Trajectory planning method, device, equipment and storage medium for autonomous driving vehicle

Abstract

The present application discloses a trajectory planning method, apparatus, device, and storage medium for an autonomous vehicle, and relates to the field of computer technology. The method comprises: obtaining driving data of the autonomous vehicle, driving data of a wrong-traffic obstacle, and road data; classifying the wrong-traffic obstacle using a preset scenario classification strategy based on the driving data of the autonomous vehicle, the driving data of the wrong-traffic obstacle, and the road data; in response to the wrong-traffic obstacle being a wrong-traffic obstacle in a first scenario, obtaining a first planned trajectory of the autonomous vehicle based on the driving data of the autonomous vehicle and the driving data of the wrong-traffic obstacle in the first scenario using a first planning model; in response to the wrong-traffic obstacle being a wrong-traffic obstacle in a second scenario, obtaining a second planned trajectory of the autonomous vehicle based on the driving data of the autonomous vehicle and the driving data of the wrong-traffic obstacle in the second scenario using a second planning model.

Description

Technical Field

[0001] The present application relates to the field of computer technology, specifically to technical fields such as intelligent transportation and autonomous driving, and in particular to a trajectory planning method, device, equipment and storage medium for an autonomous driving vehicle. Background Art

[0002] Generally, in driving scenarios filled with a large number of oncoming obstacles, such as unstructured roads or non-motorized vehicle lanes, due to the complex driving environment, autonomous vehicles need to effectively deal with oncoming obstacles to ensure vehicle driving safety.

[0003] Currently, most technical solutions for autonomous vehicles planning paths around oncoming obstacles use the SL-ST iterative optimization method to circumvent them. This effectively addresses the interaction issues with oncoming obstacles in most scenarios. However, in narrow road conditions, where lateral avoidance space is limited and longitudinal speed planning is more dependent, these technical solutions are relatively conservative in this scenario, resulting in poor smoothness and interactivity. Summary of the Invention

[0004] This application provides a trajectory planning method, apparatus, device, and storage medium for an autonomous vehicle, which optimizes the reliability of the planned trajectory of the vehicle in dealing with wrong-way obstacles. The technical solution is as follows:

[0005] In a first aspect, a trajectory planning method for an autonomous driving vehicle is provided, the method comprising:

[0006] Acquire driving data of the autonomous vehicle, driving data of the wrong-way obstacle, and road data;

[0007] Based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, the wrong-traffic obstacle is classified using a preset scene classification strategy;

[0008] In response to the wrong-way obstacle being a first-scenario wrong-way obstacle, obtaining a first planned trajectory of the autonomous driving vehicle using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the first-scenario wrong-way obstacle;

[0009] In response to the wrong-way obstacle being a second-scenario wrong-way obstacle, a second planned trajectory of the autonomous driving vehicle is obtained using a second planning model based on the driving data of the autonomous driving vehicle and the driving data of the second-scenario wrong-way obstacle.

[0010] In one possible implementation, obtaining a first planned trajectory of the autonomous driving vehicle using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the wrong-way obstacle in the first scenario includes:

[0011] Determining a motion vector of the autonomous driving vehicle using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the wrong-way obstacle in the first scenario;

[0012] Based on the driving data of the autonomous driving vehicle and the motion vector, a first planned trajectory of the autonomous driving vehicle is obtained.

[0013] In one possible implementation, determining the motion vector of the autonomous driving vehicle using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the wrong-way obstacle in the first scenario includes:

[0014] Determining a state vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle;

[0015] Inferring a state vector of the first-scenario wrong-traffic obstacle based on the driving data of the autonomous driving vehicle and the driving data of the first-scenario wrong-traffic obstacle;

[0016] Optimizing the objective function of the first planning model based on the state vector of the autonomous driving vehicle, the state vector of the wrong-way obstacle in the first scenario, and the constraints of the first planning model;

[0017] Based on the result of the optimization process, an action vector of the autonomous driving vehicle is determined.

[0018] In one possible implementation, the driving data of the second-scenario wrong-way obstacle includes a first predicted trajectory and a second predicted trajectory. Obtaining the second planned trajectory of the autonomous vehicle using a second planning model based on the driving data of the autonomous vehicle and the driving data of the second-scenario wrong-way obstacle includes:

[0019] determining a motion vector of the autonomous vehicle using a second planning model based on the driving data of the autonomous vehicle, the first predicted trajectory, and the second predicted trajectory;

[0020] Based on the driving data of the autonomous driving vehicle and the motion vector, a second planned trajectory of the autonomous driving vehicle is obtained.

[0021] In one possible implementation, determining the motion vector of the autonomous driving vehicle using a second planning model based on the driving data of the autonomous driving vehicle, the first predicted trajectory, and the second predicted trajectory includes:

[0022] Determining a state vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle;

[0023] determining a state vector of the first predicted trajectory based on the first predicted trajectory of the second-scenario wrong-way obstacle;

[0024] determining a state vector of the second predicted trajectory based on the second predicted trajectory of the wrong-way obstacle in the second scenario;

[0025] Optimizing an objective function of the second planning model based on the state vector of the autonomous driving vehicle, the state vector of the first predicted trajectory, the state vector of the second predicted trajectory, and the constraints of the second planning model;

[0026] Based on the result of the optimization process, an action vector of the autonomous driving vehicle is determined.

[0027] In one possible implementation, the classifying process of the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data using a preset scene classification strategy includes:

[0028] Determining whether the driving data of the wrong-way obstacle matches the driving data of the autonomous driving vehicle;

[0029] In response to the driving data of the wrong-traffic obstacle matching the driving data of the autonomous driving vehicle, determining that the wrong-traffic obstacle is a first-scenario wrong-traffic obstacle;

[0030] In response to the driving data of the wrong-traffic obstacle not matching the driving data of the autonomous driving vehicle, the wrong-traffic obstacle is determined to be a second-scenario wrong-traffic obstacle.

[0031] In one possible implementation, the classifying the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data using a preset scenario classification strategy further includes:

[0032] Determining a lane area corresponding to the autonomous driving vehicle based on the driving data and road data of the autonomous driving vehicle;

[0033] Determining whether the driving data of the wrong-way obstacle matches the lane area corresponding to the autonomous driving vehicle;

[0034] In response to the driving data of the wrong-traffic obstacle matching the lane area corresponding to the autonomous driving vehicle, determining that the wrong-traffic obstacle is a first-scenario wrong-traffic obstacle;

[0035] In response to the driving data of the wrong-traffic obstacle not matching the lane area corresponding to the autonomous driving vehicle, the wrong-traffic obstacle is determined to be a second-scenario wrong-traffic obstacle.

[0036] In one possible implementation, after obtaining the driving data of the autonomous driving vehicle, the driving data of the wrong-way obstacle, and the road data, the method further includes:

[0037] Based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle, a safety distance is calculated using a safety detection algorithm;

[0038] Calculating a relative distance between the autonomous driving vehicle and the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle;

[0039] When the relative distance and the safety distance meet preset conditions, the braking acceleration is calculated based on the safety distance, the driving data of the autonomous driving vehicle and the driving data of the reverse obstacle to control the driving of the autonomous driving vehicle based on the braking acceleration.

[0040] In a second aspect, a trajectory planning device for an autonomous driving vehicle is provided, the device comprising:

[0041] a first acquisition unit, configured to acquire driving data of the autonomous driving vehicle, driving data of a wrong-traffic obstacle, and road data;

[0042] a first classification unit, configured to classify the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, using a preset scenario classification strategy;

[0043] a first planning unit, configured to, in response to the wrong-way obstacle being a first-scenario wrong-way obstacle, obtain a first planned trajectory of the autonomous vehicle using a first planning model based on the driving data of the autonomous vehicle and the driving data of the first-scenario wrong-way obstacle;

[0044] The second planning unit is used to obtain a second planned trajectory of the autonomous driving vehicle by using a second planning model based on the driving data of the autonomous driving vehicle and the driving data of the second-scenario wrong-traffic obstacle in response to the wrong-traffic obstacle being a second-scenario wrong-traffic obstacle.

[0045] In a third aspect, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction, and the at least one instruction is loaded and executed by a processor to implement the method of the above-mentioned aspect and any possible implementation manner.

[0046] In a fourth aspect, an electronic device is provided, including:

[0047] at least one processor; and

[0048] a memory communicatively connected to the at least one processor; wherein,

[0049] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any possible implementation manner and the aspects described above.

[0050] In a fifth aspect, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the above-mentioned aspects and any possible implementation method.

[0051] In a sixth aspect, an autonomous driving vehicle is provided, comprising the electronic device as described above.

[0052] The beneficial effects of the technical solution provided by this application include at least:

[0053] It can be seen from the above technical solution that the embodiment of the present application can obtain the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, and then classify the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, using a preset scene classification strategy. In response to the wrong-traffic obstacle being a wrong-traffic obstacle of the first scene, based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle of the first scene, a first planning model is used to obtain a first planned trajectory of the autonomous driving vehicle. In response to the wrong-traffic obstacle being a wrong-traffic obstacle of the second scene, based on the wrong-traffic obstacle of the autonomous driving vehicle The second planning model is used to obtain the second planned trajectory of the autonomous driving vehicle based on the driving data and the driving data of the reverse obstacle in the second scenario. Since the reverse obstacles can be classified into scenarios first, the autonomous driving vehicle can then plan trajectories for reverse obstacles in different types of scenarios respectively, taking into account the bounded rational interactive game of reverse obstacles, it can effectively solve the problem of narrow road passage of autonomous driving vehicles while avoiding sudden braking and avoidance behaviors caused by overly conservative trajectory planning of autonomous driving vehicles, taking into account the robustness and exploratory nature of the trajectory planning of autonomous driving vehicles, and improving the continuity of the planned trajectory of autonomous driving vehicles, thereby ensuring the reliability of autonomous driving vehicles.

[0054] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0056] Figure 1 This is a flow chart of a trajectory planning method for an autonomous driving vehicle provided by one embodiment of the present application;

[0057] Figure 2 is a flowchart of a trajectory planning method for an autonomous driving vehicle provided by another embodiment of the present application;

[0058] Figure 3 is a schematic diagram of a first scenario in a trajectory planning method for an autonomous driving vehicle provided by another embodiment of the present application;

[0059] Figure 4 is a schematic diagram of a second scenario in a trajectory planning method for an autonomous driving vehicle provided by another embodiment of the present application;

[0060] Figure 5A and 5B to Figure 9A and 9B This is a schematic diagram of the planned trajectory and speed points for a first scenario of a retrograde obstacle in a trajectory planning method for an autonomous driving vehicle provided by another embodiment of the present application;

[0061] Figure 10A and 10B to Figure 14A and 14B This is a schematic diagram of a planning result of a trajectory planning method for an autonomous driving vehicle in response to a second scenario of a reverse obstacle, provided by another embodiment of the present application;

[0062] Figure 15 This is a structural block diagram of a trajectory planning device for an autonomous driving vehicle provided in yet another embodiment of the present application;

[0063] Figure 16 This is a block diagram of an electronic device used to implement the trajectory planning method for an autonomous driving vehicle according to an embodiment of the present application. DETAILED DESCRIPTION

[0064] The following description of exemplary embodiments of the present application is made in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0065] Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0066] It should be noted that the terminal devices involved in the embodiments of the present application may include but are not limited to mobile phones, personal digital assistants (PDAs), wireless handheld devices, tablet computers and other smart devices; display devices may include but are not limited to personal computers, televisions and other devices with display functions.

[0067] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the related objects are in an "or" relationship.

[0068] Generally, in the actual operation of autonomous driving, the driving environment is relatively complex. For example, the terminal delivery scenes of autonomous driving vehicles are mostly located on unstructured roads or non-motorized vehicle lanes. This type of scene is full of a large number of reverse obstacles, such as delivery personnel, non-motorized vehicles, etc.

[0069] For handling oncoming obstacles, related technologies often adopt or optimize methods from open-source solutions. For example, the SL-ST iterative optimization method, which bypasses oncoming obstacles, effectively solves the interaction problem of oncoming obstacles in most scenarios. However, autonomous vehicle delivery scenarios often have narrow roads with limited lateral avoidance space, requiring a greater reliance on longitudinal speed planning. However, the ST graph search method is relatively conservative in this scenario, resulting in poor smoothness and interactivity.

[0070] Therefore, there is an urgent need for a trajectory planning method for autonomous vehicles that can enable effective interaction between the vehicle and oncoming obstacles, thereby improving the continuity and reliability of the vehicle's planned trajectory.

[0071] Please refer to Figure 1, which shows a flow chart of a trajectory planning method for an autonomous vehicle provided by one embodiment of the present application. The trajectory planning method for an autonomous vehicle may specifically include:

[0072] Step 101: Acquire driving data of the autonomous driving vehicle, driving data of the wrong-traffic obstacle, and road data.

[0073] Step 102: Based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, the wrong-traffic obstacle is classified using a preset scene classification strategy.

[0074] Step 103: In response to the wrong-way obstacle being a first-scenario wrong-way obstacle, a first planned trajectory of the autonomous driving vehicle is obtained using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the first-scenario wrong-way obstacle.

[0075] Step 104: In response to the wrong-way obstacle being a wrong-way obstacle in the second scenario, a second planned trajectory of the autonomous driving vehicle is obtained using a second planning model based on the driving data of the autonomous driving vehicle and the driving data of the wrong-way obstacle in the second scenario.

[0076] At this point, the driving of the autonomous driving vehicle can be controlled based on the first planned trajectory of the autonomous driving vehicle, or the second planned trajectory of the autonomous driving vehicle.

[0077] It should be noted that the autonomous vehicle here refers to the ego vehicle. The driving data of the wrong-way obstacle may include driving state data such as the location and speed of the wrong-way obstacle perceived by the ego vehicle, as well as the predicted trajectory of the wrong-way obstacle obtained by the autonomous vehicle's prediction module based on the driving state data.

[0078] It should be noted that the road data may be data of the road area where the autonomous vehicle is currently traveling. The road data may include, but is not limited to, lane markings, lane lines, lane width, etc.

[0079] It should be noted that the first scenario's reverse-traffic obstacle may be a reverse-traffic obstacle on the planned path of the autonomous vehicle, while the second scenario's reverse-traffic obstacle may be a reverse-traffic obstacle that may conflict with the autonomous vehicle in the future.

[0080] It should be noted that part or all of the execution entities of steps 101 to 104 may be applications located in the local terminal, or may be functional units such as plug-ins or software development kits (SDKs) provided in the applications located in the local terminal, or may be processing engines located in network-side servers, or may be distributed systems located on the network side, for example, processing engines or distributed systems in autonomous driving platforms on the network side, etc. This embodiment does not specifically limit this.

[0081] It is understandable that the application may be a native program (nativeApp) installed on the local terminal, or may be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0082] In this way, by first classifying the scenarios of the reverse obstacles, the autonomous driving vehicle can then plan the trajectory for the reverse obstacles in different types of scenarios respectively. Taking into account the bounded rational interactive game of the reverse obstacles, it can effectively solve the problem of the autonomous driving vehicle passing through narrow roads while avoiding the sudden braking and avoidance behavior caused by the overly conservative trajectory planning of the autonomous driving vehicle. It takes into account the robustness and exploratory nature of the trajectory planning of the autonomous driving vehicle, improves the continuity of the trajectory of the autonomous driving vehicle, and thus ensures the reliability of the autonomous driving vehicle.

[0083] Optionally, in one possible implementation of this embodiment, in step 103, first, a motion vector of the autonomous vehicle is determined using a first planning model based on the driving data of the autonomous vehicle and the driving data of the wrong-way obstacle in the first scenario. Then, a first planned trajectory of the autonomous vehicle can be obtained based on the driving data of the autonomous vehicle and the motion vector.

[0084] In this implementation, the motion vector of the autonomous driving vehicle may be the acceleration corresponding to the trajectory point planned by the autonomous driving vehicle.

[0085] In a specific implementation process of this implementation method, first, the state vector of the autonomous driving vehicle can be determined based on the driving data of the autonomous driving vehicle. Secondly, the state vector of the first scenario's reverse-traffic obstacle can be inferred based on the driving data of the autonomous driving vehicle and the driving data of the first scenario's reverse-traffic obstacle. Thirdly, the objective function of the first planning model can be optimized based on the state vector of the autonomous driving vehicle, the state vector of the first scenario's reverse-traffic obstacle, and the constraints of the first planning model. Thirdly, the action vector of the autonomous driving vehicle can be determined based on the result of the optimization process.

[0086] In this implementation, the autonomous vehicle's driving data may include the autonomous vehicle's posture, speed, planned path, and other information. The posture of the autonomous vehicle includes its position and orientation. The driving data of the wrong-way obstacle may include the posture, speed, and other data of the wrong-way obstacle as perceived by the autonomous vehicle. The posture of the wrong-way obstacle includes its position and orientation.

[0087] One scenario of this specific implementation process is that the position, speed and planned path of the autonomous driving vehicle can be used to determine the position, speed and acceleration of each position point on the planned path of the autonomous driving vehicle, and the position, speed and acceleration of each position point on the planned path of the autonomous driving vehicle can be used as the state vector of the autonomous driving vehicle.

[0088] Another scenario of this specific implementation process is that based on the position of the autonomous driving vehicle, the planned path and the position of the retrograde obstacle in the first scenario, the motion path of the retrograde obstacle in the first scenario is inferred, and then based on the motion path of the retrograde obstacle in the first scenario, the position, velocity and acceleration of each position point on the motion path of the retrograde obstacle in the first scenario can be determined, and the position, velocity and acceleration of each position point on the motion path are used as the state vector of the retrograde obstacle in the first scenario.

[0089] In the specific implementation process, based on the current position of the autonomous driving vehicle and the current position of the retrograde obstacle in the first scene, the movement path of the retrograde obstacle in the first scene is reversely deduced according to the planned path of the autonomous driving vehicle.

[0090] In this implementation, the first planning model may be a Model Predictive Control Game (MPCG) model.

[0091] In another specific implementation, the objective function of the first planning model can be optimized based on the state vector of the autonomous vehicle, the state vector of the wrong-way obstacle in the first scenario, and the constraints of the model predictive control game model. The action vector of the autonomous vehicle can then be obtained based on the optimization result.

[0092] Another implementation of this process involves first obtaining the probability of acceleration or deceleration for the first wrong-way obstacle in the current frame. Then, based on the state vector of the autonomous vehicle, the probability of acceleration or deceleration for the first wrong-way obstacle, the state vector of the first wrong-way obstacle, and the constraints of the model predictive control game model, the objective function of the first planning model is optimized. Finally, based on the optimization results, the action vector of the autonomous vehicle is obtained.

[0093] Exemplarily, the objective function and constraints of the first planning model may be as shown in formula (7).

[0094] In another specific implementation process of this implementation method, the planned trajectory, position, speed and motion vector of the autonomous driving vehicle can be fused to obtain a first planned trajectory of the autonomous driving vehicle to deal with a reverse obstacle in a first scenario.

[0095] In this way, when it is determined that the reverse obstacle that the autonomous driving vehicle is dealing with is the obstacle of the first scene, the first planning model can be used to calculate the motion vector of the autonomous driving vehicle in dealing with the reverse obstacle in the first scene based on the driving data of the autonomous driving vehicle and the driving data of the reverse obstacle in the first scene, and the driving data and motion vector of the autonomous driving vehicle can be used to obtain a more accurate first planning trajectory of the autonomous driving vehicle in dealing with the reverse obstacle in the first scene, further improving the continuity of the planned trajectory of the autonomous driving vehicle, thereby ensuring the reliability of the autonomous driving vehicle's driving.

[0096] It should be noted that the various specific implementation processes provided in this implementation can be combined with each other to implement the steps in the trajectory planning method for an autonomous vehicle of this embodiment. For a detailed description, please refer to the relevant content of this implementation and will not be repeated here.

[0097] Optionally, in one possible implementation of this embodiment, the driving data of the wrong-way obstacle in the second scenario may include a first predicted trajectory and a second predicted trajectory. In step 104, first, a motion vector of the autonomous vehicle may be determined using a second planning model based on the driving data of the autonomous vehicle, the first predicted trajectory, and the second predicted trajectory. Second, a second planned trajectory of the autonomous vehicle may be obtained based on the driving data of the autonomous vehicle and the motion vector.

[0098] In this implementation, at least one of the first predicted trajectory and the second predicted trajectory of the second scenario's wrong-way obstacle conflicts with the planned path of the autonomous driving vehicle.

[0099] In this implementation, the second planning model may be a contigency model predictive control model (CMPC).

[0100] In a specific implementation process of this implementation method, first, the state vector of the autonomous driving vehicle can be determined based on the driving data of the autonomous driving vehicle. Secondly, the state vector of the first predicted trajectory can be determined based on the first predicted trajectory of the second scenario's retrograde obstacle. Thirdly, the state vector of the second predicted trajectory can be determined based on the second predicted trajectory of the second scenario's retrograde obstacle. Thirdly, the objective function of the second planning model can be optimized based on the state vector of the autonomous driving vehicle, the state vector of the first predicted trajectory, the state vector of the second predicted trajectory, and the constraints of the second planning model. Thirdly, based on the result of the optimization process, the action vector of the autonomous driving vehicle is determined.

[0101] In this implementation, the state vector of the autonomous vehicle may include the position, velocity, and acceleration of each location point on the planned path of the autonomous vehicle. The state vector of the first predicted trajectory may include the position, velocity, and acceleration of each trajectory point on the first predicted trajectory. The state vector of the second predicted trajectory may include the position, velocity, and acceleration of each trajectory point on the second predicted trajectory.

[0102] In one specific implementation, the probability of a conflict between the first predicted trajectory of the current frame and the planned path of the autonomous vehicle can be obtained. Based on the state vector of the autonomous vehicle, the probability of a conflict between the first predicted trajectory of the current frame and the planned path of the autonomous vehicle, the state vector of the first predicted trajectory, the state vector of the second predicted trajectory, and the constraints of the second planning model, the objective function of the second planning model can be optimized. Finally, based on the optimization results, the action vector of the autonomous vehicle is determined.

[0103] Exemplarily, the objective function and constraints of the second planning model may be as shown in formula (11).

[0104] In another specific implementation process of this implementation method, it can be determined whether the position relationship between the autonomous driving vehicle and the positional relationship of the reverse obstacle in the second scene meets the preset relationship conditions. If the position relationship between the autonomous driving vehicle and the reverse obstacle in the second scene meets the preset relationship conditions, the motion vector of the autonomous driving vehicle can be determined based on the driving data of the autonomous driving vehicle, the first predicted trajectory and the second predicted trajectory using the second planning model, and then the second planned trajectory of the autonomous driving vehicle can be obtained based on the driving data of the autonomous driving vehicle and the motion vector.

[0105] Here, the preset relationship condition may be that the second scenario wrong-traffic obstacle is located in front of the autonomous vehicle. For example, when the difference between the position of the autonomous vehicle and the position of the second scenario wrong-traffic obstacle is less than zero, it can be indicated that the second scenario wrong-traffic obstacle is located in front of the autonomous vehicle.

[0106] In this way, when it is determined that the reverse obstacle that the autonomous driving vehicle is dealing with is an obstacle in the second scenario, the second planning model can be used to calculate the motion vector of the autonomous driving vehicle dealing with the reverse obstacle in the second scenario based on the driving data of the autonomous driving vehicle and the first predicted trajectory and the second predicted trajectory of the reverse obstacle in the second scenario. The driving data and the motion vector of the autonomous driving vehicle can be used to obtain a more accurate second planning trajectory of the autonomous driving vehicle dealing with the reverse obstacle in the second scenario, further improving the continuity of the planned trajectory of the autonomous driving vehicle in this scenario, thereby ensuring the reliability of the autonomous driving vehicle's driving.

[0107] It should be noted that the various specific implementation processes provided in this implementation can be combined with the aforementioned implementations to implement the steps in the trajectory planning method for an autonomous vehicle of this embodiment. For a detailed description, please refer to the relevant content of this implementation and will not be repeated here.

[0108] Optionally, in a possible implementation of this embodiment, in step 102, first, it may be determined whether the driving data of the wrong-traffic obstacle matches the driving data of the autonomous vehicle. Second, in response to the driving data of the wrong-traffic obstacle matching the driving data of the autonomous vehicle, the wrong-traffic obstacle may be determined to be a wrong-traffic obstacle in the first scenario. Third, in response to the driving data of the wrong-traffic obstacle not matching the driving data of the autonomous vehicle, the wrong-traffic obstacle may be determined to be a wrong-traffic obstacle in the second scenario.

[0109] In a specific implementation process of this implementation method, first, the position of the wrong-way obstacle and the planned path of the autonomous driving vehicle can be matched. When the position of the wrong-way obstacle is on the planned path of the autonomous driving vehicle, the wrong-way obstacle can be classified as a first-scenario wrong-way obstacle; when the position of the wrong-way obstacle is not on the planned path of the autonomous driving vehicle, the wrong-way obstacle can be classified as a second-scenario wrong-way obstacle.

[0110] It is understood that, here, the position of the wrong-way obstacle may be the current position of the wrong-way obstacle in the current planning frame, and the planned path of the autonomous driving vehicle may be the current planned path of the autonomous driving vehicle in the current planning frame.

[0111] Optionally, in a possible implementation of this embodiment, in step 102, further, based on the driving data and road data of the autonomous driving vehicle, the lane area corresponding to the autonomous driving vehicle can be determined, and then it can be determined whether the driving data of the wrong-moving obstacle matches the lane area corresponding to the autonomous driving vehicle. In response to the driving data of the wrong-moving obstacle matching the lane area corresponding to the autonomous driving vehicle, the wrong-moving obstacle is determined to be a first-scenario wrong-moving obstacle. In response to the driving data of the wrong-moving obstacle not matching the lane area corresponding to the autonomous driving vehicle, the wrong-moving obstacle is determined to be a second-scenario wrong-moving obstacle.

[0112] In a specific implementation process of this implementation method, first, the lane area of the lane where the autonomous vehicle is located can be determined based on the position and planned path of the autonomous vehicle, and the lane lines and lane widths in the road data. Secondly, the position of the wrong-way obstacle can be matched with the lane area corresponding to the autonomous vehicle. Thirdly, if the position of the wrong-way obstacle is within the lane area corresponding to the autonomous vehicle, the wrong-way obstacle can be classified as a first-scenario wrong-way obstacle; if the position of the wrong-way obstacle is not within the lane area corresponding to the autonomous vehicle, the wrong-way obstacle can be classified as a second-scenario wrong-way obstacle.

[0113] In another specific implementation process of this implementation method, the first-scenario reverse obstacles determined by the above two implementation methods can be merged to obtain all the first-scenario reverse obstacles perceived by the autonomous driving vehicle. The second-scenario reverse obstacles determined by the above two implementation methods can also be merged to obtain all the second-scenario reverse obstacles perceived by the autonomous driving vehicle.

[0114] In this way, by classifying the driving scenarios of the wrong-way obstacles, it is possible to plan the trajectory of the autonomous driving vehicle to deal with wrong-way obstacles of different scene types, which can further improve the continuity of the planned trajectory and thus improve the driving reliability of the autonomous driving vehicle.

[0115] It should be noted that the various specific implementation processes provided in this implementation can be combined with the aforementioned implementations to implement the steps in the trajectory planning method for an autonomous vehicle of this embodiment. For a detailed description, please refer to the relevant content of this implementation and will not be repeated here.

[0116] Optionally, in a possible implementation of this embodiment, after step 101, first, a safety distance can be calculated based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle using a safety detection algorithm. Secondly, a relative distance between the autonomous driving vehicle and the wrong-traffic obstacle can be calculated based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle. Thirdly, when the relative distance and the safety distance meet preset conditions, a braking acceleration is calculated based on the safety distance, the driving data of the autonomous driving vehicle, and the driving data of the wrong-traffic obstacle, so as to control the driving of the autonomous driving vehicle based on the braking acceleration.

[0117] In this implementation, the preset condition may include that the relative distance is less than the safety distance.

[0118] In another specific implementation of this method, first, the relative distance between the autonomous driving vehicle and the oncoming obstacle can be calculated based on the position of the autonomous driving vehicle and the position of the oncoming obstacle. Next, if it is determined that the relative distance is less than the safe distance, the braking acceleration is calculated based on the safe distance, the speed of the autonomous driving vehicle, and the speed of the oncoming obstacle. Next, the driving of the autonomous driving vehicle can be controlled based on the braking acceleration. Again, if it is determined that the relative distance is not less than the safe distance, step 102 can be continued.

[0119] It is understandable that before planning the trajectory of an autonomous vehicle, it is possible to first determine whether the current driving situation of the autonomous vehicle is a dangerous driving situation, that is, to determine whether the distance between the autonomous vehicle and the oncoming obstacle meets the safety distance. If it is determined that the autonomous vehicle is in a dangerous driving situation, the braking acceleration can be directly calculated, and the autonomous vehicle can be braked based on the braking acceleration.

[0120] Furthermore, after planning the trajectory of the autonomous vehicle, it is also possible to determine whether the current driving situation of the autonomous vehicle is dangerous, that is, to determine whether the distance between the autonomous vehicle and the oncoming obstacle meets the safe distance. If it is determined that the autonomous vehicle is in a dangerous driving situation, the braking acceleration can be directly calculated, and the autonomous vehicle can be braked based on the braking acceleration. If it is determined that the autonomous vehicle is in a safe driving situation, the autonomous vehicle can be controlled based on the first planned trajectory or the second planned trajectory.

[0121] In this way, the safety of the autonomous driving vehicle can be further guaranteed by performing safety detection and processing on the autonomous driving vehicle.

[0122] It should be noted that the various specific implementation processes provided in this implementation can be combined with the aforementioned implementations to implement the steps in the trajectory planning method for an autonomous vehicle of this embodiment. For a detailed description, please refer to the relevant content of this implementation and will not be repeated here.

[0123] In order to better understand the method of the embodiment of the present application, the method of the embodiment of the present application is described below with reference to the accompanying drawings and specific application scenarios.

[0124] Figure 2 This is a flow chart of a trajectory planning method for an autonomous driving vehicle provided by another embodiment of the present application. Figure 2 shown.

[0125] Step 201: Acquire driving data of the autonomous driving vehicle, driving data of the wrong-traffic obstacle, and road data.

[0126] In this embodiment, the driving data of the autonomous driving vehicle may include but is not limited to posture data, speed, planned path, etc. The posture data of the autonomous driving vehicle may include position data and orientation data.

[0127] In this embodiment, the driving data for wrong-way obstacle information may include data sensed by the autonomous vehicle and related data obtained by the autonomous vehicle after processing the sensed data. The driving data for wrong-way obstacle information may include, but is not limited to, position data, speed, and predicted trajectory. The predicted trajectory may be multimodal, meaning that there may be multiple predicted trajectories.

[0128] In this embodiment, the road data may include but is not limited to lane lines, lane markings, lane width, etc.

[0129] Step 202: Based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, the wrong-traffic obstacle is classified using a preset scene classification strategy.

[0130] In this embodiment, it is determined whether the driving data of the wrong-traffic obstacle matches the driving data of the autonomous driving vehicle. If the driving data of the wrong-traffic obstacle matches the driving data of the autonomous driving vehicle, the wrong-traffic obstacle is classified as a first-scenario wrong-traffic obstacle; if the driving data of the wrong-traffic obstacle does not match the driving data of the autonomous driving vehicle, the wrong-traffic obstacle is classified as a second-scenario wrong-traffic obstacle.

[0131] Preferably, the position data of the wrong-way obstacle and the planned path of the autonomous driving vehicle are matched and processed. When the position data of the wrong-way obstacle is on the planned path of the autonomous driving vehicle, the wrong-way obstacle can be classified as a first-scenario wrong-way obstacle; when the position data of the wrong-way obstacle is not on the planned path of the autonomous driving vehicle, the wrong-way obstacle can be classified as a second-scenario wrong-way obstacle.

[0132] In this embodiment, the lane area corresponding to the autonomous driving vehicle can also be determined based on the driving data and road data of the autonomous driving vehicle, and then it can be determined whether the driving data of the wrong-moving obstacle matches the lane area corresponding to the autonomous driving vehicle. When the driving data of the wrong-moving obstacle matches the lane area corresponding to the autonomous driving vehicle, the wrong-moving obstacle can be classified as a first-scenario wrong-moving obstacle; when the driving data of the wrong-moving obstacle does not match the lane area corresponding to the autonomous driving vehicle, the wrong-moving obstacle can be classified as a second-scenario wrong-moving obstacle.

[0133] Preferably, first, the lane area of the lane where the autonomous vehicle is located can be determined based on the autonomous vehicle's position data and planned path, as well as the lane lines and lane widths in the road data. Second, the position data of the wrong-way obstacle can be matched with the lane area corresponding to the autonomous vehicle. Third, if the position data of the wrong-way obstacle is within the lane area corresponding to the autonomous vehicle, the wrong-way obstacle can be classified as a wrong-way obstacle in the first scenario; if the position data of the wrong-way obstacle is not within the lane area corresponding to the autonomous vehicle, the wrong-way obstacle can be classified as a wrong-way obstacle in the second scenario.

[0134] Furthermore, the first-scenario reverse obstacles determined by the above two methods can be combined to obtain all the first-scenario reverse obstacles perceived by the autonomous driving vehicle. The second-scenario reverse obstacles determined by the above two methods can also be combined to obtain all the second-scenario reverse obstacles perceived by the autonomous driving vehicle.

[0135] Here, the first-scenario wrong-way obstacles may include wrong-way obstacles on the planned path of the autonomous vehicle and wrong-way obstacles in the lane where the autonomous vehicle is located. The first-scenario wrong-way obstacles may be wrong-way obstacles that will definitely cause the autonomous vehicle to collide. The first-scenario wrong-way obstacles may be wrong-way obstacles located on the path or lane of the autonomous vehicle.

[0136] The second-scenario oncoming obstacle can be an oncoming obstacle that may conflict with the autonomous vehicle in the future. The second-scenario oncoming obstacle can be an obstacle that is traveling in the opposite direction, outside the path or lane of the autonomous vehicle, but has the intention of intruding.

[0137] It is understandable that both the first and second scenarios are driving environments where the ego vehicle needs to deal with a wrong-traffic obstacle. The first scenario may refer to a scenario where the road is relatively narrow, the obstacle is within the ego vehicle's driving path, and there is a conflict between wrong-traffic and the ego vehicle's driving. The wrong-traffic obstacle in the first scenario may be a wrong-traffic obstacle that the ego vehicle needs to deal with in the first scenario. The second scenario may refer to a scenario where the obstacle is outside the ego vehicle's path, but its future predicted trajectory has the potential to intrude into the ego vehicle's path, and there is a conflict between wrong-traffic and the ego vehicle's driving. The wrong-traffic obstacle in the second scenario may be a wrong-traffic obstacle that the ego vehicle needs to deal with in the second scenario.

[0138] Step 203: When the wrong-traffic obstacle is a wrong-traffic obstacle in the first scenario, the motion vector of the automatic driving vehicle is calculated using the first planning model based on the driving data of the automatic driving vehicle and the driving data of the wrong-traffic obstacle in the first scenario.

[0139] Step 204: Obtain a first planned trajectory of the autonomous driving vehicle based on the driving data and motion vector of the autonomous driving vehicle.

[0140] In this embodiment, the first planning model may be a model predictive control game model.

[0141] Preferably, when the reverse-traffic obstacle is a reverse-traffic obstacle in the first scenario, i.e., the autonomous vehicle is in the driving environment of the first scenario, first, the position data and speed of the reverse-traffic obstacle in the first scenario can be extracted from the driving data of the reverse-traffic obstacle in the first scenario, and the planned path of the autonomous vehicle can be extracted from the driving data of the autonomous vehicle. Secondly, based on the position data of the reverse-traffic obstacle in the first scenario, the motion path of the reverse-traffic obstacle is inferred in the opposite direction of the planned path of the autonomous vehicle. Thirdly, based on the position data, speed, and motion path of the reverse-traffic obstacle, a state vector of the reverse-traffic obstacle is obtained. Thirdly, the probability of the reverse-traffic obstacle accelerating / decelerating in the previous frame of the current planning frame is obtained. Based on the probability of the reverse-traffic obstacle accelerating / decelerating in the previous frame, a preset step size, and a preset maximum acceleration or deceleration, a preset first probability estimation algorithm is used to calculate the probability of the reverse-traffic obstacle accelerating / decelerating in the current planning frame. Thirdly, based on the position data, speed, and planned path of the autonomous vehicle, a state vector of the autonomous vehicle is obtained. Next, based on the state vector of the autonomous vehicle, the state vector of the wrong-way obstacle in the first scenario, and the constraints of the first planning model, the objective function of the first planning model can be optimized and solved to calculate the motion vector of the autonomous vehicle. Finally, based on the driving data and motion vector of the autonomous vehicle, a first planned trajectory for the autonomous vehicle to deal with the wrong-way obstacle in the first scenario is obtained.

[0142] In this implementation, Figure 3Schematic diagram of the first scenario in the trajectory planning method of an autonomous driving vehicle provided by another embodiment of the present application. Figure 3 As shown in the figure, ego represents the autonomous vehicle, v1 represents the speed of the autonomous vehicle, i.e., the speed of the ego vehicle. obs represents the opposite-traffic obstacle, i.e., the opposite-traffic obstacle in the first scenario. v2 represents the speed of the opposite-traffic obstacle in the first scenario. s represents the distance between the autonomous vehicle and the opposite-traffic obstacle in the first scenario.

[0143] For example, for the first type of wrong-way scenario, i.e., the first scenario, the model predictive control game model is used to plan the trajectory of the ego vehicle to deal with the wrong-way obstacle in the first scenario. In the process of determining the objective function of the model optimization problem of the model predictive control game model, it is assumed that the current state vector of the ego vehicle is , To obtain the current position of the vehicle, To obtain the current speed of the vehicle, To obtain the current acceleration of the vehicle, the state transition equation is as follows (1):

[0144] (1)

[0145] in, is the state vector, is the action vector, and is the coefficient matrix, which can be expressed as:

[0146]

[0147]

[0148] Assume that the reverse obstacle is currently located in front of the vehicle Distance position, the initial state of the retrograde obstacle is defined as , is the initial position of the retrograde obstacle, is the initial velocity of the reverse obstacle, is the initial acceleration of the reverse obstacle. The state transfer equation of the reverse obstacle is consistent with that of the ego vehicle, as shown in the following formula (2):

[0149] (2)

[0150] The coefficient matrix is expressed as:

[0151]

[0152]

[0153] Furthermore, in the Stackelberg game model, the ego vehicle (ego) is the leader, the opposite obstacle vehicle (obs) is the follower, and the follower action semantic types are divided into acceleration (accel) and deceleration (decel). The ego vehicle, as the leader, responds to the opposite obstacle's response to the ego vehicle, and the action semantic types can be divided into acceleration and deceleration. It can be expressed as , It can be expressed as .

[0154] Therefore, the state vector set of the ego vehicle and the oncoming obstacle includes , the action vector set includes .

[0155] It is understandable that the following objective functions and constraint design parts involving the same acceleration and deceleration actions of the ego vehicle (leader) and the oncoming obstacle vehicle (follower) are no longer expressed separately.

[0156] On the one hand, for the ego vehicle, the objective function of the ego vehicle can be defined as formula (3):

[0157] (3)

[0158] in, It can represent the state cost, It can represent the cost of an action. and is the weight parameter matrix. For your car The reference state at the moment, the reference state can be defined as:

[0159]

[0160] in, The speed limit of the road where the vehicle is currently located. Plan the position of the trajectory point for the previous frame, with an expected acceleration of 0.

[0161] The constraints of the ego vehicle can include equality constraints and inequality constraints. Equality constraints include initial state constraints and state transition constraints, which can be defined as:

[0162]

[0163]

[0164] Inequality constraints can include collision constraints and reversing constraints, which can be defined as:

[0165]

[0166]

[0167] in, It can be a preset minimum safety distance.

[0168] The bound constraints of variables can be defined as:

[0169]

[0170]

[0171]

[0172]

[0173] in, The jump of the car, that is, the acceleration.

[0174] Therefore, the above optimization problem of the vehicle can be summarized as an optimal control problem, which can be expressed as formula (4):

[0175] (4)

[0176] st

[0177] ,

[0178]

[0179] in, and are the follower's reaction state vector and reaction action vector for the leader's current possible actions, and is the leader’s response vector to the follower’s reaction state vector and action. is the equality constraint function above, is the above inequality constraint.

[0180] On the other hand, for the vehicle with the opposite direction to the obstacle, the objective function of the vehicle with the opposite direction to the obstacle can be defined as formula (5):

[0181] (5)

[0182] Among them, the reverse obstacle vehicle and the ego vehicle can share the weight parameter matrix and , that is, assuming that the opposite-travel obstacle vehicle and the ego vehicle have the same cost evaluation method. The reference state of the opposite-travel obstacle vehicle It can be defined as:

[0183]

[0184] in, The speed limit of the road where the vehicle with the wrong-way obstacle is located. Plan the position for the previous frame, with an expected acceleration of 0.

[0185] The constraints of vehicles with obstacles on the other side of the road can also include equality constraints and inequality constraints. Equality constraints include initial state constraints and state transition constraints, which can be defined as:

[0186]

[0187]

[0188] Inequality constraints include collision constraints and backing constraints, which can be defined as:

[0189]

[0190]

[0191]

[0192] in, and It can be a preset minimum safety distance.

[0193] Variable bound constraints can be defined as:

[0194]

[0195]

[0196]

[0197]

[0198] in, The jump of the obstacle can be reversed, that is, the acceleration.

[0199] Therefore, the above optimization problem of the retrograde obstacle is summarized as an optimal control problem, which can be expressed as formula (6):

[0200] (6)

[0201] st

[0202] ,

[0203]

[0204] in, The state vector that can be the follower's guess at the action taken by the leader, and The state vector and action vector of the follower’s response to this guess can be expressed as, and are the above equality constraints and inequality constraints respectively.

[0205] Here, the leader vehicle and the follower obstacle each represent their own optimal control problems. However, the variables of both vehicles are nested, and the optimization problem has duality. Therefore, to solve a nested optimization problem, the optimal control problem of the follower can be expressed as the following Lagrangian function using the Karush-Kuhn-Tucker (KKT) condition:

[0206]

[0207] in, is the Lagrangian multiplier corresponding to the equality constraint, is the KKT multiplier corresponding to the inequality constraint. KKT conditions can include:

[0208] ,

[0209]

[0210]

[0211]

[0212]

[0213] Therefore, the overall model optimization problem, that is, the objective function of the first planning model, can be defined as formula (7):

[0214] (7)

[0215] st

[0216]

[0217]

[0218]

[0219]

[0220]

[0221]

[0222]

[0223]

[0224]

[0225]

[0226]

[0227]

[0228]

[0229]

[0230]

[0231] in, The prior probability of the human driver taking a deceleration reaction can be set based on the probability of deceleration when encountering a reverse obstacle, 1- This is the prior probability set based on the human driver's acceleration response, that is, the probability of accelerating in the face of a reverse obstacle. is the leader’s objective function. is the leader’s equality constraint function, is the leader’s inequality constraint, and are the equality constraints and inequality constraints of the followers respectively. Can be a strain constraint.

[0232] In this way, by adding strain constraints, the continuity of the ego vehicle's reaction process can be guaranteed, and the planning results of the ego vehicle's acceleration or deceleration decision can be better integrated.

[0233] In this embodiment, in addition, the prior probability can be set based on the intention of the human driver to take action, and the intention of the human driver to take action can be summarized and statistically obtained through historical behavior. The prior probability of each planning frame is updated, and the prior probability of the current frame is The estimation is defined as formula (8):

[0234] , ;

[0235] , (8)

[0236] in, The upper limit of the probability can be set, preferably, , The prior probability of acceleration or deceleration can be taken for the reverse obstacle in the previous frame. Can be the acceleration state of the reverse obstacle, The maximum possible step size can be set, preferably, . and It can be the maximum acceleration / deceleration of the retrograde obstacle, preferably the maximum acceleration / deceleration is ,

[0237] Step 205: When the wrong-traffic obstacle is a wrong-traffic obstacle in the second scenario, determine whether the positional relationship between the wrong-traffic obstacle in the second scenario and the autonomous driving vehicle meets a preset relationship condition.

[0238] In this embodiment, the preset relationship condition may be that the wrong-way obstacle is located in front of the autonomous driving vehicle.

[0239] Specifically, conflict judgment can be performed based on the position data of the autonomous driving vehicle and the multimodal predicted trajectory of the planned path and the reverse obstacle.

[0240] Step 206: When the positional relationship between the second-scenario wrong-traffic obstacle and the autonomous driving vehicle satisfies a preset relationship condition, the motion vector of the autonomous driving vehicle is calculated using the second planning model based on the driving data of the autonomous driving vehicle, the first predicted trajectory and the second predicted trajectory of the second-scenario wrong-traffic obstacle.

[0241] In this embodiment, at least one of the first predicted trajectory and the second predicted trajectory of the wrong-way obstacle in the second scenario overlaps with the planned path of the autonomous driving vehicle.

[0242] Step 207: Obtain a second planned trajectory of the autonomous driving vehicle based on the driving data and motion vector of the autonomous driving vehicle.

[0243] In this embodiment, the second planning model may be a contingency model predictive control model.

[0244] In this embodiment, a strain-based model predictive control model can be constructed to plan the response trajectory of the vehicle to the second scenario's reverse obstacle based on the driving data of the autonomous driving vehicle and the first predicted trajectory and the second predicted trajectory of the reverse obstacle in the second scenario.

[0245] Optionally, first, based on the position data, speed and planned path of the autonomous driving vehicle, the state vector of the autonomous driving vehicle is determined. The state vector of the autonomous driving vehicle may include the position, speed and acceleration of each position point on the planned path. Secondly, the state vector of the first predicted trajectory of the second scenario's reverse obstacle may be determined. The state vector of the first predicted trajectory may include the position, speed and acceleration of each trajectory point on the first predicted trajectory. Thirdly, the state vector of the second predicted trajectory of the second scenario's reverse obstacle is determined. The state vector of the second predicted trajectory may include the position, speed and acceleration of each trajectory point on the second predicted trajectory. Thirdly, based on the state vector of the autonomous driving vehicle, the state vector of the first predicted trajectory, the state vector of the second predicted trajectory, and the constraints of the second planning model, the objective function of the second planning model is optimized and solved to calculate the action vector of the autonomous driving vehicle. Thirdly, based on the driving data and action vector of the autonomous driving vehicle, the second planned trajectory of the autonomous driving vehicle is obtained.

[0246] Furthermore, the probability of a wrong-way obstacle entering the planned path of the autonomous driving vehicle can be obtained, and then based on the state vector of the autonomous driving vehicle, the state vector of the first predicted trajectory, the state vector of the second predicted trajectory, the constraints of the second planning model, and the probability of a wrong-way obstacle entering the planned path of the autonomous driving vehicle, the objective function of the second planning model can be optimized and solved to calculate the action vector of the autonomous driving vehicle.

[0247] The probability of a wrong-way obstacle entering the planned path of the autonomous driving vehicle may be a probability that a predicted trajectory of the wrong-way obstacle conflicts with the planned path of the autonomous driving vehicle.

[0248] In this implementation, Figure 4 Schematic diagram of the second scenario in the trajectory planning method of an autonomous driving vehicle provided by another embodiment of the present application. Figure 4 As shown in the figure, ego represents the autonomous vehicle, i.e., the ego vehicle. v1 represents the speed of the autonomous vehicle, i.e., the speed of the ego vehicle. obs represents the reverse obstacle, i.e., the reverse obstacle in the second scenario. v2 represents the speed of the reverse obstacle in the second scenario.

[0249] For example, for the second type of reverse driving scenario, i.e., the second scenario, the strain-type model predictive control model is used to plan the response trajectory of the ego vehicle. In the process of constructing the objective function of the optimal control problem of the strain-type model predictive control model, similarly, the current state of the ego vehicle is , the state transfer equation is consistent with that in formula (1). The multimodal prediction trajectory of the reverse obstacle is a set of trajectory points in the Cartesian coordinate system. Then, the lane center reference line of the ego vehicle is used as the reference to convert it into the Frenet coordinate system. Since the reverse obstacle is in front of the ego vehicle, the reverse obstacle in the second scenario should also satisfy . is the current position of the vehicle, is the position of the opposite obstacle corresponding to the current state of the vehicle.

[0250] Furthermore, the first predicted trajectory in the multimodal predicted trajectory of the reverse obstacle may be: The second predicted trajectory can be: .

[0251] in, is the state vector of the retrograde obstacle, defined as:

[0252] Then, the objective function of the vehicle in response to the first predicted trajectory is And the objective function of the vehicle to deal with the second predicted trajectory It can be expressed as formula (9) and (10)

[0253] (9)

[0254] (10)

[0255] in, and is the first predicted trajectory of the vehicle to deal with the reverse obstacle The response state vector and action vector of and is the second predicted trajectory of the vehicle to deal with the reverse obstacle The response state vector and action vector, and Can be a weight coefficient, The first predicted trajectory for the vehicle is The reference state at the moment, For the vehicle to respond to the second predicted trajectory The reference state at the moment.

[0256] It can be understood that the updating process of the reference state and the value determination process of the weight coefficient here are consistent with the relevant processes of the first scenario. You can refer to the aforementioned relevant content and will not repeat them here.

[0257] Here, the constraints can also include equality constraints and inequality constraints. Equality constraints include initial state constraints and state transition constraints, which can be defined as:

[0258]

[0259]

[0260]

[0261]

[0262] Inequality constraints include collision constraints and reversing constraints, which can be defined as:

[0263]

[0264]

[0265]

[0266]

[0267] in, The minimum safe distance.

[0268] Variable bound constraints can be defined as:

[0269]

[0270]

[0271]

[0272]

[0273]

[0274]

[0275]

[0276]

[0277] in, The vehicle can cope with the jump in the first predicted trajectory, i.e., the acceleration. The vehicle can cope with the jump of the second predicted trajectory, that is, the acceleration.

[0278] The above optimization problem is summarized as an optimal control problem, that is, the objective function of the second programming model, which can be defined as formula (11):

[0279] (11)

[0280] st

[0281] ,

[0282] ,

[0283] ,

[0284]

[0285]

[0286] in, Can represent the first predicted trajectory of the retrograde obstacle The probability of entering the planned path or driving lane of the vehicle, and They are the first predicted trajectories of the vehicle above The equality and inequality constraints of and The second predicted trajectory of the vehicle is equality and inequality constraints. is the strain constraint.

[0287] In addition, since the focus is on the possibility that the oncoming obstacle actually invades the vehicle's driving path, the priority of considering the obstacle's actual running path is much higher than considering its speed.

[0288] Here, illustratively, the first predicted trajectory of the obstacle is ,estimate The probability of actually invading the driving path of the vehicle is , collect the actual running trajectory coordinates of the reverse obstacle for 5 consecutive frames (0.5s) The first predicted trajectory of the corresponding frame and the second predicted trajectory For comparison, take the minimum distance between the trajectory point and the two predicted trajectories. If the actual running trajectory point is within the trajectory in 5 consecutive frames, The closer the trajectory, the more points there are. If the match is better, then we can calculate it by formula (12): The probability of , from the trajectory The closer the trajectory, the more points there are. If the match is better, then we can calculate it by formula (13): Probability , as shown below:

[0289] (12)

[0290] (13)

[0291] in, The upper limit of the probability that can be set is The maximum probability step size that can be set is the maximum step size.

[0292] Step 208: Perform safety detection processing on the autonomous driving vehicle using the safety detection strategy, so that when the result of the detection processing is that the autonomous driving vehicle is driving dangerously, the braking acceleration is calculated based on the safety distance, the driving data of the autonomous driving vehicle, and the driving data of the oncoming obstacle, and the driving of the autonomous driving vehicle is controlled based on the braking acceleration. Alternatively, when the result of the detection processing is that the autonomous driving vehicle is driving safely, the driving of the autonomous driving vehicle is controlled based on the first planned trajectory or the second planned trajectory of the autonomous driving vehicle.

[0293] In this embodiment, first, a safety distance can be calculated based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle using a safety detection algorithm. Secondly, based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle, the relative distance between the autonomous driving vehicle and the wrong-traffic obstacle is calculated. Thirdly, it is determined whether the relative distance is less than the safety distance. If the relative distance is less than the safety distance, it can be determined that the result of the detection process is that the autonomous driving vehicle is driving dangerously. Thirdly, in the case where the result of the detection process is that the autonomous driving vehicle is driving dangerously, the braking acceleration is calculated based on the safety distance, the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle, and the driving of the autonomous driving vehicle is controlled based on the braking acceleration. If the relative distance is not less than the safety distance, it can be determined that the result of the detection process is that the autonomous driving vehicle is driving safely. Thirdly, in the case where the result of the detection process is that the autonomous driving vehicle is driving safely, the driving of the autonomous driving vehicle can be controlled based on the first planned trajectory or the second planned trajectory of the autonomous driving vehicle.

[0294] In this embodiment, preferably, the security detection algorithm can be expressed as formula (14):

[0295] (14)

[0296] in, A safe distance, Can be the speed of the current reverse obstacle, can be the current speed of the vehicle, The maximum deceleration preset for the vehicle.

[0297] Furthermore, the relative distance between the current vehicle and the obstacle can be calculated based on the current position of the vehicle and the position of the obstacle. , if the current relative distance between the vehicle and the reverse obstacle is , then the vehicle should immediately decelerate to avoid the impact. The deceleration can be calculated using formula (15), that is, the braking acceleration :

[0298] (15)

[0299] in, A safe distance, Can be the speed of the current reverse obstacle, It can be the current speed of the vehicle.

[0300] It is understood that safety checks can be performed on the autonomous vehicle directly after step 201. If the autonomous vehicle is determined to be in a dangerous driving situation, the braking acceleration of the autonomous vehicle can be directly calculated and the autonomous vehicle can be braked. If the autonomous vehicle is determined to be in a safe driving situation, subsequent steps 202 to 207 can be performed to obtain the first planned trajectory or the second planned trajectory of the autonomous vehicle.

[0301] Understandably, the logic behind the safety check strategy is that when the ego vehicle brakes to a stop and an oncoming obstacle strikes it, the ego vehicle has decelerated and braked to a complete stop, effectively ignoring the collision. The safety check algorithm can calculate the safe distance without considering the obstacle's response time. This avoids conservative driving decisions caused by overly stringent safety checks, and avoids insufficient deceleration due to overly optimistic predictions.

[0302] In this implementation, Figure 5A and 5B to Figure 9A and 9B This is a schematic diagram of the trajectory and speed points planned by the trajectory planning method of the autonomous driving vehicle in response to the first scenario of the reverse obstacle provided by another embodiment of the present application. Figure 5A and 5B to Figure 9A and 9B In the figure above, the actual driving trajectory is shown. The purple-red dot (traj_L) on the left side of the figure above represents the actual driving trajectory of the vehicle, and the light red dotted line in front represents the planned trajectory of the vehicle's selected acceleration (acc) / deceleration (dec). Figure 5A and 5B to Figure 9A and 9BIn the planning process shown in Figure 1, the trajectory planned for acceleration (acc) becomes increasingly shallow, while the trajectory planned for deceleration (dec) becomes increasingly clear. The blue dot (traj_F) on the right side of the top figure represents the actual trajectory of the oncoming obstacle. The bottom figure shows the speed planning graph. The purple dot (vxL) below represents the actual speed of the ego vehicle, and the blue dot (vxF) above represents the planned speed of the oncoming obstacle. Figure 5A and 5B to Figure 9A and 9B The figure shows the changes of the planned trajectory and planned velocity points over time t. For example, the initial position of the reverse obstacle can be defined as 40 meters (m) in front of the vehicle, with a speed of 6 meters / second (m / s) and an acceleration of 0 meters / square second (m / s). 2 ); the initial position of the vehicle is defined as 0m, the initial velocity is 5m / s, and the acceleration is 0m / s 2 The planning time step is 0.5s, the prediction period (horizon) is 10s, and the initial confidence deceleration probability of the obstacle is 0.5. Figure 5A In the example, the initial time t=0.0 seconds (s), the probability that the opposite obstacle takes deceleration is Pd=0.50, and the probability that the opposite obstacle takes acceleration is Pa=0.50. Figure 5B In the example, at time t=0.5 s, the probability that the opposite obstacle takes deceleration is Pd=0.50, and the probability that the opposite obstacle takes acceleration is Pa=0.50; Figure 6A In the example, at time t=2.0s, the probability that the opposite obstacle will decelerate is Pd=0.40, and the probability that the opposite obstacle will accelerate is Pa=0.60. Figure 6B In the example, at time t=2.5s, the probability that the obstacle moving against the vehicle will decelerate is Pd=0.50, and the probability that the obstacle moving against the vehicle will accelerate is Pa=0.50; Figure 7A In the example, at time t=4.0s, the probability that the opposite obstacle will decelerate is Pd=0.80, and the probability that the opposite obstacle will accelerate is Pa=0.20. Figure 7B In the example, at time t=4.5 s, the probability that the opposite obstacle takes deceleration is Pd=0.90, and the probability that the opposite obstacle takes acceleration is Pa=0.10; Figure 8A In the example, at time t=7.0s, the probability that the opposite obstacle will decelerate is Pd=0.95, and the probability that the opposite obstacle will accelerate is Pa=0.05. Figure 8B In the example, at time t=7.5 s, the probability that the opposite obstacle takes deceleration is Pd=0.95, and the probability that the opposite obstacle takes acceleration is Pa=0.05; Figure 9A In the example, at time t=9.0s, the probability that the opposite obstacle takes deceleration is Pd=0.95, and the probability that the opposite obstacle takes acceleration is Pa=0.05. Figure 9B At time t=9.5 s, the probability that the opposite obstacle takes deceleration is Pd=0.95, and the probability that the opposite obstacle takes acceleration is Pa=0.05.

[0303] like Figure 5A and 5B to Figure 9A and 9B As can be seen from the figure, when the initial distance between the ego vehicle and the oncoming obstacle vehicle is far, they choose to speed up to approach the expected speed and improve efficiency. When the distance is close, the two vehicles begin to slow down to ensure longitudinal safety.

[0304] In this implementation, Figure 10A and 10B to Figure 14A and 14B This is a schematic diagram of the planning results of the trajectory planning method for an autonomous driving vehicle according to another embodiment of the present application for dealing with a reverse obstacle in the second scenario. Figure 10A and 10B to Figure 14A and 14B The planning results may include planned trajectory, planned speed point, planned acceleration, and planned acceleration. The upper left figure shows the actual driving trajectory point diagram. The purple-red point (traj_L) on the left side of the upper left figure represents the actual driving trajectory point of the vehicle. The two light red lines 1 and 2, one deep and one light, represent the response trajectory of the vehicle in response to different modal prediction paths of the reverse obstacle, that is, the planned trajectory. The blue point on the right (traj_F) represents the actual driving trajectory point of the obstacle vehicle. In this embodiment, the actual driving trajectory of the reverse obstacle is the predicted trajectory that conflicts with the vehicle. The upper right figure shows the driving speed diagram of the vehicle and the reverse obstacle. The blue point (vF) below represents the obstacle. The actual speed of the obstructing vehicle is shown in Figure 1. The purple dot (vL) above represents the planned speed of the ego vehicle. The lower left figure shows the action vector, or jerk, record. Red line 1 represents the planned action vector for the ego vehicle's response to the predicted trajectory with a conflict with the oncoming obstacle, while green line 2 represents the planned action vector for the ego vehicle's response to the non-conflict path with the oncoming obstacle. The lower right figure shows the acceleration record. Red line 1 represents the planned acceleration for the predicted trajectory with a conflict with the oncoming obstacle, while green line 2 represents the planned acceleration for the non-conflict path with the oncoming obstacle. It can be seen that green line 2 in the lower left and right figures becomes shallower with planning time, indicating that the probability of the predicted non-conflict trajectory with the oncoming obstacle decreases.

[0305] Figure 10A and 10B to Figure 14A and 14BThe figure shows the planned trajectory, planned velocity points, planned jerk, and planned acceleration as a function of time t for the ego vehicle in the second scenario, when dealing with an oncoming obstacle. The oncoming obstacle's initial position is defined as 50 meters in front of the ego vehicle, with a velocity of 5 m / s and an acceleration of 0. The ego vehicle's initial position is defined as 0 meters, with an initial velocity of 5 m / s and an acceleration of 0. The planning time step is 0.5 seconds, and the horizon is 10 seconds. The probability that the obstacle's potentially conflicting predicted trajectory initially intersects the ego vehicle's path is 0.7. Figure 10A In the initial time t=0.0s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1=0.70, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2=0.30. Figure 10B In the example, at time t = 0.5 s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1 = 0.70, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2 = 0.30; Figure 11A At time t=1.0s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1=0.80, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2=0.20. Figure 11B In the example, at time t = 1.5 s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1 = 0.90, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2 = 0.10; Figure 12A At time t=5.0s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1=0.95, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2=0.05. Figure 12B In the example, at time t = 5.5 s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1 = 0.95, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2 = 0.05; Figure 13A At time t=7.0s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1=0.95, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2=0.05. Figure 13B In the example, at time t=7.5s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1=0.95, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2=0.05; Figure 14A At time t=9.0s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1=0.95, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2=0.05. Figure 14B At time t = 9.5 s, the probability that the reverse obstacle may have a conflicting predicted trajectory is P1 = 0.95, and the probability that the reverse obstacle may have a non-conflicting predicted trajectory is P2 = 0.05.

[0306] like Figure 10A and 10B to Figure 14A and 14BAs shown in the figure, it can be seen that the vehicle predicts the paths of two opposite conflicting obstacles, effectively judges the actual possible motion trajectories of the opposite obstacles, and responds to the opposite intentions and behaviors of the obstacles using a contingency processing method.

[0307] In this way, by adopting the technical solution in this embodiment, it is possible to analyze and classify the reverse-travel obstacle scenarios when passing through narrow roads, handle the reverse-travel obstacles located on the planned path of the self-vehicle based on an interactive game method, and handle the reverse-travel obstacles with intrusion intentions located outside the planned path of the self-vehicle based on a strain processing method based on multimodal prediction of obstacles. This can achieve a balance between robustness and exploratoryness when the self-vehicle makes decisions and plans the trajectory to deal with reverse-travel obstacles.

[0308] In addition, the technical solution in this embodiment can avoid sudden braking and evasive behavior caused by overly conservative decision-making of the vehicle, improve the continuity of the vehicle's decision-making, and prevent step-like mutations in the decision-making process.

[0309] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0310] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0311] Figure 15 FIG. 1 shows a structural block diagram of a trajectory planning device for an autonomous driving vehicle provided by an embodiment of the present application. Figure 15As shown. The trajectory planning device 1500 of the autonomous driving vehicle of this embodiment may include a first acquisition unit 1501, a first classification unit 1502, a first planning unit 1503, and a second planning unit 1504. The first acquisition unit 1501 is used to acquire the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data; the first classification unit 1502 is used to classify the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data using a preset scenario classification strategy; the first planning unit 1503 is used to, in response to the wrong-traffic obstacle being a wrong-traffic obstacle in a first scenario, obtain a first planned trajectory of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle in the first scenario using a first planning model; the second planning unit 1504 is used to, in response to the wrong-traffic obstacle being a wrong-traffic obstacle in a second scenario, obtain a second planned trajectory of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle in the second scenario using a second planning model.

[0312] It should be noted that part or all of the trajectory planning device of the autonomous driving vehicle of this embodiment may be an application located in the local terminal, or may also be a functional unit such as a plug-in or software development kit (SDK) set in the application located in the local terminal, or may also be a processing engine located in a network-side server, or may also be a distributed system located on the network side, for example, a processing engine or distributed system in an autonomous driving platform on the network side, etc. This embodiment does not specifically limit this.

[0313] It is understandable that the application may be a native program (nativeApp) installed on the local terminal, or may be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0314] Optionally, in a possible implementation of this embodiment, the first planning unit 1503 can be used to determine the motion vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle and the driving data of the reverse obstacle in the first scene using a first planning model; and obtain the first planned trajectory of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle and the motion vector.

[0315] Optionally, in a possible implementation of this embodiment, the first planning unit 1503 can be used to determine the state vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle; infer the state vector of the first-scene retrograde obstacle based on the driving data of the autonomous driving vehicle and the driving data of the first-scene retrograde obstacle; optimize the objective function of the first planning model based on the state vector of the autonomous driving vehicle, the state vector of the first-scene retrograde obstacle, and the constraints of the first planning model; and determine the action vector of the autonomous driving vehicle based on the result of the optimization process.

[0316] Optionally, in a possible implementation of this embodiment, the driving data of the second-scenario wrong-way obstacle includes a first predicted trajectory and a second predicted trajectory, and the second planning unit 1504 can be used to determine the motion vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle, the first predicted trajectory and the second predicted trajectory using a second planning model; and obtain the second planned trajectory of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle and the motion vector.

[0317] Optionally, in a possible implementation of this embodiment, the second planning unit 1504 can be used to determine the state vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle; determine the state vector of the first predicted trajectory based on the first predicted trajectory of the wrong-way obstacle in the second scenario; determine the state vector of the second predicted trajectory based on the second predicted trajectory of the wrong-way obstacle in the second scenario; optimize the objective function of the second planning model based on the state vector of the autonomous driving vehicle, the state vector of the first predicted trajectory, the state vector of the second predicted trajectory, and the constraints of the second planning model; and determine the action vector of the autonomous driving vehicle based on the result of the optimization.

[0318] Optionally, in a possible implementation of this embodiment, the first classification unit 1502 is used to determine whether the driving data of the wrong-traffic obstacle matches the driving data of the autonomous driving vehicle; in response to the driving data of the wrong-traffic obstacle matching the driving data of the autonomous driving vehicle, determine that the wrong-traffic obstacle is a first-scenario wrong-traffic obstacle; in response to the driving data of the wrong-traffic obstacle not matching the driving data of the autonomous driving vehicle, determine that the wrong-traffic obstacle is a second-scenario wrong-traffic obstacle.

[0319] Optionally, in a possible implementation of this embodiment, the first classification unit 1502 is used to determine the lane area corresponding to the autonomous driving vehicle based on the driving data and road data of the autonomous driving vehicle; determine whether the driving data of the wrong-moving obstacle matches the lane area corresponding to the autonomous driving vehicle; in response to the driving data of the wrong-moving obstacle matching the lane area corresponding to the autonomous driving vehicle, determine that the wrong-moving obstacle is a first-scenario wrong-moving obstacle; in response to the driving data of the wrong-moving obstacle not matching the lane area corresponding to the autonomous driving vehicle, determine that the wrong-moving obstacle is a second-scenario wrong-moving obstacle.

[0320] Optionally, in a possible implementation of this embodiment, the first acquisition unit 1501 is used to calculate a safety distance based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle using a safety detection algorithm; calculate the relative distance between the autonomous driving vehicle and the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle; and calculate a braking acceleration based on the safety distance, the driving data of the autonomous driving vehicle, and the driving data of the wrong-traffic obstacle when the relative distance and the safety distance meet preset conditions, so as to control the driving of the autonomous driving vehicle based on the braking acceleration.

[0321] In this embodiment, the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data can be obtained by the first acquisition unit. Then, the first classification unit can classify the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data using a preset scenario classification strategy. The first planning unit, in response to the wrong-traffic obstacle being a first-scenario wrong-traffic obstacle, obtains a first planned trajectory of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle and the driving data of the first-scenario wrong-traffic obstacle using a first planning model. The second planning unit, in response to the wrong-traffic obstacle being a second-scenario wrong-traffic obstacle, obtains a first planned trajectory of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle and the driving data of the first-scenario wrong-traffic obstacle. The driving data of the autonomous driving vehicle and the driving data of the second-scenario wrong-way obstacle are used to obtain the second planned trajectory of the autonomous driving vehicle using the second planning model. Since the wrong-way obstacles can be classified into scenarios first, the autonomous driving vehicle can then plan trajectories for wrong-way obstacles in different types of scenarios respectively, taking into account the bounded rational interactive game of wrong-way obstacles, it can effectively solve the problem of narrow road passage of the autonomous driving vehicle while avoiding the sudden braking avoidance behavior caused by overly conservative trajectory planning of the autonomous driving vehicle, taking into account the robustness and exploratory nature of the trajectory planning of the autonomous driving vehicle, and improving the continuity of the planned trajectory of the autonomous driving vehicle, thereby ensuring the reliability of the autonomous driving vehicle.

[0322] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved, such as user images and attribute data, comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0323] According to an embodiment of the present application, the present application also provides an electronic device, a readable storage medium and a computer program product.

[0324] According to an embodiment of the present application, an unmanned vehicle including the provided electronic device is further provided. The autonomous driving vehicle may include vehicles of level 2 or above. For example, the autonomous driving vehicle may include, but is not limited to, autonomous logistics vehicles, autonomous inspection vehicles, autonomous delivery vehicles, and autonomous large vehicles.

[0325] Figure 16 A schematic block diagram of an example electronic device 1600 that can be used to implement embodiments of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0326] like Figure 16 As shown, electronic device 1600 includes a computing unit 1601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1602 or a computer program loaded from a storage unit 1608 into a random access memory (RAM) 1603. RAM 1603 may also store various programs and data required for the operation of electronic device 1600. Computing unit 1601, ROM 1602, and RAM 1603 are interconnected via a bus 1604. An input / output (I / O) interface 1605 is also connected to bus 1604.

[0327] Multiple components in electronic device 1600 are connected to I / O interface 1605, including: an input unit 1606, such as a keyboard, mouse, etc.; an output unit 1607, such as various types of displays, speakers, etc.; a storage unit 1608, such as a magnetic disk, optical disk, etc.; and a communication unit 1609, such as a network card, modem, wireless communication transceiver, etc. The communication unit 1609 allows electronic device 1600 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0328] Computing unit 1601 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of computing unit 1601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Computing unit 1601 performs the various methods and processes described above, such as the trajectory planning method for an autonomous vehicle. For example, in some embodiments, the trajectory planning method for an autonomous vehicle can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 1608. In some embodiments, part or all of the computer program can be loaded and / or installed onto electronic device 1600 via ROM 1602 and / or communication unit 1609. When the computer program is loaded into RAM 1603 and executed by computing unit 1601, one or more steps of the trajectory planning method for an autonomous vehicle described above can be performed. Alternatively, in other embodiments, the computing unit 1601 may be configured to execute the trajectory planning method for the autonomous driving vehicle in any other appropriate manner (e.g., by means of firmware).

[0329] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-a-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0330] The program code for implementing the methods of the present application can be written in any combination of one or more programming languages. Such program code can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the program code is executed by the processor or controller, the functions / operations specified in the flow charts and / or block diagrams are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0331] In the context of this application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0332] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0333] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0334] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0335] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this application can be achieved. This is not a limitation herein.

[0336] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A trajectory planning method for an autonomous driving vehicle, characterized in that: The method comprises: Acquire driving data of the autonomous vehicle, driving data of the wrong-way obstacle, and road data; Calculating a safe distance using a safety detection algorithm based on a preset maximum deceleration of the autonomous driving vehicle, the speed in the driving data of the autonomous driving vehicle, and the speed in the driving data of the oncoming obstacle; Calculating a relative distance between the autonomous driving vehicle and the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle; When the relative distance and the safety distance satisfy a preset condition, calculating a braking acceleration based on the safety distance, a speed in the driving data of the autonomous driving vehicle, and a speed in the driving data of the oncoming obstacle, so as to control the driving of the autonomous driving vehicle based on the braking acceleration; If the relative distance and the safety distance do not meet the preset conditions, classify the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, using a preset scene classification strategy; In response to the wrong-way obstacle being a first-scenario wrong-way obstacle, obtaining a first planned trajectory of the autonomous vehicle using a first planning model based on the driving data of the autonomous vehicle and the driving data of the first-scenario wrong-way obstacle; the first-scenario wrong-way obstacle is a wrong-way obstacle on the planned path of the autonomous vehicle; In response to the wrong-way obstacle being a second-scenario wrong-way obstacle, a second planned trajectory of the autonomous vehicle is obtained using a second planning model based on the probability of the second-scenario wrong-way obstacle entering the planned path of the autonomous vehicle, the driving data of the autonomous vehicle, and the driving data of the second-scenario wrong-way obstacle. The second-scenario wrong-way obstacle is a wrong-way obstacle that may conflict with the autonomous vehicle in the future.

2. The method according to claim 1, characterized in that The method of obtaining a first planned trajectory of the autonomous driving vehicle by using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the wrong-way obstacle in the first scenario includes: Determining a motion vector of the autonomous driving vehicle using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the wrong-way obstacle in the first scenario; Based on the driving data of the autonomous driving vehicle and the motion vector, a first planned trajectory of the autonomous driving vehicle is obtained.

3. The method according to claim 2, characterized in that The determining of the motion vector of the autonomous driving vehicle using a first planning model based on the driving data of the autonomous driving vehicle and the driving data of the wrong-way obstacle in the first scenario includes: Determining a state vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle; Inferring a state vector of the first-scenario wrong-traffic obstacle based on the driving data of the autonomous driving vehicle and the driving data of the first-scenario wrong-traffic obstacle; Optimizing the objective function of the first planning model based on the state vector of the autonomous driving vehicle, the state vector of the wrong-way obstacle in the first scenario, and the constraints of the first planning model; Based on the result of the optimization process, an action vector of the autonomous driving vehicle is determined.

4. The method according to claim 1, wherein The driving data of the second-scenario wrong-way obstacle includes a first predicted trajectory and a second predicted trajectory. The second planned trajectory of the autonomous driving vehicle is obtained by using a second planning model based on the driving data of the autonomous driving vehicle and the driving data of the second-scenario wrong-way obstacle, including: determining a motion vector of the autonomous vehicle using a second planning model based on the driving data of the autonomous vehicle, the first predicted trajectory, and the second predicted trajectory; Based on the driving data of the autonomous driving vehicle and the motion vector, a second planned trajectory of the autonomous driving vehicle is obtained.

5. The method according to claim 4, characterized in that The determining, based on the driving data of the autonomous driving vehicle, the first predicted trajectory, and the second predicted trajectory, using a second planning model, of a motion vector of the autonomous driving vehicle includes: Determining a state vector of the autonomous driving vehicle based on the driving data of the autonomous driving vehicle; determining a state vector of the first predicted trajectory based on the first predicted trajectory of the second scenario's reverse obstacle; determining a state vector of the second predicted trajectory based on the second predicted trajectory of the wrong-way obstacle in the second scenario; Optimizing an objective function of the second planning model based on the state vector of the autonomous driving vehicle, the state vector of the first predicted trajectory, the state vector of the second predicted trajectory, and the constraints of the second planning model; Based on the result of the optimization process, an action vector of the autonomous driving vehicle is determined.

6. The method according to claim 1, characterized in that The classifying process of the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data using a preset scene classification strategy includes: Determining whether the driving data of the wrong-way obstacle matches the driving data of the autonomous driving vehicle; In response to the driving data of the wrong-traffic obstacle matching the driving data of the autonomous driving vehicle, determining that the wrong-traffic obstacle is a first-scenario wrong-traffic obstacle; In response to the driving data of the wrong-traffic obstacle not matching the driving data of the autonomous driving vehicle, the wrong-traffic obstacle is determined to be a second-scenario wrong-traffic obstacle.

7. The method according to claim 1, characterized in that The method of classifying the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data using a preset scene classification strategy further includes: Determining a lane area corresponding to the autonomous driving vehicle based on the driving data and road data of the autonomous driving vehicle; Determining whether the driving data of the wrong-way obstacle matches the lane area corresponding to the autonomous driving vehicle; In response to the driving data of the wrong-traffic obstacle matching the lane area corresponding to the autonomous driving vehicle, determining that the wrong-traffic obstacle is a first-scenario wrong-traffic obstacle; In response to the driving data of the wrong-traffic obstacle not matching the lane area corresponding to the autonomous driving vehicle, the wrong-traffic obstacle is determined to be a second-scenario wrong-traffic obstacle.

8. A trajectory planning device for an autonomous vehicle, characterized in that: The device comprises: a first acquisition unit configured to acquire driving data of the autonomous driving vehicle, driving data of the wrong-traffic obstacle, and road data; calculate a safety distance using a safety detection algorithm based on a preset maximum deceleration of the autonomous driving vehicle, a speed in the driving data of the autonomous driving vehicle, and a speed in the driving data of the wrong-traffic obstacle; calculate a relative distance between the autonomous driving vehicle and the wrong-traffic obstacle based on the driving data of the autonomous driving vehicle and the driving data of the wrong-traffic obstacle; and calculate a braking acceleration based on the safety distance, the speed in the driving data of the autonomous driving vehicle, and the speed in the driving data of the wrong-traffic obstacle, if the relative distance and the safety distance meet preset conditions, so as to control the driving of the autonomous driving vehicle based on the braking acceleration; a first classification unit, configured to classify the wrong-traffic obstacle using a preset scenario classification strategy based on the driving data of the autonomous driving vehicle, the driving data of the wrong-traffic obstacle, and the road data, when the relative distance and the safety distance do not satisfy a preset condition; a first planning unit configured to, in response to the wrong-way obstacle being a first-scenario wrong-way obstacle, obtain a first planned trajectory of the autonomous vehicle using a first planning model based on the driving data of the autonomous vehicle and the driving data of the first-scenario wrong-way obstacle; the first-scenario wrong-way obstacle being a wrong-way obstacle on the planned path of the autonomous vehicle; The second planning unit is configured to, in response to the wrong-way obstacle being a second-scenario wrong-way obstacle, obtain a second planned trajectory of the autonomous vehicle by using a second planning model based on the probability of the second-scenario wrong-way obstacle entering the planned path of the autonomous vehicle, the driving data of the autonomous vehicle, and the driving data of the second-scenario wrong-way obstacle, wherein the second-scenario wrong-way obstacle is a wrong-way obstacle that may conflict with the autonomous vehicle in the future.

9. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

11. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

12. An autonomous driving vehicle comprising the electronic device according to claim 9.

Images