Method for generating lightweight human-vehicle interaction behavior of test background vehicles in unprotected left turn scenarios
By combining Prescan and Simulink to design a lightweight test background vehicle generation method for unprotected left turns, the problem that traditional models cannot simulate real driver behavior is solved, enabling efficient and economical testing of autonomous driving algorithms and improving the reliability and accuracy of the algorithms.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HEFEI UNIV OF TECH
- Filing Date
- 2025-03-04
- Publication Date
- 2026-04-10
AI Technical Summary
In existing autonomous vehicle testing, traditional micro-traffic models are difficult to simulate real driver behavior, while high-precision models are computationally expensive and economical, and cannot effectively verify the reliability of autonomous driving algorithms.
In the scenario of unprotected left turns, a lightweight generation method was designed by combining Prescan and Simulink to generate the test background vehicle using IDM car-following mode and human-like interaction mode switching. The driver behavior was simulated using TIS sensors and a finite state machine game model to achieve lightweight generation of the background vehicle.
It improves the realism and reliability of autonomous driving algorithm testing, reduces computing resource requirements, achieves high-precision lightweight simulation, has appropriate computing power requirements, and enhances the economy and intelligence of testing.
Smart Images
Figure CN120087078B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of automatic driving vehicle testing methods, in particular to a test background vehicle human-like interaction behavior lightweight generation method under a left turn without protection scene. BACKGROUND
[0002] In recent years, the development of the field of automatic driving is changing rapidly, and virtual simulation based on scenes has become a research hotspot due to its significant advantages such as high efficiency and safety, as one of the three pillars of automatic driving testing and deployment. In the simulation testing of automatic driving, the reliability of the automatic driving vehicle algorithm needs to be verified, which puts forward higher requirements for the background vehicle testing system, so the human-like interaction behavior method of the background vehicle in dangerous working conditions is of great significance. In the past automatic driving vehicle testing, traditional microscopic traffic models (such as IDM, improved ACC) are generally used to generate background vehicle trajectories, and these models rarely consider the behavior characteristics of drivers, and the testing authenticity is limited; or high-precision human-like interaction models (such as LSTM, GAN models based on deep learning) are used, which can help simulate real driving behavior, but require high CPU / GPU energy consumption and computing power, and have poor economy and model interpretability. SUMMARY
[0003] The present application provides a test background vehicle human-like interaction behavior lightweight generation method under a left turn without protection scene to solve the problems existing in the prior art background vehicle testing system.
[0004] In order to achieve the above purpose, the technical scheme adopted by the present application is:
[0005] The test background vehicle human-like interaction behavior lightweight generation method under a left turn without protection scene comprises the following steps:
[0006] Step 1, set the simulation working condition to be a left turn without protection scene on an urban road, the scene has two closed loops, which are respectively set as route 1 and route 2, and route 1 and route 2 have a shared intersection section;
[0007] Step 2, set a plurality of test background vehicles on route 1, each test background vehicle is arranged in sequence along the driving direction, the adjacent test background vehicles on route 1 have equal spacing, and each test background vehicle can circulate on route 1;
[0008] An automatic driving host vehicle is set on route 2, the automatic driving host vehicle can drive on route 2 and turn left at the intersection, so that the test background vehicle and the left-turning automatic driving host vehicle have a conflict at the intersection area;
[0009] Step 3, perform simulation, each test background vehicle has two different control modes during simulation, which are IDM following mode and human-like interaction mode;
[0010] Before the simulation starts, each test background vehicle and the automatic driving host vehicle are stationary on their respective routes;
[0011] After the simulation starts, each test background vehicle is made to circulate on route 1, and the automatic driving host vehicle is made to drive on route 2 and turn left at the intersection; when the test background vehicles do not drive to the intersection area, the IDM following mode is adopted to control the first test background vehicle to drive at the IDM free flow speed, and the remaining test background vehicles follow in the driving direction; when any test background vehicle drives to the intersection area, the human-like interaction mode is switched to, so that the test background vehicle driving to the intersection area randomly performs the behavior of cutting in / letting go / gaming, thereby interfering with the automatic driving host vehicle, and the algorithm performance reliability of the automatic driving host vehicle is verified;
[0012] Step 4, after passing through the intersection area, the test background vehicle is switched to the initial IDM following mode again, and the test background vehicle continues to drive on route 1;
[0013] Step 5, after completing a test cycle according to steps 3 and 4, the test background vehicle continues to drive on route 1; after completing a test cycle according to steps 3 and 4, the automatic driving host vehicle continues to drive on route 2. Until the test background vehicle reaches the intersection area next time, and then steps 3 and 4 are repeated, thereby the algorithm performance reliability of the automatic driving host vehicle is continuously monitored.
[0014] Further, the scene in step 1 is built by Prescan.
[0015] Further, in steps 2-5, each test background vehicle is equipped with a TIS sensor, thereby realizing safe following of the vehicle.
[0016] Further, in steps 3 and 4, the switch module in simulink is used for mode switching of the test background vehicle, the coordinates of the test background vehicle in the global coordinate system are used as input, the switching of different control modes is realized, and then two signals of throttle and brake are output to the dynamics model of the test background vehicle for longitudinal speed control.
[0017] Compared with the prior art, the present application has the following advantages:
[0018] The present application is based on automatic driving simulation test, and is built by Prescan and simulink, realizes a human-like interaction lightweight generation method of test background vehicles in a left turn scene without protection, the algorithm is flexible and simple, can realize human-like decision making, makes the driving interaction behavior more real and accurate, and has the advantages of lightweight, small algorithm model, appropriate required computing power and high precision, can be used as an important means for testing the reliability of automatic driving vehicles, improves the intelligence of the test background vehicle, and verifies the safety and reliability of the automatic driving algorithm. BRIEF DESCRIPTION OF DRAWINGS
[0019] Figure 1 is a method flowchart of an embodiment of the present application.
[0020] Figure 2 is a scene schematic diagram built by an embodiment of the present application. DETAILED DESCRIPTION
[0021] The present application is further illustrated below in conjunction with the accompanying drawings and embodiments.
[0022] As shown in the drawings, Figure 1 the present embodiment discloses a lightweight generation method for test background vehicle human-computer interaction behavior in a left-turn scene without protection, comprising the following steps:
[0023] Step 1, establish a scene.
[0024] As shown in the drawings, Figure 2 Prescan is used to build a simulation working condition for a left-turn scene without protection on an urban road. The scene has two closed loops, which are respectively set as route 1 and route 2, and route 1 and route 2 have a shared intersection section, which has a cross-shaped intersection area and a T-shaped intersection area
[0025] The present embodiment is applicable to a left-turn scene without protection at an urban road intersection, and the design speed is 30-50 km / h. The present embodiment is intended to detect the reliability of the automatic driving host vehicle algorithm in the case of left-turn without protection, so only the cross-shaped intersection area is considered as the conflict area, and the T-shaped intersection area below is not considered as the conflict area.
[0026] Step 2, set a number of test background vehicles on route 1, and set an automatic driving host vehicle on route 2.
[0027] In the present embodiment, a total of 8 test background vehicles are set on route 1, and the 8 test background vehicles are arranged in sequence along the driving direction. The distance between adjacent test background vehicles on route 1 is equal and is 20 m, and each test background vehicle can circulate on route 1. Since the simulation scene is at an urban road intersection, the speed is low, and close-range perception is required, so each test background vehicle team is equipped with a TIS sensor, which is conducive to realizing safe car following.
[0028] The automatic driving host vehicle, as the vehicle under test, can drive on route 2 and turn left at the intersection, so Figure 2 As can be seen from the route shown in the drawings, the test background vehicles and the left-turn automatic driving host vehicle will have a conflict in the cross-shaped intersection area.
[0029] Step 3, perform simulation.
[0030] There are two different control modes for each test background vehicle in the simulation, which are IDM car-following mode and human-like interaction mode.
[0031] Before the simulation starts, each test background vehicle and the automatic driving host vehicle are stationary on their respective routes.
[0032] After the simulation starts, each test background vehicle is made to circulate on route 1, and the automatic driving host vehicle is made to drive on route 2 and turn left at the intersection; when the test background vehicles do not drive to the intersection area, the IDM car-following mode is used to control the first test background vehicle to drive at the IDM free-flow speed, and the remaining test background vehicles follow in the driving direction; when any test background vehicle drives to the intersection area, the human-like interaction mode is switched on to make the test background vehicle that drives to the intersection area randomly perform the behavior of cutting in, yielding, or game playing, thereby interfering with the automatic driving host vehicle and testing the reliability of the algorithm performance of the automatic driving host vehicle.
[0033] In this embodiment, in the intersection area, the decision-making behavior of the test background vehicle will interfere with the automatic driving host vehicle, which needs to judge the behavior (cutting in, yielding, or game playing) of the test background vehicle according to its position, speed, and other information, and use it as input to make decisions.
[0034] Specifically, in this embodiment, the specific setting process of the IDM car-following mode of the test background vehicle is as follows:
[0035] The IDM formula is implemented using the MATLAB Function Block in Simulink, and the speed of the current test background vehicle, the speed of the preceding vehicle, and the distance from the preceding vehicle are input into the IDM controller, and then the acceleration is calculated by the formula.
[0036] The information of the preceding vehicle in the test background vehicle is transmitted by the TIS sensor (in the initial state, the first vehicle has no preceding vehicle information input and drives at the IDM free-flow speed), and each test background vehicle is equipped with two TIS sensors (short and long distances) and sets the detection range to ensure coverage of the following distance (short distance 50m, 120°, long distance 100m, 12°) and direction (both are the front of the vehicle).
[0037] In the IDM car-following mode, the safety time T is set to 1.6s, the static safety distance s0 is set to 4m, and the maximum acceleration is set to 1.0m / s 2 .
[0038] In the test background vehicle, an independent subsystem is created for each rear vehicle, including an input signal module (front vehicle distance and speed, obtained by TIS sensor), a signal processing module (relative speed and relative distance calculation, calculated by MATLAB Subtract module), an IDM controller (MATLAB Function Block connected to the input signal), a vehicle dynamics model (Integrator module for acceleration, speed integration, and displacement), and an independent controller for each rear vehicle to achieve signal connection, thereby realizing the test background vehicle platoon following.
[0039] In this embodiment, the specific setting process of the human-like interaction mode of the test background vehicle is as follows:
[0040] The behavior decision layer of the human-like interaction mode is a finite state machine (FSM) + game model. The state transition condition is set to three (triggering overtaking: when the main vehicle is far from the conflict point, the background vehicle takes the probability P agg acceleration; triggering yielding: when the main vehicle approaches the conflict point and the speed is fast, the test background vehicle takes the probability P yield deceleration; game mode: based on the relative distance and speed of both sides, the Nash equilibrium is calculated, and the acceleration is dynamically adjusted) The game model is realized by MATLAB Function code. The control execution layer is based on the improved IDM model, and the acceleration correction of the game decision is added. Then a random number is generated by the Random Number Generator module (MATLAB Function code) to trigger different decisions.
[0041] The input of the test background vehicle intelligent controller includes: input main vehicle position / velocity and its own state (position, velocity), and the decision is made using the finite state machine (Stateflow) game model (MATLAB Function). The control layer is an improved IDM longitudinal control, and the throttle / brake instruction is output to the vehicle dynamics model.
[0042] In the human-like interaction mode, the state transition logic is realized by the Stateflow module; the conflict point distance and random trigger are responded; the MATLAB Function Block embeds the game model and IDM formula; the Vehicle Dynamics uses the vehicle model provided by Prescan; and the vehicle state is synchronized through the Prescan-Simulink interface module.
[0043] In this embodiment, position triggering is adopted, that is, when the test background vehicle does not reach the trigger area of the intersection, the IDM model is used for control, and when any test background vehicle drives to the trigger area of the intersection, it is switched to the human-like interaction mode for random decision. The specific switching process is as follows:
[0044] Prescan setup: Define trigger area in the scene and set trigger point coordinates. Test background vehicles use TIS sensor to output their absolute position, and the host vehicle is configured with sensors (camera or radar) to perceive the test background vehicles.
[0045] Simulink model setup: Read the test background vehicle position from Prescan, determine whether the test background vehicle has entered the trigger area (Relational Operator module), and switch the mode using the Switch module. Input 1 (default mode): IDM following model output acceleration. Input 2 (trigger mode): Intelligent game model output acceleration. Control signal: Boolean signal generated by position trigger, and avoid repeated triggering to ensure that once triggered, the mode remains switched.
[0046] Each test background vehicle processes position detection and mode switching independently to ensure that its trigger logic and control model do not interfere with each other.
[0047] Finally, perform Prescan and Simulink co-simulation. Prescan and Simulink interface binding ensures that vehicle and sensor signals are correctly mapped. Then route the position signal (X, Y) of each background vehicle to the corresponding trigger judgment module, and send the acceleration instruction after mode switching to the vehicle dynamics module to realize mode switching.
[0048] Step 4, After passing through the intersection area, switch the test background vehicle to the initial IDM following mode again, and let the test background vehicle continue to drive on Route 1.
[0049] Specifically, the overall switching logic for the second switch is: initial mode, IDM following model (default). First switch, enter trigger area → human-like interaction mode. Second switch, enter second trigger area → switch back to IDM mode.
[0050] Therefore, after passing through the conflict area, the test background vehicle is switched to the initial IDM following mode again, and continues to drive on Route 1, i.e. define the second trigger area in Prescan and use the Simulink switch module for trigger judgment.
[0051] Step 5, After completing a test cycle according to steps 3 and 4, let the test background vehicle continue to drive on Route 1; after completing a test cycle according to steps 3 and 4, let the host vehicle continue to drive on Route 2. Repeat steps 3 and 4 when the test background vehicle reaches the intersection area again, and switch the mode again. This cycle continues to monitor the performance and reliability of the host vehicle's algorithm.
[0052] The preferred embodiments of the present application are described in detail above with reference to the accompanying drawings, and the embodiments described in the present application are merely a description of the preferred embodiments of the present application, and are not intended to limit the concept and scope of the present application. In the above specific embodiments, various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction, and such combination should also be considered as disclosed by the present disclosure, as long as it does not deviate from the concept of the present application. In order to avoid unnecessary repetition, the present application will not further describe various possible combinations.
[0053] The present application is not limited to the specific details described in the above embodiments, and various modifications and improvements to the technical solutions of the present application made by those skilled in the art within the scope of the technical concept of the present application and without departing from the design concept of the present application should fall within the protection scope of the present application. The technical content claimed by the present application has been fully recorded in the claims.
Claims
1. A method for lightweight generation of human-like interaction behavior of test background vehicles in unprotected left turn scenarios, characterized in that, The method comprises the following steps: Step 1: Set the simulation working condition as an unprotected left turn scene on an urban road, and the scene has two closed loops, which are respectively set as route 1 and route 2, and route 1 and route 2 have a shared intersection section; Step 2: A plurality of test background vehicles are arranged on route 1, and the test background vehicles are arranged in sequence along the driving direction, the adjacent test background vehicles on route 1 have equal spacing, and each test background vehicle can circularly drive on route 1; an automatic driving host vehicle is arranged on route 2, the automatic driving host vehicle can drive on route 2 and turn left at the intersection, and the test background vehicles and the left-turn automatic driving host vehicle have a conflict at the intersection area; Step 3: Simulation is performed, each test background vehicle has two different control modes, which are IDM following mode and human-like interaction mode; before the simulation starts, each test background vehicle and the automatic driving host vehicle are stationary on the respective routes; after the simulation starts, each test background vehicle circularly drives on route 1, and the automatic driving host vehicle drives on route 2 and turns left at the intersection; when the test background vehicle does not drive to the intersection area, the IDM following mode is adopted to control the first test background vehicle to drive at the IDM free flow speed, and the remaining test background vehicles follow in the driving direction; when any test background vehicle drives to the intersection area, the human-like interaction mode is switched to, so that the test background vehicle driving to the intersection area randomly performs the behavior of cutting in, yielding or game, thereby interfering with the automatic driving host vehicle, and the algorithm performance reliability of the automatic driving host vehicle is verified; Step 4: After passing through the intersection area, the test background vehicle is switched to the initial IDM following mode again, and the test background vehicle continues to drive on route 1; Step 5: After completing a test cycle according to steps 3 and 4, the test background vehicle continues to drive on route 1; after completing a test cycle according to steps 3 and 4, the automatic driving host vehicle continues to drive on route 2; until the test background vehicle arrives at the intersection area next time, and then steps 3 and 4 are repeated, thereby the algorithm performance reliability of the automatic driving host vehicle is continuously monitored; In steps 3 and 4, the mode switching of the test background vehicle adopts a switch module in simulink, and the coordinates of the test background vehicle in the global coordinate system are input, the switching of different control modes is realized, and then two signals of throttle and brake are output to the dynamics model of the test background vehicle for longitudinal speed control.
2. The method of claim 1, wherein the method further comprises: The scene in step 1 is built by Prescan.
3. The method of claim 1, wherein the method further comprises: determining a first position of the test vehicle; determining a second position of the test vehicle; and determining a third position of the test vehicle. In steps 2-5, each test background vehicle is equipped with a TIS sensor, thereby realizing safe following of the vehicle.