maximum speed, minimum turning radius, etc.). In step S50, the predicted trajectory points of the obstacles bound to the multiple time layers can be determined based on the predicted trajectory points of the obstacles in the two time layers corresponding to each connecting edge, that is, the possible positions of the obstacles in the two time layers can be determined, thereby calculating the collision risk, thereby helping to plan the driving trajectory of the vehicle, avoid collisions, and improve safety. The purpose of step S50 is to address a vehicle planning trajectory that meets the lane boundary constraints and has the lowest cost. The vehicle planning trajectory is formed by connecting edges between different time layers. Based on the above embodiment method, vehicle trajectory planning for dynamic obstacle avoidance scenarios can be achieved.

[0062] In some embodiments, since vehicle travel may include movement in both longitudinal and lateral directions, the predicted trajectory points of the vehicle and the obstacle in this embodiment include lateral predicted trajectory points and longitudinal predicted trajectory points.

[0063] The predicted lateral trajectory points of the vehicle are the predicted future trajectory points after the vehicle's lateral movement, and the predicted longitudinal trajectory points of the vehicle are the predicted future trajectory points after the vehicle's longitudinal movement. Specifically, the lateral trajectory point prediction can be obtained by evenly distributing points from the centerline of the vehicle's lane to the centerline of the adjacent lane, or from the centerline of the vehicle's lane to the boundary of the adjacent lane.

[0064] The predicted lateral trajectory points of the obstacle are the predicted future trajectory points after the obstacle moves laterally, and the predicted longitudinal trajectory points of the obstacle are the predicted future trajectory points after the obstacle moves longitudinally. Specifically, the longitudinal trajectory point prediction can be obtained by evenly distributing points along the longitudinal direction of the reference line.

[0065] In some embodiments, the cost of each connecting edge includes at least one of an acceleration cost, an acceleration change rate cost, a reference line distance cost, a collision cost, and a lateral boundary cost.

[0066] Among them, the acceleration cost is the cost corresponding to the acceleration required for the vehicle to move from one point on the connecting edge to another point; the calculation method of the acceleration cost can be defined by yourself. For example, the greater the acceleration required for the vehicle to move from one point on the connecting edge to another point, the greater the corresponding acceleration cost; the smaller the acceleration required for the vehicle to move from one point on the connecting edge to another point, the smaller the corresponding acceleration cost, because if an emergency occurs in front during the acceleration process, it is easy to have no time to avoid it.

[0067] For example, it can be set to the following function:

[0068]

[0069] Among them, cost(a) represents the acceleration cost, and a represents the acceleration.

[0070] The acceleration change rate cost is the cost corresponding to the acceleration change rate (jerk) required for the vehicle to move from one point on the connecting edge to another point.

[0071] The calculation method of the acceleration rate cost can be customized. For example, the greater the acceleration required to move the vehicle from one point on the connecting edge to another, the greater the corresponding acceleration rate cost. The smaller the acceleration required to move the vehicle from one point on the connecting edge to another, the smaller the corresponding acceleration rate cost. This is because if an unexpected situation occurs ahead during the acceleration change, it may be difficult to avoid it in time.

[0072] For example, it can be set to the following function:

[0073]

[0074] Among them, cost(j) represents the acceleration change rate cost, and j represents the acceleration change rate.

[0075] The reference line distance cost is the cost corresponding to the distance from one point on the connecting edge to another point on the lane where the vehicle is located, as shown in the following function:

[0076] cost((x,y),(x ref ,y ref ))=(xx ref ) 2 +(yy ref ) 2

[0077] Among them, cost((x,y),(x ref ,y ref )) represents the reference line distance cost, x, y are the vertical coordinates and horizontal coordinates of the other point respectively, x ref 、y ref are the longitudinal coordinate and lateral coordinate of the point on the lane centerline closest to the other point respectively.

[0078] The collision cost is the cost corresponding to whether the vehicle collides with an obstacle when moving from one point on the connecting edge to another point, for example:

[0079] If the distance d between the ego vehicle and the obstacle is greater than the preset safe distance dis_safe, the collision cost cost(ego,obs) = 0, where ego is the ego vehicle and obs is the obstacle.

[0080] If the distance d between the vehicle and the obstacle is less than or equal to the preset safe distance dis_safe, then further calculation is performed to determine whether a collision occurs. If a collision occurs, the collision cost cost(ego,obs) = ∞. If no collision occurs, the collision cost cost(ego,obs) is calculated based on the distance d between the vehicle and the obstacle, for example, 1 / d - 1 / dis_safe.

[0081] The lateral boundary cost is the cost corresponding to the distance between the vehicle and the left and right lane boundaries when the vehicle moves from one point on the connecting edge to another point, such as the following function:

[0082]

[0083] l safe_buffer It is the preset lateral safety distance, which can be set to 0.2 times ego_width. l_min refers to the right lateral boundary, l_max refers to the left lateral boundary, and ego_width is the width of the vehicle.

[0084] In some embodiments, the constraints include lane boundary constraints and obstacle linear boundary constraints;

[0085] The lane boundary constraint means that the vehicle can cross the dotted line or drive on the dotted line, but cannot cross the solid line or drive on the solid line. Specifically, according to road traffic rules, the dotted line can be crossed, but the solid line cannot be crossed. Based on this, the vehicle's drivable range is formed. The drivable range is, for example, Figure 3 As shown, Figure 3 The drivable range is the drivable position range of the vehicle center. When the vehicle center is within this range, there is no compaction line.

[0086] The constraint conditions include obstacle linear boundary constraints, which refer to representing the obstacle boundary with a straight line, considering a collision on one side of the straight line and a safe collision-free state on the other side of the straight line. The vehicle must travel on the other side of the straight line. Specifically, as described above, obstacle information can be obtained through perception units such as vehicle cameras, ultrasonic radars, and lidars. The obstacle information includes the surface structure of the obstacle. In this embodiment, obstacle boundary information is extracted from the obstacle information output by the perception unit, and the obstacle boundary is described by a straight line. The obstacle linear boundary constraint can be understood as a linear inequality, considering a collision on one side of the inequality and a safe collision-free state on the other side.

[0087] In some embodiments, the multiple time layers include multiple time layers of a first time period and multiple time layers of a second time period, and any two adjacent time layers of the multiple time layers of the first time period are separated by t1 time, and any two adjacent time layers of the multiple time layers of the second time period are separated by t2 time; wherein, t2 time and t1 time are both preset values, and t2 time is less than t1 time.

[0088] Specifically, the t2 time is, for example, 0.2 seconds, the t1 time is, for example, 1 second, the first time period is, for example, 0 to 5 seconds, and the second time period is, for example, 5 to 6 seconds. The purpose of this is that if it is too dense, too many search lines will lead to excessive computational consumption; if it is too sparse, the effective sampling range will be too small and the flexibility will be poor. Therefore, in the spatiotemporal union of this embodiment, each layer corresponds to a time, the 5.2-second layer and the 5.4-second layer are different layers, the original connection is sampled at 1 second, and can only be connected from the 5-second layer to the 6-second layer. However, in order to ensure that the last section can pass quickly or stop as soon as possible, it is necessary to allow a connection line from 5 seconds to 5.2 seconds to avoid excessive computational consumption while also improving flexibility.

[0089] Corresponding to the vehicle trajectory planning method of the above embodiment, another embodiment of the present application further provides a vehicle trajectory planning device, the device comprising:

[0090] A reference line acquisition module is used to search for reference lines based on the three-lane model; the reference lines include the center lines of the lane and the two lanes to the left, the center lines of the lane and the two lanes to the right, or the center lines of the lane and the left and right adjacent lanes;

[0091] A trajectory point prediction module, configured to generate a plurality of predicted trajectory points of the vehicle and obstacles around the vehicle within a preset distance range based on the reference line;

[0092] A spatiotemporal discretization module is used to bind the current position of the vehicle and the multiple predicted trajectory points to multiple preset time layers, with a certain time interval between two adjacent time layers; wherein the first time layer is bound to the current position of the vehicle, and the other time layers are bound to the predicted trajectory points of the vehicle and obstacles;

[0093] The connection edge generation module is used to connect the current position points or predicted trajectory points of the vehicle bound to any two adjacent time layers to obtain multiple connection edges. Any two points are connected to form a connection edge;

[0094] The optimization solution module is used to calculate the cost of each connecting edge using a preset cost function, and perform trajectory planning based on the cost of each connecting edge, preset constraints, and the predicted trajectory points of obstacles bound to the multiple time layers to obtain a vehicle planning trajectory that satisfies the lane boundary constraints and minimizes the cost.

[0095] In some embodiments, the predicted trajectory points include horizontal predicted trajectory points and vertical predicted trajectory points;

[0096] The predicted lateral trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves laterally, and the predicted longitudinal trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves longitudinally.

[0097] The predicted lateral trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the lateral movement, and the predicted longitudinal trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the longitudinal movement.

[0098] In some embodiments, the cost of each connecting edge includes at least one of an acceleration cost, an acceleration change rate cost, a reference line distance cost, a collision cost, and a lateral boundary cost;

[0099] The acceleration cost is the cost corresponding to the acceleration required for the vehicle to move from one point on the connecting edge to another point;

[0100] The acceleration change rate cost is the cost corresponding to the acceleration change rate required for the vehicle to move from one point on the connecting edge to another point;

[0101] The reference line distance cost is the cost corresponding to the distance from one point on the connecting edge to another point on the lane where the vehicle is located;

[0102] The collision cost is the cost corresponding to whether the vehicle collides with an obstacle when moving from one point on the connecting edge to another point;

[0103] The lateral boundary cost is the cost corresponding to the distance between the vehicle and the left and right lane boundaries when the vehicle moves from one point on the connecting edge to another point.

[0104] In some embodiments, the constraints include lane boundary constraints and obstacle linear boundary constraints;

[0105] The lane boundary constraint means that the vehicle can cross or press the dashed line, but cannot cross or press the solid line;

[0106] The constraint conditions include obstacle linear boundary constraints, which refer to representing the obstacle boundary with a straight line. A collision is considered on one side of the straight line, and a safe collision-free state is considered on the other side of the straight line. The vehicle must travel on the other side of the straight line.

[0107] In some embodiments, the multiple time layers include multiple time layers of a first time period and multiple time layers of a second time period, any two adjacent time layers of the multiple time layers of the first time period are separated by t1 time, and any two adjacent time layers of the multiple time layers of the second time period are separated by t2 time, and t2 time is less than t1 time.

[0108] It should be noted that the vehicle trajectory planning device of this embodiment corresponds to the vehicle trajectory planning method of the above embodiment. Therefore, the undescribed parts of the vehicle trajectory planning device of this embodiment can be obtained by referring to the contents of the vehicle trajectory planning method of the above embodiment, and will not be repeated here.

[0109] Furthermore, if the vehicle trajectory planning device of the above embodiment is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0110] Another embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the vehicle trajectory planning method as described in the above embodiment is implemented.

[0111] Specifically, the computer-readable storage medium may include: any entity or recording medium that can carry the computer program instructions, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0112] While various embodiments of the present application have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or technological improvements in the marketplace, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A vehicle trajectory planning method, characterized in that: The method comprises: Searching for reference lines based on a three-lane model; the reference lines include the center lines of the lane and the two lanes to the left, or the center lines of the lane and the two lanes to the right, or the center lines of the lane and the left and right adjacent lanes; generating a plurality of predicted trajectory points of the vehicle and obstacles around the vehicle within a preset distance range based on the reference line; Binding the current position of the vehicle and the multiple predicted trajectory points to multiple preset time layers, with a certain time interval between two adjacent time layers; wherein the first time layer is bound to the current position of the vehicle, and the other time layers are bound to the predicted trajectory points of the vehicle and the obstacles; Connect the current position points or predicted trajectory points of the vehicle bound to any two adjacent time layers to obtain multiple connection edges, and connect any two points to form a connection edge; A preset cost function calculates the cost of each connecting edge, and based on the cost of each connecting edge, preset constraints, and the predicted trajectory points of obstacles bound to the multiple time layers, trajectory planning is performed to obtain a vehicle planning trajectory that satisfies the lane boundary constraints and has the minimum cost.

2. The method according to claim 1, characterized in that The predicted trajectory points include horizontal predicted trajectory points and vertical predicted trajectory points; The predicted lateral trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves laterally, and the predicted longitudinal trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves longitudinally. The predicted lateral trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the lateral movement, and the predicted longitudinal trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the longitudinal movement.

3. The method according to claim 1, characterized in that The cost of each connecting edge includes at least one of an acceleration cost, an acceleration change rate cost, a reference line distance cost, a collision cost, and a lateral boundary cost; The acceleration cost is the cost corresponding to the acceleration required for the vehicle to move from one point on the connecting edge to another point; The acceleration change rate cost is the cost corresponding to the acceleration change rate required for the vehicle to move from one point on the connecting edge to another point; The reference line distance cost is the cost corresponding to the distance from one point on the connecting edge to another point on the lane where the vehicle is located; The collision cost is the cost corresponding to whether the vehicle collides with an obstacle when moving from one point on the connecting edge to another point; The lateral boundary cost is the cost corresponding to the distance between the vehicle and the left and right lane boundaries when the vehicle moves from one point on the connecting edge to another point.

4. The method according to claim 1, wherein The preset constraints include lane boundary constraints and obstacle linear boundary constraints; The lane boundary constraint means that the vehicle can cross or press the dashed line, but cannot cross or press the solid line. The constraint conditions include obstacle linear boundary constraints, which refer to representing the obstacle boundary with a straight line. A collision is considered on one side of the straight line, and a safe collision-free state is considered on the other side of the straight line. The vehicle must travel on the other side of the straight line.

5. The method according to claim 1, wherein The multiple time layers include multiple time layers of a first time period and multiple time layers of a second time period, any two adjacent time layers of the multiple time layers of the first time period are separated by t1 time, and any two adjacent time layers of the multiple time layers of the second time period are separated by t2 time; wherein, t2 time and t1 time are both preset values, and t2 time is less than t1 time.

6. A vehicle trajectory planning device, characterized in that: The device comprises: A reference line acquisition module is used to search for reference lines based on the three-lane model; the reference lines include the center lines of the lane and the two lanes to the left, the center lines of the lane and the two lanes to the right, or the center lines of the lane and the left and right adjacent lanes; A trajectory point prediction module, configured to generate a plurality of predicted trajectory points of the vehicle and obstacles around the vehicle within a preset distance range based on the reference line; A spatiotemporal discretization module is used to bind the current position of the vehicle and the multiple predicted trajectory points to multiple preset time layers, with a certain time interval between two adjacent time layers; wherein the first time layer is bound to the current position of the vehicle, and the other time layers are bound to the predicted trajectory points of the vehicle and obstacles; The connection edge generation module is used to connect the current position points or predicted trajectory points of the vehicle bound to any two adjacent time layers to obtain multiple connection edges. Any two points are connected to form a connection edge; The optimization solution module is used to calculate the cost of each connecting edge using a preset cost function, and perform trajectory planning based on the cost of each connecting edge, preset constraints, and the predicted trajectory points of obstacles bound to the multiple time layers to obtain a vehicle planning trajectory that satisfies lane boundary constraints and minimizes the cost.

7. The device according to claim 6, characterized in that The predicted trajectory points include horizontal predicted trajectory points and vertical predicted trajectory points; The predicted lateral trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves laterally, and the predicted longitudinal trajectory points of the vehicle are the predicted future trajectory points after the vehicle moves longitudinally. The predicted lateral trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the lateral movement, and the predicted longitudinal trajectory points of the obstacle are the predicted future trajectory points of the obstacle after the longitudinal movement.

8. The device according to claim 6, characterized in that The cost of each connecting edge includes at least one of an acceleration cost, an acceleration change rate cost, a reference line distance cost, a collision cost, and a lateral boundary cost; The acceleration cost is the cost corresponding to the acceleration required for the vehicle to move from one point on the connecting edge to another point; The acceleration change rate cost is the cost corresponding to the acceleration change rate required for the vehicle to move from one point on the connecting edge to another point; The reference line distance cost is the cost corresponding to the distance from one point on the connecting edge to another point on the lane where the vehicle is located; The collision cost is the cost corresponding to whether the vehicle collides with an obstacle when moving from one point on the connecting edge to another point; The lateral boundary cost is the cost corresponding to the distance between the vehicle and the left and right lane boundaries when the vehicle moves from one point on the connecting edge to another point.

9. The device according to claim 6, characterized in that The preset constraints include lane boundary constraints and obstacle linear boundary constraints; The lane boundary constraint means that the vehicle can cross or press the dashed line, but cannot cross or press the solid line. The constraint conditions include obstacle linear boundary constraints, which refer to representing the obstacle boundary with a straight line. A collision is considered on one side of the straight line, and a safe collision-free state is considered on the other side of the straight line. The vehicle must travel on the other side of the straight line.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the vehicle trajectory planning method according to any one of claims 1 to 5 is implemented.

Images