Closed-loop simulation test method for automatic driving algorithm

Abstract

The invention discloses a closed-loop simulation test method for an automatic driving algorithm, and relates to the technical field of automatic driving test. The method comprises the following steps: obtaining each to-be-measured algorithm module corresponding to a to-be-measured automatic driving algorithm; for each to-be-measured algorithm module, determining a module type corresponding to the to-be-measured algorithm module, and closing a functional module corresponding to the module type in a preset closed-loop simulation test system; accessing the to-be-measured algorithm module to the position of a functional module corresponding to the module type in a preset closed-loop simulation test system, acquiring input data through a data bus of the preset closed-loop simulation test system, and outputting output data through the data bus of the preset closed-loop simulation test system; according to the simulation test state of each to-be-measured algorithm module in the preset closed-loop simulation test system, the simulation test performance of each to-be-measured algorithm module is evaluated, the closed-loop performance test of one or more to-be-measured algorithm modules under a complete automatic driving framework can be realized, and the parallel development and test efficiency of an automatic driving algorithm is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving testing, and particularly to a closed-loop simulation testing method for autonomous driving algorithms. Background Art

[0002] With the continuous development of autonomous driving technology, the testing of autonomous driving algorithms has received increasing attention. The testing methods of autonomous driving algorithms are mainly divided into two types: real vehicle testing and simulation testing. Although the credibility of real vehicle testing is higher, its disadvantages such as small scene coverage, low safety, and high cost make it impossible to complete the testing of autonomous driving algorithms. Due to advantages such as large coverage, high safety, and low cost, simulation testing algorithms are widely used in the field of autonomous driving testing.

[0003] In the process of implementing the present invention, the inventors found that the autonomous driving simulation testing methods provided by related technologies have the following technical problems:

[0004] The autonomous driving simulation testing methods provided by related technologies can only perform closed-loop testing on complete autonomous driving algorithms, or can only perform separate open-loop testing on a single functional module or multiple functional modules of autonomous driving algorithms respectively. As a result, the design or update of autonomous driving algorithms is likely to be affected by defects in single or multiple functional modules, leading to low testing efficiency. Moreover, the test results cannot truly reflect the fault problems caused by the closed-loop linkage between functional modules. Summary of the Invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides a closed-loop simulation testing method for autonomous driving algorithms. By providing access interfaces for each functional module corresponding to the algorithm module to be tested in a preset complete closed-loop autonomous driving algorithm that meets the test criteria, the closed-loop performance testing of single or multiple algorithm modules to be tested in the autonomous driving algorithm can be realized under the complete autonomous driving framework, thereby improving the parallel development and testing efficiency of autonomous driving algorithms. The technical solutions provided by the present invention are as follows:

[0006] According to one aspect of an embodiment of the present invention, there is provided a closed-loop simulation testing method for autonomous driving algorithms, characterized in that the closed-loop simulation testing method includes:

[0007] S10: Obtain each algorithm module to be tested corresponding to the autonomous driving algorithm to be tested;

[0008] S20: For each algorithm module to be tested, determine the module type corresponding to the algorithm module to be tested, and turn off the functional module corresponding to the module type in a preset closed-loop simulation testing system, where the preset closed-loop simulation testing system is a complete closed-loop autonomous driving algorithm that meets the test criteria;

[0009] S30: connecting the algorithm module to be tested to the position of the functional module corresponding to the module type in the preset closed-loop simulation test system, acquiring input data through the data bus of the preset closed-loop simulation test system, and outputting output data through the data bus of the preset closed-loop simulation test system;

[0010] S40: Evaluate the simulation test performance of each algorithm module to be tested in the autonomous driving algorithm to be tested according to the simulation test status of each algorithm module to be tested in the preset closed-loop simulation test system.

[0011] Preferably, the preset closed-loop simulation test system includes a task input module, a perception module, a lateral decision module, a longitudinal decision module, a lateral planning module, a longitudinal planning module, a lateral control module, a longitudinal control module, a data bus, a module switch and an access interface for the algorithm to be tested;

[0012] Wherein, the task input module is used to obtain driving task information and driving map information;

[0013] The perception module is used to obtain perception information;

[0014] The lateral decision module is used to determine the lateral planner number to be called according to the driving task information, the driving map information and the perception information, and then output the lateral planner number to the lateral planning module;

[0015] The longitudinal decision module is used to determine the longitudinal planner number to be called according to the driving task information, the driving map information and the perception information, and then output the longitudinal planner number to the longitudinal planning module;

[0016] The lateral planning module is used to call the corresponding lateral planner according to the lateral planner number, calculate the expected path data according to the lateral planner, and then output the expected path data to the lateral control module;

[0017] The longitudinal planning module is used to call the corresponding longitudinal planner according to the longitudinal planner number, calculate the expected speed data and the expected acceleration data according to the longitudinal planner, and then output the expected speed data and the expected acceleration data to the longitudinal control module;

[0018] The lateral control module is used to calculate the steering wheel angle data according to the expected path data, and control the steering wheel to perform corresponding steering operations according to the steering wheel angle data;

[0019] The longitudinal control module is used to calculate the brake master cylinder pressure data and the throttle opening data according to the expected speed data and the expected acceleration data, and control the brake pedal and the accelerator pedal to perform corresponding braking operations or acceleration / deceleration operations according to the brake master cylinder pressure data and the throttle opening data;

[0020] The lateral decision-making module, the longitudinal decision-making module, the lateral planning module, the longitudinal planning module, the lateral control module, and the longitudinal control module are replaceable modules that can be replaced by the algorithm module to be tested;

[0021] The data bus is used to transmit data between the algorithm module to be tested and the preset closed-loop simulation test system; the module switch is used to open and close at least one replaceable module in the preset closed-loop simulation test system; the algorithm access interface to be tested is used to connect the algorithm module to be tested to the preset closed-loop simulation test system.

[0022] Preferably, the module type corresponding to the algorithm module to be tested connected to the preset closed-loop simulation test system is any one of the replaceable module types; or,

[0023] The module type corresponding to the algorithm module to be tested connected to the preset closed-loop simulation test system is at least two of the replaceable module types.

[0024] Preferably, the types of the lateral planners include a lane change path planner, an intersection path planner, and an empty planner;

[0025] The lane change path planner uses a fifth-degree polynomial curve to implement path planning for straight roads and curves. The fitting formula of the fifth-degree polynomial curve is as shown in Equation 1:

[0026] y d (x d )=a0x d 5 +a1x d 4 +a2x d 3 +a3x d 2 +a4x d +a5 Equation 1

[0027] Where, a4 = 0, a5 = 0, W represents the lane change width, and D represents the lane change distance;

[0028] The intersection path planner uses a Bezier curve to perform path planning for left turns, straight runs, right turns, and U-turns. The fitting formula of the Bezier curve is as shown in Equation 2:

[0029]

[0030] Among them, P i is a control point, is the quadratic coefficient, t is a parameter, and each parameter can generate a trajectory point;

[0031] The empty planner is used to be called when there is no planning task.

[0032] Preferably, the types of the longitudinal planner include a constant speed cruise planner, a stable car-following planner, a fixed-point parking planner, and an emergency braking planner;

[0033] The constant speed cruise planner is used to realize the function of the test vehicle cruising at a preset speed;

[0034] The stable car-following planner is used to realize the function of the test vehicle following the vehicle in front at a preset spacing interval;

[0035] The fixed-point parking planner is used to realize the function of the test vehicle stopping at a preset position in front or decelerating to a preset speed range;

[0036] The emergency braking planner is used to realize the emergency obstacle avoidance function of the test vehicle.

[0037] Preferably, the longitudinal control module outputs the brake master cylinder pressure data through the first control loop and outputs the throttle opening data through the second control loop;

[0038] The first control loop adopts a single-layer PID plus open-loop control scheme; the single-layer PID controller is used to calculate the expected deceleration data according to the expected vehicle speed data and the actual vehicle speed data of the test vehicle, and then calculate the brake master cylinder pressure data by the look-up table method according to the expected deceleration speed;

[0039] The second control loop adopts an inner and outer two-layer cascade PID controller scheme; among them, the input of the outer-layer PID controller is the expected vehicle speed data and the actual vehicle speed data of the test vehicle, and the output is the expected acceleration data; the input of the inner-layer PID controller is the expected acceleration data and the actual acceleration data of the test vehicle, and the output is the throttle opening data.

[0040] Preferably, when the expected vehicle speed is less than 1 m / s, the outer-layer PID controller of the second control loop is integrally cleared; when the brake switch is turned on, the inner-layer PID of the second control loop is integrally cleared.

[0041] Preferably, the lateral control module dynamically selects a preview point according to the actual vehicle speed of the test vehicle by using a pure tracking algorithm, then calculates the feedforward value of the steering wheel angle, and then uses a model predictive control algorithm to calculate the expected steering wheel angle data according to the feedforward value combined with the expected path and the vehicle state.

[0042] Preferably, when the stable following planner detects that the distance between the test vehicle and the vehicle ahead is higher than the preset following distance, the speed of the vehicle ahead plus the preset following speed value is directly output as the expected speed data;

[0043] When the stable following planner detects that the distance between the test vehicle and the vehicle in front is not higher than the preset following distance, the stable following planner calculates and outputs the expected speed data in real time according to the expected following distance input by the user and the actual following distance.

[0044] Preferably, the longitudinal decision module also includes a speed envelope generator, which is used to obtain the road curvature, and then calculate the maximum vehicle speed supported by the test vehicle under the road curvature, and when it is detected that the expected speed data is greater than the maximum speed, the speed of the test vehicle is reduced to the maximum speed.

[0045] Compared with the prior art, the closed-loop simulation test method for autonomous driving algorithms provided by the present invention has the following advantages:

[0046] The present invention provides a closed-loop simulation test method for an autonomous driving algorithm, which comprises the following steps: obtaining each algorithm module to be tested corresponding to the autonomous driving algorithm to be tested; determining, for each algorithm module to be tested, the module type corresponding to the algorithm module to be tested, and shutting down the functional module corresponding to the module type in a preset closed-loop simulation test system, wherein the preset closed-loop simulation test system is a complete closed-loop autonomous driving algorithm that meets the test standard; connecting the algorithm module to be tested to the position of the functional module corresponding to the module type in the preset closed-loop simulation test system, obtaining input data through a data bus of the preset closed-loop simulation test system, and outputting output data through a data bus of the preset closed-loop simulation test system; evaluating the simulation test performance of each algorithm module to be tested in the autonomous driving algorithm to be tested according to the simulation test status of each algorithm module to be tested in the preset closed-loop simulation test system, and realizing closed-loop performance testing of a single or multiple algorithm modules to be tested in the autonomous driving algorithm under a complete autonomous driving framework, thereby improving the parallel development and testing efficiency of the autonomous driving algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0048] Figure 1 It is a method flow chart of a closed-loop simulation test method for an autonomous driving algorithm according to an exemplary embodiment of the present invention.

[0049] Figure 2It is a schematic diagram of the implementation principle of the closed-loop simulation test method for the automatic driving algorithm provided by the present invention. Detailed implementation manners

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0051] Figure 1 It is a flowchart of a method for a closed-loop simulation test method for an automatic driving algorithm according to an exemplary embodiment of the present invention. As Figure 1 shown, the closed-loop simulation test method for the automatic driving algorithm includes:

[0052] S10: Obtain each algorithm module to be tested corresponding to the automatic driving algorithm to be tested.

[0053] S20: For each algorithm module to be tested, determine the module type corresponding to the algorithm module to be tested, and turn off the function module corresponding to the module type in the preset closed-loop simulation test system. The preset closed-loop simulation test system is a complete closed-loop automatic driving algorithm that meets the test standards.

[0054] Preferably, the preset closed-loop simulation test system includes a task input module, a perception module, a lateral decision-making module, a longitudinal decision-making module, a lateral planning module, a longitudinal planning module, a lateral control module, a longitudinal control module, a data bus, a module switch, and an access interface for the algorithm under test; wherein, the task input module is used to obtain driving task information and driving map information; the perception module is used to obtain perception information; the lateral decision-making module is used to determine the number of the lateral planner to be called according to the driving task information, the driving map information, and the perception information, and then output the number of the lateral planner to the lateral planning module; the longitudinal decision-making module is used to determine the number of the longitudinal planner to be called according to the driving task information, the driving map information, and the perception information, and then output the number of the longitudinal planner to the longitudinal planning module; the lateral planning module is used to call the corresponding lateral planner according to the number of the lateral planner, calculate the expected path data according to the lateral planner, and then output the expected path data to the lateral control module; the longitudinal planning module is used to call the corresponding longitudinal planner according to the number of the longitudinal planner, calculate the expected speed data and the expected acceleration data according to the longitudinal planner, and then output the expected speed data and the expected acceleration data to the longitudinal control module; the lateral control module is used to calculate the steering wheel angle data according to the expected path data, and control the steering wheel to perform corresponding steering operations according to the steering wheel angle data; the longitudinal control module is used to calculate the brake master cylinder pressure data and the throttle opening data according to the expected speed data and the expected acceleration data, and control the brake pedal and the accelerator pedal to perform corresponding braking operations or acceleration / deceleration operations according to the brake master cylinder pressure data and the throttle opening data; the lateral decision-making module, the longitudinal decision-making module, the lateral planning module, the longitudinal planning module, the lateral control module, and the longitudinal control module are replaceable modules that can be replaced by the algorithm module under test; the data bus is used to transmit data between the algorithm module under test and the preset closed-loop simulation test system; the module switch is used to open and close at least one replaceable module in the preset closed-loop simulation test system; the access interface for the algorithm under test is used to connect the algorithm module under test to the preset closed-loop simulation test system.

[0055] It should be noted that the task input module can transmit the driving map information from the Internet or an external connected device, and obtain the driving task information input by the user through an external interaction device. Among them, the driving map information may include traffic information such as a miniature map information, road lane information, and road speed limit information; the driving task information may include starting point information, target point information, departure time information, etc.

[0056] The perception information obtained by the perception module includes the perception of dynamic traffic participants, the perception of static traffic participants, the perception of traffic lights, and the perception of traffic signs. Among them, dynamic traffic participants refer to dynamic traffic element information such as pedestrians, bicycles, and various forms of vehicles participating in traffic, and static traffic participants refer to static traffic element information such as cones and railings.

[0057] Preferably, the types of the lateral planner include a lane-changing path planner, an intersection path planner, and an empty planner;

[0058] The lane-changing path planner uses a fifth-degree polynomial curve to implement the path planning of straight roads and curves. The fitting formula of the fifth-degree polynomial curve is as shown in Equation 1:

[0059] y d (x d )=a0x d 5 +a1x d 4 +a2x d 3 +a3x d 2 +a4x d +a5 Equation 1

[0060] Where, a3 = 0, a4 = 0, a5 = 0, W represents the lane-changing width, and D represents the lane-changing distance.

[0061] When planning the lane-changing trajectory, the lane-changing path planner can first convert the trajectory in the vehicle coordinate system or the earth coordinate system into the Frenet coordinate system. Then, use the above method to plan the lane-changing trajectory in the Frenet coordinate system. Finally, convert the straight-road lane-changing trajectory in the Frenet coordinate system into the vehicle coordinate system or the earth coordinate system to meet the lane-changing trajectory planning of roads with different curvatures.

[0062] The intersection path planner uses a Bezier curve to perform path planning for left turns, straight runs, right turns, and U-turns. The fitting formula of the Bezier curve is as shown in Equation 2:

[0063]

[0064] Where, P i is the control point, is the quadratic coefficient, t is the parameter, and each parameter can generate a trajectory point.

[0065] In the cases of left turns, right turns, and straight runs, the Bezier curve is fitted using three control points; in the case of a U-turn, the Bezier curve is fitted using four control points.

[0066] The empty planner is used to be called when there is no planning task.

[0067] Preferably, the types of the longitudinal planners include a constant speed cruise planner, a stable car-following planner, a fixed-point parking planner, and an emergency braking planner;

[0068] The constant speed cruise planner is used to implement the function of enabling the test vehicle to cruise at a preset speed;

[0069] The stable car-following planner is used to implement the function of enabling the test vehicle to follow the vehicle ahead at a preset spacing interval;

[0070] The fixed-point parking planner is used to implement the function of enabling the test vehicle to park at a preset position ahead or decelerate to a preset speed interval;

[0071] The emergency braking planner is used to implement the function of enabling the test vehicle to avoid obstacles emergently.

[0072] Preferably, the longitudinal control module outputs the brake master cylinder pressure data through the first control loop and outputs the throttle opening data through the second control loop;

[0073] The first control loop adopts a single-layer PID plus open-loop control scheme; the single-layer PID controller is used to calculate the desired deceleration data according to the desired vehicle speed data and the actual vehicle speed data of the test vehicle, and then calculate the brake master cylinder pressure data by means of a look-up table method according to the desired deceleration speed;

[0074] The second control loop adopts an inner and outer two-layer cascade PID controller scheme; wherein, the input of the outer-layer PID controller is the desired vehicle speed data and the actual vehicle speed data of the test vehicle, and the output is the desired acceleration data; the input of the inner-layer PID controller is the desired acceleration data and the actual acceleration data of the test vehicle, and the output is the throttle opening data.

[0075] Preferably, when the desired vehicle speed is less than 1 m / s, the outer-layer PID controller of the second control loop is cleared of integration; when the brake switch is turned on, the inner-layer PID of the second control loop is cleared of integration.

[0076] Among them, the integration clearing measure is to avoid the integral saturation of the PID controller.

[0077] Preferably, the lateral control module adopts a pure tracking algorithm to dynamically select a preview point according to the actual vehicle speed of the test vehicle, then calculates the feedforward value of the steering wheel angle, and then adopts a model predictive control algorithm to calculate the desired steering wheel angle data according to the feedforward value in combination with the desired path and the vehicle state.

[0078] Preferably, when the stable car - following planner detects that the distance between the test vehicle and the vehicle ahead is higher than the preset car - following distance, it directly outputs the speed of the vehicle ahead plus the preset car - following speed value as the desired speed data.

[0079] When the stable car - following planner detects that the distance between the test vehicle and the vehicle ahead is not higher than the preset car - following distance, the stable car - following planner calculates and outputs the desired speed data in real - time according to the desired car - following distance input by the user and the actual car - following distance.

[0080] Preferably, the longitudinal decision - making module further includes a speed envelope generator. The speed envelope generator is used to obtain the road curvature, then calculate the maximum vehicle speed supported by the test vehicle under the road curvature, and when it detects that the desired speed data is greater than the maximum vehicle speed, reduce the speed of the test vehicle to the maximum vehicle speed.

[0081] S30: Connect the algorithm module to be tested to the position of the function module corresponding to the module type in the preset closed - loop simulation test system, obtain input data through the data bus of the preset closed - loop simulation test system, and output output data through the data bus of the preset closed - loop simulation test system.

[0082] Preferably, the module type corresponding to the algorithm module to be tested connected to the preset closed - loop simulation test system is any one of the replaceable module types; or,

[0083] The module type corresponding to the algorithm module to be tested connected to the preset closed - loop simulation test system is at least two of the replaceable module types.

[0084] When the module type corresponding to the algorithm module to be tested connected to the preset closed - loop simulation test system is any one of the replaceable module types, the present invention can realize the closed - loop continuous operation performance test of a single algorithm module to be tested in the closed - loop simulation test system, thus avoiding the performance problems of a single algorithm module to be tested in the algorithm for autonomous driving to be tested.

[0085] When the module type corresponding to the algorithm module to be tested connected to the preset closed - loop simulation test system is at least two of the replaceable module types, the present invention can realize the closed - loop continuous operation performance test of multiple algorithm modules to be tested in parallel development in the closed - loop simulation test system, thus timely discovering the performance compatibility problems between multiple algorithm modules to be tested in parallel development.

[0086] The user can select the type of the algorithm module to be tested connected to the preset closed - loop simulation test system according to the actual development situation.

[0087] S40: Evaluate the simulation test performance of each algorithm module to be tested in the algorithm for autonomous driving to be tested according to the simulation test status of each algorithm module to be tested in the preset closed - loop simulation test system.

[0088] Exemplarily, the schematic diagram showing the implementation principle of the closed-loop simulation test method for the autonomous driving algorithm provided by the present invention is as Figure 2 shown.

[0089] To better illustrate the implementation process of the closed-loop simulation test method for the autonomous driving algorithm provided by the present invention, the following embodiments are shown for exemplary illustration.

[0090] Embodiment 1

[0091] S10: Obtain the algorithm module to be tested corresponding to the autonomous driving algorithm to be tested.

[0092] S20: If it is determined that the replaceable module type corresponding to the algorithm module to be tested in the preset closed-loop simulation test system is the lateral decision-making module, then the lateral decision-making module of the preset closed-loop simulation test system is turned off through the module switch.

[0093] S30: Connect the algorithm module to be tested to the position of the lateral decision-making module in the preset closed-loop simulation test system through the interface of the algorithm to be tested, and obtain the perception information uploaded by the perception module, as well as the driving task information and driving map information uploaded by the task input module through the data bus of the preset closed-loop simulation test system. After being executed by the algorithm to be tested according to the perception information, driving task information and driving map information, the number of the lateral planner to be called is obtained, and the number of the lateral planner is output through the data bus of the preset closed-loop simulation test system. The lateral planning module in the preset closed-loop simulation test system reads the number of the called lateral scheduler from the data bus and executes it; among them, other functional modules of the preset closed-loop simulation test system operate normally.

[0094] S40: Evaluate the simulation test performance of the algorithm module to be tested according to the simulation test state of the algorithm module to be tested in the preset closed-loop simulation test system.

[0095] Embodiment 2

[0096] S10: Obtain the algorithm module to be tested corresponding to the autonomous driving algorithm to be tested.

[0097] S20: If it is determined that the replaceable module type corresponding to the algorithm module to be tested in the preset closed-loop simulation test system is the longitudinal decision-making module, then the longitudinal decision-making module of the preset closed-loop simulation test system is turned off through the module switch.

[0098] S30: Connect the algorithm module to be tested to the position of the longitudinal decision-making module in the preset closed-loop simulation test system through the interface of the algorithm to be tested, and obtain the perception information uploaded by the perception module, as well as the driving task information and driving map information uploaded by the task input module, through the data bus of the preset closed-loop simulation test system. After being executed by the algorithm to be tested according to the perception information, driving task information and driving map information, obtain the serial number of the longitudinal planner to be called, and output the serial number of the longitudinal planner through the data bus of the preset closed-loop simulation test system. The longitudinal planning module in the preset closed-loop simulation test system reads the serial number of the called longitudinal scheduler from the data bus and executes it; wherein, other functional modules of the preset closed-loop simulation test system operate normally.

[0099] S40: Evaluate the simulation test performance of the algorithm module to be tested according to the simulation test status of the algorithm module to be tested in the preset closed-loop simulation test system.

[0100] Embodiment 3

[0101] S10: Obtain the algorithm module to be tested corresponding to the algorithm for autonomous driving to be tested.

[0102] S20: If it is determined that the replaceable module type corresponding to the algorithm module to be tested in the preset closed-loop simulation test system is the lateral planning module, then turn off its own lateral planning module in the preset closed-loop simulation test system through the module switch.

[0103] S30: Connect the algorithm module to be tested to the position of the lateral planning module in the preset closed-loop simulation test system through the interface of the algorithm to be tested, and obtain the serial number of the lateral planner uploaded by the lateral decision-making module through the data bus of the preset closed-loop simulation test system. After being executed by the algorithm to be tested according to the serial number of the lateral planner, calculate the expected path data, and output the expected path data through the data bus of the preset closed-loop simulation test system. The lateral control module in the preset closed-loop simulation test system reads the expected path data from the data bus and executes it; wherein, other functional modules of the preset closed-loop simulation test system operate normally.

[0104] S40: Evaluate the simulation test performance of the algorithm module to be tested according to the simulation test status of the algorithm module to be tested in the preset closed-loop simulation test system.

[0105] Embodiment 4

[0106] S10: Obtain the algorithm module to be tested corresponding to the algorithm for autonomous driving to be tested.

[0107] S20: If it is determined that the replaceable module type corresponding to the algorithm module to be tested in the preset closed-loop simulation test system is the longitudinal planning module, then turn off its own longitudinal planning module in the preset closed-loop simulation test system through the module switch.

[0108] S30: Connect the algorithm module to be tested to the position of the longitudinal planning module in the preset closed-loop simulation test system through the interface of the algorithm to be tested, and obtain the longitudinal planner number uploaded by the longitudinal decision-making module through the data bus of the preset closed-loop simulation test system. After the algorithm to be tested is executed according to the longitudinal planner number, calculate the expected speed data and expected acceleration data, and output the expected speed data and expected acceleration data through the data bus of the preset closed-loop simulation test system. The longitudinal control module in the preset closed-loop simulation test system reads the expected speed data and expected acceleration data from the data bus and executes them; among them, other functional modules of the preset closed-loop simulation test system operate normally.

[0109] S40: Evaluate the simulation test performance of the algorithm module to be tested according to the simulation test status of the algorithm module to be tested in the preset closed-loop simulation test system.

[0110] Embodiment 5

[0111] S10: Obtain the algorithm module to be tested corresponding to the autonomous driving algorithm to be tested.

[0112] S20: If it is determined that the replaceable module type corresponding to the algorithm module to be tested in the preset closed-loop simulation test system is the lateral control module, then turn off its own lateral control module in the preset closed-loop simulation test system through the module switch.

[0113] S30: Connect the algorithm module to be tested to the position of the lateral control module in the preset closed-loop simulation test system through the interface of the algorithm to be tested, and obtain the expected path data uploaded by the lateral planning module through the data bus of the preset closed-loop simulation test system. After the algorithm to be tested is executed according to the expected path data, calculate the steering wheel angle data, and output the calculated steering wheel angle data through the data bus of the preset closed-loop simulation test system. The steering wheel of the test vehicle reads the steering wheel angle data from the data bus and performs the corresponding steering operation; among them, other functional modules of the preset closed-loop simulation test system operate normally.

[0114] S40: Evaluate the simulation test performance of the algorithm module to be tested according to the simulation test status of the algorithm module to be tested in the preset closed-loop simulation test system.

[0115] Embodiment 6

[0116] S10: Obtain the algorithm module to be tested corresponding to the autonomous driving algorithm to be tested.

[0117] S20: If it is determined that the replaceable module type corresponding to the algorithm module to be tested in the preset closed-loop simulation test system is the longitudinal control module, then turn off its own longitudinal control module in the preset closed-loop simulation test system through the module switch.

[0118] S30: Connect the algorithm module to be tested to the position of the longitudinal control module in the preset closed-loop simulation test system through the interface of the algorithm to be tested, and obtain the expected speed data and expected acceleration data uploaded by the longitudinal planning module through the data bus of the preset closed-loop simulation test system. After the algorithm to be tested executes according to the expected speed data and expected acceleration data, calculate the brake master cylinder pressure data and throttle opening data, and output the brake master cylinder pressure data and throttle opening data through the data bus of the preset closed-loop simulation test system. The brake pedal reads the brake master cylinder pressure data from the data bus and performs corresponding braking operations, or the accelerator pedal reads the throttle opening data from the data bus and performs corresponding acceleration and deceleration operations; among them, other functional modules of the preset closed-loop simulation test system operate normally.

[0119] S40: Evaluate the simulation test performance of the algorithm module to be tested according to the simulation test status of the algorithm module to be tested in the preset closed-loop simulation test system.

[0120] It should be noted that when the number of algorithm modules to be tested corresponding to the algorithm to be tested for autonomous driving is at least two, since in the closed-loop simulation test method for the algorithm for autonomous driving provided by the present invention, the data interaction between the algorithm module to be tested and the preset closed-loop simulation test system is realized through the interface and bus for accessing the algorithm to be tested, multiple algorithms to be tested can also be simultaneously accessed and detected by using the access method provided in the above Embodiments 1-6.

[0121] Among them, when multiple algorithm modules to be tested are connected to the preset closed-loop simulation test system and performance failures occur, the compatibility performance between multiple algorithm modules to be tested can be tested and troubleshot by sequentially removing one or more of the algorithm modules to be tested in the preset closed-loop simulation test system and enabling the corresponding functional modules of the removed algorithm modules to be tested in the preset closed-loop simulation test system.

[0122] In summary, the closed-loop simulation test method for the algorithm for autonomous driving provided by the present invention has the following advantages:

[0123] A closed-loop simulation test method for an autonomous driving algorithm provided by the present invention includes obtaining each algorithm module to be tested corresponding to the autonomous driving algorithm to be tested; for each algorithm module to be tested, determining the module type corresponding to the algorithm module to be tested, and turning off the function module corresponding to the module type in a preset closed-loop simulation test system, where the preset closed-loop simulation test system is a complete closed-loop autonomous driving algorithm that meets the test standard; connecting the algorithm module to be tested to the position of the function module corresponding to the module type in the preset closed-loop simulation test system, obtaining input data through the data bus of the preset closed-loop simulation test system, and outputting output data through the data bus of the preset closed-loop simulation test system; evaluating the simulation test performance of each algorithm module to be tested in the autonomous driving algorithm to be tested according to the simulation test status of each algorithm module to be tested in the preset closed-loop simulation test system, and realizing the closed-loop performance test of single or multiple algorithm modules to be tested in the autonomous driving algorithm under a complete autonomous driving framework, thereby improving the parallel development and test efficiency of the autonomous driving algorithm.

[0124] Although the present invention has been described in detail above with general descriptions, specific embodiments and experiments, it can be modified or improved based on the present invention, which is obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention fall within the scope of the present invention claimed.

[0125] After considering the specification and practicing the present invention herein, those skilled in the art will readily conceive of other embodiments of the present invention. The present invention is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include known common knowledge or conventional technical means in the technical field not disclosed by the present invention. It should be understood that the present invention is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A closed-loop simulation test method for an autonomous driving algorithm, characterized in that, The closed-loop simulation test method includes: S10: Obtain each algorithm module to be tested corresponding to the autonomous driving algorithm to be tested; S20: For each algorithm module to be tested, determine the module type corresponding to the algorithm module to be tested, and turn off the function module corresponding to the module type in the preset closed-loop simulation test system, where the preset closed-loop simulation test system is a complete closed-loop autonomous driving algorithm that meets the test standards; S30: Connect the algorithm module to be tested to the position of the function module corresponding to the module type in the preset closed-loop simulation test system, obtain input data through the data bus of the preset closed-loop simulation test system, and output output data through the data bus of the preset closed-loop simulation test system; S40: Evaluate the simulation test performance of each algorithm module to be tested in the autonomous driving algorithm to be tested according to the simulation test status of each algorithm module to be tested in the preset closed-loop simulation test system.

2. The closed-loop simulation test method according to claim 1, characterized in that The preset closed-loop simulation test system includes a task input module, a perception module, a lateral decision-making module, a longitudinal decision-making module, a lateral planning module, a longitudinal planning module, a lateral control module, a longitudinal control module, a data bus, a module switch, and an access interface for the algorithm to be tested; Among them, the task input module is used to obtain driving task information and driving map information; The perception module is used to obtain perception information; The lateral decision-making module is used to determine the number of the lateral planner to be called according to the driving task information, the driving map information, and the perception information, and then output the number of the lateral planner to the lateral planning module; The longitudinal decision-making module is used to determine the number of the longitudinal planner to be called according to the driving task information, the driving map information, and the perception information, and then output the number of the longitudinal planner to the longitudinal planning module; The lateral planning module is used to call the corresponding lateral planner according to the number of the lateral planner, calculate the expected path data according to the lateral planner, and then output the expected path data to the lateral control module; The longitudinal planning module is used to call the corresponding longitudinal planner according to the number of the longitudinal planner, calculate the expected speed data and the expected acceleration data according to the longitudinal planner, and then output the expected speed data and the expected acceleration data to the longitudinal control module; The lateral control module is used to calculate the steering wheel angle data according to the expected path data, and control the steering wheel to perform corresponding steering operations according to the steering wheel angle data; The longitudinal control module is used to calculate the brake master cylinder pressure data and the throttle opening data according to the expected speed data and the expected acceleration data, and control the brake pedal and the accelerator pedal to perform corresponding braking operations or acceleration / deceleration operations according to the brake master cylinder pressure data and the throttle opening data; The lateral decision-making module, the longitudinal decision-making module, the lateral planning module, the longitudinal planning module, the lateral control module, and the longitudinal control module are replaceable modules that can be replaced by the algorithm module to be tested; The data bus is used to transmit data between the algorithm module to be tested and the preset closed-loop simulation test system; the module switch is used to open and close at least one replaceable module in the preset closed-loop simulation test system; the algorithm access interface to be tested is used to connect the algorithm module to be tested to the preset closed-loop simulation test system.

3. The closed-loop simulation test method according to claim 1, characterized in that The module type corresponding to the algorithm module to be tested connected to the preset closed-loop simulation test system is any one of the replaceable module types; or, The module type corresponding to the algorithm module to be tested connected to the preset closed-loop simulation test system is at least two of the replaceable module types.

4. The closed-loop simulation test method according to claim 2, wherein The types of the lateral planner include a lane change path planner, an intersection path planner, and an empty planner; The lane change path planner uses a fifth-degree polynomial curve to implement path planning for straight roads and curves. The fitting formula of the fifth-degree polynomial curve is as shown in Equation 1: y d (x d ) = a0x d 5 + a1x d 4 + a2x d 3 + a3x d 2 + a4x d + a5 Equation 1 Among them, a3 = 0, a4 = 0, a5 = 0, W represents the lane-changing width, and D represents the lane-changing distance; The intersection path planner uses a Bezier curve to perform path planning for left turns, straight runs, right turns, and U-turns. The fitting formula of the Bezier curve is as shown in Equation 2: Among them, P i is a control point, is the quadratic coefficient, t is a parameter, and each parameter can generate a trajectory point; The empty planner is used to be called when there is no planning task.

5. The closed-loop simulation test method according to claim 2, characterized in that The types of the longitudinal planner include a constant speed cruise planner, a stable car-following planner, a fixed-point parking planner, and an emergency braking planner; The constant speed cruise planner is used to implement the function of the test vehicle cruising at a preset speed; The stable car-following planner is used to implement the function of the test vehicle following the vehicle in front while maintaining a preset spacing interval; The fixed-point parking planner is used to implement the function of the test vehicle stopping at a preset position in front or decelerating to a preset speed range; The emergency braking planner is used to implement the emergency obstacle avoidance function of the test vehicle.

6. The closed-loop simulation test method according to claim 2, wherein The longitudinal control module outputs brake master cylinder pressure data through the first control loop and throttle opening data through the second control loop; The first control loop adopts a single-layer PID plus open-loop control scheme; the single-layer PID controller is used to calculate the expected deceleration data according to the expected vehicle speed data and the actual vehicle speed data of the test vehicle, and then calculate the brake master cylinder pressure data through the look-up table method according to the expected deceleration speed; The second control loop adopts an inner and outer two-layer cascade PID controller scheme; among them, the input of the outer-layer PID controller is the expected vehicle speed data and the actual vehicle speed data of the test vehicle, and the output is the expected acceleration data; the input of the inner-layer PID controller is the expected acceleration data and the actual acceleration data of the test vehicle, and the output is the throttle opening data.

7. The closed-loop simulation test method according to claim 6, wherein When the expected vehicle speed is less than 1 m / s, the integral of the outer-layer PID controller of the second control loop is cleared; when the brake switch is turned on, the integral of the inner layer of the second control loop is cleared.

8. The closed-loop simulation test method according to claim 2, characterized in that The lateral control module uses a pure tracking algorithm to dynamically select a preview point according to the actual vehicle speed of the test vehicle, then calculates the feedforward value of the steering wheel angle, and then uses a model predictive control algorithm to calculate the expected steering wheel angle data according to the feedforward value in combination with the expected path and vehicle state.

9. The closed-loop simulation test method according to claim 5, characterized in that, When the stable car-following planner detects that the distance between the test vehicle and the vehicle in front is higher than the preset car-following distance, the vehicle speed of the vehicle in front plus the preset car-following speed value is directly output as the expected speed data; When the stable car-following planner detects that the distance between the test vehicle and the vehicle ahead is not higher than the preset car-following distance, the stable car-following planner calculates and outputs the expected speed data in real time based on the expected car-following distance input by the user and the actual car-following distance.

10. The closed-loop simulation test method according to claim 2, characterized in that, The longitudinal decision-making module further includes a speed envelope generator, which is used to obtain the road curvature, then calculate the maximum vehicle speed supported by the test vehicle under the road curvature, and when it detects that the expected speed data is greater than the maximum vehicle speed, reduce the vehicle speed of the test vehicle to the maximum vehicle speed.

Images