Low-cost dynamic triggering test method and system for automatic driving multi-target vehicle conflict sequence in closed field

Abstract

The application discloses a kind of closed field automatic driving multi-target vehicle conflict sequence low-cost dynamic trigger test method and system, belong to automatic driving test technical field.The problem that single test is difficult to efficiently and low cost to verify the continuous processing ability of multiple vehicle conflicts of automatic driving vehicle VUT is solved.The gist is: predefine VUT route and conflict point corresponding target vehicle VTi;VUT basic speed-time curve is obtained by probing operation;For each VTi, pre-trigger position Ai is calculated in reverse according to VUT dynamic speed curve and VTi calibration time ti, and the distance L and time margin TTC of VUT to conflict point are calculated in real time;When VUT reaches Ai or TTC is less than or equal to the preset threshold Ti of corresponding conflict point (including VTi response time and ti), trigger stationary VTi to start manufacturing conflict;Sequentially trigger each VTi to form continuous conflict sequence.It is used for single test to efficiently and low cost verify the multi-target conflict processing capability of VUT.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving test verification. More specifically, the present application relates to a low-cost dynamic trigger test method and system for closed field automatic driving multi-target vehicle conflict sequence. BACKGROUND

[0002] The rapid development of automatic driving technology puts high requirements on the safety and reliability of vehicles in various complex traffic scenarios. Closed field testing is an important means to verify the performance and safety boundaries of automatic driving systems, and simulating common multi-vehicle interaction conflict scenarios in real roads (such as continuous intersection conflict, multi-vehicle merging conflict, etc.) is particularly critical. Such scenarios can effectively test the system's perception, decision-making, path planning, and emergency avoidance capabilities. However, it is still a significant challenge to efficiently, cost-effectively, and reliably construct continuous, dynamic multi-target vehicle conflict sequences in a closed field to fully cover the complex situations that the system may encounter.

[0003] The current mainstream method for closed field testing of multi-vehicle conflict scenarios mainly relies on pre-set fixed trigger points or manual remote control of target vehicles. These methods can usually only test single or a small number of conflict points independently. To construct a sequence of multiple conflict scenarios occurring in sequence, multiple independent test runs are often required. This not only leads to low test efficiency and prolonged test cycle, but also significantly increases the time cost of the field, equipment (such as target vehicles), and personnel, making it difficult to meet the rapid iteration development test requirements.

[0004] Further, the above-mentioned methods also have the problem of insufficient dynamic adaptability. After the first conflict target is encountered and the avoidance behavior (such as braking, turning) is taken by the measured vehicle, its speed trajectory will change significantly. If the subsequent conflict target vehicle still triggers according to the pre-set fixed position or time trigger, it cannot accurately simulate the dynamic process of the measured vehicle encountering subsequent conflicts after the avoidance of the new state (such as reduced speed) in the real world. This may lead to distortion of the test scenario, either too lenient conflict (measured vehicle speed has been reduced), or too severe conflict (triggered too early, measured vehicle has not recovered to the expected state), which cannot accurately evaluate the system's ability to handle continuous dynamic conflicts.

[0005] In addition, even if dynamic triggering is attempted in the prior art, there are often problems of precision and reliability. For example, a single initial speed curve is used to predict the triggering point, which cannot adapt to changes in the trajectory of the vehicle under test due to pre-sequenced evasive behavior; or only a single real-time collision time threshold is used to determine the triggering time, ignoring the response time and movement time required by the target vehicle itself from a stationary start, which may result in the target vehicle triggering too late to accurately reach the predetermined conflict point, or triggering too early to cause an invalid or unexpected conflict scenario. These factors limit the possibility of efficiently, cost-effectively, and accurately completing a continuous multi-target conflict sequence verification in a single test run. Therefore, there is an urgent need for a low-cost test method that can dynamically, accurately, and efficiently trigger a multi-target vehicle conflict sequence. SUMMARY

[0006] An object of the present application is to solve at least the above problems and provide at least the advantages to be described later.

[0007] Another object of the present application is to provide a low-cost dynamic triggering test method for automatic driving multi-target vehicle conflict sequences in a closed field, which can complete multi-target vehicle continuous conflict sequence verification in a single test, greatly reducing costs and increasing efficiency; dynamically calculates the triggering point of the subsequent target vehicle according to the actual speed of the VUT after evasive behavior, improving the authenticity of the scene; combines the pre-trigger position (Ai) and the real-time collision time (TTC) double threshold to determine and compensate for the response time, ensuring accurate triggering of the conflict.

[0008] In order to achieve these objects and other advantages according to the present application, a low-cost dynamic triggering test method for automatic driving multi-target vehicle conflict sequences in a closed field is provided, comprising:

[0009] S1, defining the driving route of the vehicle under test VUT and a plurality of conflict points in a closed test field, each conflict point corresponding to an intervention position of a target vehicle VTi, i = 1, 2, 3, …, and executing the following data collection loop:

[0010] First, run a thorough investigation: control the VUT to drive all the way in a target-free scene, and generate a basic speed-time curve for each position point;

[0011] S2, for the i-th VTi, when i = 1, based on the basic speed-time curve and the nominal movement time ti of VTi to reach the corresponding conflict point, the pre-trigger position Ai of VUT at time ti before reaching the conflict point is calculated inversely; when i ≥ 2, after the VUT evades behavior at the i-1th conflict point, the actual speed-time curve is generated according to the actual speed and time data of the speed stabilization segment, and the pre-trigger position Ai is recalculated;

[0012] At the same time, the position and speed of the VUT are acquired in real time, the real-time distance L and time margin TTC of the VUT to the conflict point are calculated in real time, and TTC=L / VUT real-time speed;

[0013] S3, when the VUT reaches the pre-trigger position Ai or the real-time TTC≤ the preset threshold Ti of the conflict point, a trigger instruction is sent to the corresponding target vehicle VTi, the VTi starts from static and drives to the conflict point according to the preset trajectory, and forms a preset conflict scene with the VUT;

[0014] S4, according to the driving sequence of the VUT, each VTi is dynamically triggered in turn to form a continuous conflict sequence, and multiple scene verifications are completed in a single test;

[0015] The preset threshold Ti=VTi response time+ti, and the VTi response time is 50-200 ms.

[0016] Preferably, the risk avoidance behavior intensity monitoring and compensation mechanism is implemented during the recalculation of the pre-trigger position Ai in step S2, specifically including: the braking deceleration value of the VUT at the previous conflict point is collected in real time by fusing the vehicle-mounted inertial measurement unit and the wheel speed sensor, and when the measured deceleration value of any previous conflict point exceeds the limit value 3.5 m / s 2 , the compensation mechanism is activated:

[0017] The arithmetic mean value of all over-limit decelerations in the previous conflict point is calculated, and the margin compensation value is generated according to the following rules:

[0018] 3.5 m / s 2 ≤ the arithmetic mean value of all over-limit decelerations <4.5 m / s 2 , the margin compensation value is 100 ms;

[0019] 4.5 m / s 2 ≤ the arithmetic mean value of all over-limit decelerations <5.5 m / s 2 , the margin compensation value is 150 ms;

[0020] The arithmetic mean value of all over-limit decelerations >5.5 m / s 2 , the margin compensation value is 300 ms;

[0021] The generated margin compensation value is added to the preset threshold Ti of all subsequent conflict points, and the compensation continues to be effective until the end of the current test cycle.

[0022] Preferably, the method for acquiring the designated motion time ti of the VTi to the corresponding conflict point in step S2 is as follows:

[0023] Before the test, the no-load dynamics of any target vehicle VTi is calibrated, the average acceleration of the vehicle from static to the conflict point is measured, and based on the acceleration and the measured distance of VTi and the corresponding conflict point, the calibrated motion time of VTi to reach the corresponding conflict point is calculated.

[0024] Preferably, when the pre-trigger position condition and the real-time TTC condition are activated at the same time, the real-time TTC trigger instruction is selected as the effective trigger signal, and the other trigger channel is immediately locked until the current conflict scenario ends.

[0025] Preferably, a first control system is installed on the measured vehicle VUT, the first control system includes an inertial navigation positioning module, a calculation module and a first communication module arranged on the VUT, the inertial navigation positioning module at least includes an on-board inertial measurement unit and a wheel speed sensor, the calculation module is in communication connection with the inertial navigation positioning module and the first communication module, a second control system is installed on any target vehicle, the second control system includes a throttle speed execution mechanism, a VT control module and a second communication module, the VT control module is in communication connection with the throttle speed execution mechanism and the second communication module, and the first communication module is in communication connection with the second communication module.

[0026] Preferably, the throttle speed execution mechanism includes:

[0027] An electric travel push rod, one end of which is rotatably connected to the vehicle body of the target vehicle;

[0028] A throttle pedal, one end of which is rotatably connected to the other end of the electric travel push rod, and the other end is rotatably connected to the vehicle body of the target vehicle;

[0029] The VT control module controls the travel slope of the electric travel push rod according to the trigger signal.

[0030] Preferably, the load rate of the calculation module of the first control system is monitored in real time, when the load rate is greater than 90% for 3-5s, the real-time TTC trigger channel is automatically closed, the pure pre-trigger position trigger mode is adopted, and a hardware expansion warning is generated and a degradation event is recorded.

[0031] Preferably, the low-cost dynamic trigger test method for the automatic driving multi-target vehicle conflict sequence in the closed field further includes establishing a multi-target vehicle trigger conflict arbitration mechanism, and the following steps are performed:

[0032] The pre-trigger position influence area associated with each conflict point is divided by electronic fence technology, and the pre-trigger position influence area is a circular area with the pre-trigger position Ai as the center and a radius of 4-6m;

[0033] According to the current driving direction of VUT, the target vehicles are given priority according to the following rules:

[0034] First level: target vehicle corresponding to longitudinal conflict point coinciding with VUT trajectory extension line;

[0035] Second level: target vehicle corresponding to transverse conflict point orthogonal to VUT trajectory;

[0036] Third level: target vehicle corresponding to oblique conflict point forming an acute angle with VUT trajectory;

[0037] If multiple target vehicle pre-trigger position influence zones are detected to overlap, then the low-priority target vehicle trigger sequence is forced to be delayed until the high-priority target vehicle completes the conflict;

[0038] For the delayed trigger target vehicle, for each delayed priority, the calibrated motion time ti is reduced by 10%, and the compressed calibrated motion time shall not be lower than 80% of the time required for the target vehicle to accelerate from rest to the corresponding conflict point at maximum acceleration, otherwise the trigger of the target vehicle is cancelled.

[0039] The present application further claims to protect the test system of the low-cost dynamic trigger test method of the automatic driving multi-target vehicle conflict sequence in the enclosed field, comprising:

[0040] The first control system is installed on the vehicle VUT to be tested, comprising:

[0041] The inertial navigation positioning module comprises at least an on-board inertial measurement unit and a wheel speed sensor, for real-time acquisition of VUT position, speed and deceleration data;

[0042] The calculation module is used to perform:

[0043] Based on the pre-defined VUT driving route and conflict point sequence, the basic speed-time curve is generated when the first target vehicle-free operation is performed;

[0044] When i=1, the pre-trigger position of the first conflict point is calculated according to the basic speed-time curve and the calibrated motion time ti of VTi to reach the corresponding conflict point, and when i≥2, the pre-trigger position Ai of each target vehicle is dynamically calculated according to the actual speed-time curve after the previous conflict avoidance and the calibrated motion time ti of VTi to reach the corresponding conflict point;

[0045] The real-time distance L and time margin TTC of VUT to reach the ith conflict point are calculated in real time;

[0046] The first communication module is used to send the trigger instruction;

[0047] The second control system is installed on each target vehicle, comprising:

[0048] The VT control module is used to receive the trigger instruction and generate a control signal;

[0049] a throttle speed actuator, which drives the target vehicle to move along the preset trajectory according to the control signal;

[0050] a second communication module, which establishes a communication link with the first communication module;

[0051] a dynamic triggering module integrated in the computing module, for:

[0052] when the VUT reaches the pre-trigger position Ai or the TTC is less than the preset threshold Ti corresponding to the conflict point, sending a triggering instruction to the corresponding target vehicle VTi; if the pre-trigger position and the TTC condition are activated at the same time, the TTC triggering is forced to be selected and the other channel is locked;

[0053] wherein the calibration motion time ti is obtained through no-load dynamics calibration, and the calibration method comprises: measuring the average acceleration of the target vehicle VTi from static acceleration to the conflict point, and calculating the calibration motion time of VTi to the corresponding conflict point based on the acceleration and the measured distance between VTi and the corresponding conflict point.

[0054] Preferably, the computing module further performs:

[0055] Performing risk avoidance intensity monitoring: when the measured deceleration of the current sequence conflict point exceeds the limit, generating a margin compensation value and superimposing it into the preset threshold Ti of all subsequent conflict points;

[0056] Load monitoring: but when the computing module load rate is greater than 90% for 3-5s, the TTC triggering channel is closed and a hardware expansion warning is generated, and a degradation event is recorded.

[0057] The present application at least includes the following beneficial effects:

[0058] Firstly, the present application collects the speed and time data of the VUT after avoiding risk in real time, dynamically calculates the pre-trigger position Ai based on the actual speed-time curve, and combines the real-time collision time TTC with the preset threshold Ti double triggering condition, so that multiple target vehicles can be triggered in a continuous conflict sequence in a single test, improving the test efficiency while reducing the cost of the site;

[0059] Secondly, the present application also monitors the overrun value of the braking deceleration of the previous conflict point in real time and dynamically generates a margin compensation value, which is superimposed into the threshold T of the subsequent conflict points, effectively eliminating the time sequence cumulative deviation caused by strong risk avoidance behavior, and ensuring the spatiotemporal consistency of the continuous conflict scene;

[0060] Thirdly, the present application adopts a split type control system architecture: the VUT end sends a triggering instruction in real time through an inertial navigation positioning module and a computing module, the target vehicle end responds to the control through a throttle speed actuator, and the two systems are synchronized through a communication module to ensure that the end-to-end delay control of the triggering instruction to the vehicle execution is within 200ms.

[0061] Fourthly, the application further sets a calculation module load monitoring mechanism, which automatically closes the real-time TTC calculation channel when the load rate is greater than 90% for 3-5 seconds, degrades to a pure pre-trigger position mode and generates a hardware expansion warning, prevents system overload failure and improves the robustness of the test process;

[0062] Fifthly, the application further establishes a three-level conflict priority arbitration mechanism (vertical > horizontal > diagonal), which forces the low-priority target vehicle in the overlapping influence area to delay triggering and dynamically compresses its calibration motion time t, thereby ensuring the authenticity of the core scene timing while avoiding scene failure caused by spatial conflicts.

[0063] Other advantages, objects and features of the application will be apparent from the following description, and will be understood by those skilled in the art upon reading and understanding the application. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 is a schematic diagram of three target vehicle conflict sequences in one technical solution of the application;

[0065] Figure 2 is a schematic diagram of the first control system and the second control system in another technical solution of the application;

[0066] Figure 3 is a structural schematic diagram of the accelerator speed execution mechanism in another technical solution of the application. DETAILED DESCRIPTION

[0067] The application will be further described in detail below with reference to the accompanying drawings, so that those skilled in the art can implement the application according to the description.

[0068] It should be understood that the terms such as "have", "contain" and "include" used herein do not exclude the presence or addition of one or more other elements or combinations thereof.

[0069] The application provides a low-cost dynamic triggering test method for automatic driving multi-target vehicle conflict sequences in a closed field, comprising:

[0070] S1, defining a driving route of a vehicle under test (VUT) and a plurality of conflict points in a closed test field, each conflict point corresponding to an intervention position of a target vehicle (VTi), i=1,2,3……, and executing the following data acquisition cycle:

[0071] Firstly, the basic speed-time curve of each position point is generated by controlling the VUT to drive through the whole route without target vehicles;

[0072] S2, for the ith VTi, when i=1, based on the basic speed-time curve and the nominal motion time ti of VTi reaching the corresponding conflict point, the pre-trigger position Ai of VUT reaching the conflict point at ti is reversely calculated; when i≥2, after the risk avoidance behavior of VUT passing through the i-1th conflict point, the actual speed-time curve is generated according to the actual speed and time data of the speed stabilization segment, and the pre-trigger position Ai is recalculated;

[0073] At the same time, the position and speed of VUT are acquired in real time, the real-time distance L and time margin TTC of VUT reaching the conflict point are calculated in real time, and TTC=L / real-time speed of VUT;

[0074] S3, when VUT reaches the pre-trigger position Ai or real-time TTC≤the preset threshold Ti of the conflict point, a trigger instruction is sent to the corresponding target vehicle VTi, VTi starts from static and drives to the conflict point according to the preset trajectory, and forms a preset conflict scene with VUT.

[0075] S4, according to the driving sequence of VUT, each VTi is dynamically triggered in turn to form a continuous conflict sequence, and multiple scene verifications are completed in a single test.

[0076] Wherein, the preset threshold Ti=VTi response time+ti; the VTi response time is 50-200 ms.

[0077] The above technical scheme realizes low-cost testing of multiple target vehicle conflict sequences through double trigger conditions and actual speed-time curve dynamic decision. First, the driving route of the measured vehicle VUT and multiple conflict points are predefined in a closed field, and each conflict point is associated with a static target vehicle VTi. A scene model is constructed through data acquisition: according to the speed-time data of the vehicle without target, the basic speed-time curve of VUT is generated.

[0078] For the trigger of each target vehicle VTi (i=1, 2, 3…), an adaptive pre-trigger position calculation is adopted: for the first target vehicle (i=1), based on the basic speed-time curve and the nominal motion time ti of VTi, the pre-trigger position Ai of VUT reaching the conflict point at ti is reversely calculated; for the subsequent target vehicle (i≥2), the Ai is recalculated based on the actual speed-time curve after the previous risk avoidance, to ensure that the trigger point matches the real-time state of VUT. The real-time distance L and time margin TTC (TTC=L / real-time speed) of VUT to the conflict point are calculated by synchronously monitoring through real-time double channels: obtaining the position and speed of VUT. Finally, through the dynamic trigger logic: when VUT reaches Ai or TTC≤the preset threshold Ti (Ti=VTi response time+t, the response time is 50-200 ms) of the corresponding conflict point, VTi is triggered to start from static (response device: electromagnetic clutch+DC motor driving system), and drives to the conflict point according to the preset trajectory to form an accurate conflict scene.

[0079] As Figure 1 , according to the above technical solution, taking three target vehicle collision sequences as an example, a specific workflow is as follows:

[0080] Initialization and data collection: VUT performs the first target-free vehicle operation according to the predetermined route, and generates the basic speed-time curve (sampling frequency 100Hz) according to the recorded full-range data.

[0081] Dynamic trigger execution:

[0082] VT1 trigger: based on the basic curve and VT1 calibration time t1, calculate the pre-trigger position A1 (A' in the figure). In real-time monitoring, when VUT reaches A1 or TTC≤T (T= response time+ t1), trigger VT1 to start.

[0083] VT2 trigger: recalculate A2 (B' in the figure) based on the actual curve after VT1 avoidance. If the actual curve shows that the speed of VUT is reduced, then the position of A2 is moved forward compared with the calculated value of the basic curve. The real-time TTC threshold is updated synchronously to T= response time+ t2 value of VT2; the actual speed-time curve is generated by collecting at least 6 points after the speed is stable after the VT1 collision point avoidance.

[0084] VT3 trigger: similarly, calculate A3 (C' in the figure) based on the actual curve after VT2 avoidance, realize time sequence nested trigger.

[0085] Scene closed loop: trigger VT1, VT2 and VT3 in turn according to the driving order of VUT, complete three scene continuous verification in single test, and do not need to interrupt or repeat the test.

[0086] The above technical solution can accurately generate multiple target vehicle continuous collision sequences through real-time collection of avoidance speed curve and dynamic calculation of pre-trigger position based on the curve, combined with real-time TTC double condition trigger, which can compress the traditional N times independent test scene to 1 time, improve the test efficiency by several times and reduce the site cost. At the same time, it also improves the scene authenticity: for subsequent target vehicles (i≥2), the pre-trigger position Ai is dynamically updated based on the actual speed curve after the previous avoidance in operation, which ensures the accurate matching of the collision time sequence with the real-time motion state of VUT (such as the trajectory after braking deceleration), and avoids the scene distortion caused by fixed trigger point. In addition, the double trigger condition design is compatible with vehicle positioning error and response delay, which guarantees the collision space-time accuracy through the preset threshold Ti (including 50~200ms response time compensation), and the trigger reliability is more than 95%.

[0087] In one of the technical solutions, the risk aversion behavior intensity monitoring and compensation mechanism is implemented in the process of recalculating the pre-trigger position Ai in step S2. Specifically, the braking deceleration value of the VUT at the pre-sequence conflict point is collected in real time by fusing the vehicle-mounted inertial measurement unit and the wheel speed sensor. When the measured deceleration value of any pre-sequence conflict point exceeds the limit value of 3.5 m / s 2 , the compensation mechanism is activated.

[0088] The arithmetic mean of all over-limit decelerations at the pre-sequence conflict point is calculated, and the margin compensation value is generated according to the following rules:

[0089] 3.5 m / s 2 ≤ arithmetic mean of all over-limit decelerations < 4.5 m / s 2 , the margin compensation value is 100 ms;

[0090] 4.5 m / s 2 ≤ arithmetic mean of all over-limit decelerations < 5.5 m / s 2 , the margin compensation value is 150 ms;

[0091] arithmetic mean of all over-limit decelerations > 5.5 m / s 2 , the margin compensation value is 300 ms;

[0092] The generated margin compensation value is added to the preset threshold Ti of all subsequent conflict points, and the compensation continues to be effective until the end of the current test cycle.

[0093] The above technical solution further increases the risk aversion behavior intensity monitoring and compensation mechanism. The braking intensity of VUT in the pre-sequence conflict is quantified in real time by vehicle-mounted sensors, and the subsequent trigger threshold is dynamically adjusted to eliminate timing accumulation bias. The specific implementation includes:

[0094] Intensity monitoring: the actual braking deceleration value of each pre-sequence conflict point is collected by fusing the inertial measurement unit (IMU can choose Bosch BMI088) and the wheel speed sensor (can choose Hella SGM sensor) on the VUT vehicle. Set 3.5 m / s 2 as the over-limit threshold, and 3.5 m / s 2 m / s as the over-limit threshold corresponding to moderate emergency braking intensity. According to different test requirements, it can be adjusted appropriately.

[0095] Dynamic compensation: when the measured deceleration of any pre-sequence conflict point exceeds 3.5 m / s 2 , the arithmetic mean of all over-limit decelerations is calculated, and the margin compensation value is generated according to the gradient rules:

[0096] When the average value is between 3.5 and 4.5 m / s 2 , compensate 100 ms (light compensation);

[0097] Average value in 4.5~5.5m / s 2 Compensation 150ms (moderate compensation) when;

[0098] Average value in >5.5m / s 2 Compensation 300ms (severe compensation, corresponding to extreme braking).

[0099] Threshold correction: add the compensation value to the preset threshold T1 of all subsequent conflict points (i.e. T'= original T + compensation value), and the compensation continues to take effect until the end of the current test cycle.

[0100] According to the above technical scheme, taking the case of two consecutive conflict scenarios (VT1→VT2) as an example, a specific implementation process is as follows:

[0101] VT1 trigger and risk monitoring:

[0102] After VT1 is triggered, VUT takes braking at the conflict point, and the deceleration value measured by the fusion of IMU and wheel speed sensor exceeds 3.5m / s 2 limit value, but less than 4.5m / s 2 .

[0103] System activates compensation mechanism: calculate the average value of the current over-limit deceleration, which falls into the interval of 3.5~4.5m / s 2 , generate a 100ms margin compensation value.

[0104] VT2 trigger threshold correction:

[0105] The original preset threshold T2=VT2 response time+calibration motion time t2.

[0106] After adding the compensation value, the new threshold T2'= T2+100ms. The preset threshold of all subsequent conflict points is added by 100ms.

[0107] Dynamic trigger execution:

[0108] According to the VT1 speed curve after avoiding in the conflict scenario incremental learning, calculate the VT2 pre-trigger position A2 (because of braking, the speed of VUT is reduced, and the position of A2 is moved forward).

[0109] In real-time monitoring, when VUT reaches A2 or TTC≤T2', VT2 is triggered, which responds 100ms earlier than the original threshold, offsetting the effect of VUT travel shortening caused by previous braking.

[0110] The technical solution eliminates the time sequence interference of strong risk avoidance behavior, dynamically generates a compensation value by monitoring the brake deceleration overrun event in real time, corrects the subsequent conflict point trigger threshold, solves the problem of early triggering of the subsequent target vehicle caused by VUT sudden deceleration, and guarantees the space-time consistency of the continuous conflict scene. At the same time, the hierarchical compensation mechanism accurately matches the brake intensity influence: light braking (4.5 m / s 2 ) compensates for 100 ms, and heavy braking (> 5.5 m / s 2 ) compensates for 300 ms, so that the compensation amount is positively correlated with the risk avoidance intensity, avoiding overcompensation or undercompensation. The design of the compensation value being effective until the end of the test can accumulate the chain deviation between multiple conflict points and improve the restoration degree of continuous scenes.

[0111] In one of the technical solutions, the method for obtaining the nominal motion time ti of VTi reaching the corresponding conflict point in step S2 is as follows:

[0112] Before the test, the empty dynamics of any target vehicle VTi is calibrated, the average acceleration of VTi from static acceleration to the conflict point is measured, and the nominal motion time of VTi reaching the corresponding conflict point is calculated based on the acceleration and the measured distance between VTi and the corresponding conflict point. By calibrating the empty dynamics of the target vehicle VTi, the average acceleration of VTi from static acceleration to the conflict point is measured, and the nominal motion time ti is accurately calculated based on the measured distance, which eliminates the motion time deviation caused by individual performance differences (such as motor response and tire wear) of the target vehicle, ensures the space-time accuracy of the preset threshold Ti, controls the conflict scene trigger error within ±5%, and guarantees the space-time consistency of the multi-target vehicle conflict sequence.

[0113] In one of the technical solutions, when the pre-trigger position condition and the real-time TTC condition are activated at the same time, the real-time TTC trigger instruction is selected as the effective trigger signal, and the other trigger channel is immediately locked until the current conflict scene ends. When the pre-trigger position Ai and the real-time collision time TTC condition are activated at the same time, the TTC trigger instruction is selected and the other channel is locked, the real-time collision risk (TTC≤T) is prioritized, and the repeated triggering or instruction conflict caused by double-channel competition is avoided; the locking mechanism ensures the uniqueness of the trigger signal of the current conflict scene, guarantees the target vehicle VTi to intervene accurately according to the preset space-time condition, controls the scene construction error within ±5%, and significantly improves the trigger reliability of the continuous conflict sequence.

[0114] As Figure 2In one of the technical solutions, a first control system is installed on the vehicle under test VUT, the first control system comprising an inertial navigation positioning module, a calculation module and a first communication module arranged on the VUT, the inertial navigation positioning module comprising at least one vehicle-mounted inertial measurement unit and a wheel speed sensor, the calculation module being in communication connection with the inertial navigation positioning module and the first communication module, and a second control system being installed on any target vehicle, the second control system comprising a throttle speed execution mechanism, a VT control module and a second communication module, the VT control module being in communication connection with the throttle speed execution mechanism and the second communication module, and the first communication module being in communication connection with the second communication module.

[0115] The above technical solution realizes the dynamic triggering logic through a split hardware architecture, and separates the control part into a first control system arranged on the vehicle under test VUT and a second control system installed on the target vehicle, the first control system and the second control system working cooperatively through wireless communication, wherein the first control system on the VUT side comprises an inertial navigation positioning module, a calculation module and a first communication module, the inertial navigation positioning module comprising at least a vehicle-mounted IMU and a wheel speed sensor, and collecting VUT position, speed and deceleration data in real time, and being arranged at the center of the VUT chassis and the wheel hubs. The calculation module can be selected from an NVIDIA Jetson AGX Xavier industrial computer, etc., and is arranged in the VUT trunk, and performs basic curve generation, pre-trigger position (Ai) calculation, real-time TTC monitoring and triggering decision. The first communication module is arranged in the VUT roof antenna box, and is responsible for sending triggering instructions to the target vehicle. The second control system on the VTi side comprises a VT control module, a throttle speed execution mechanism and a second communication module, wherein the VT control module is used for analyzing the triggering instructions and generating motor control signals, the throttle speed execution mechanism is mechanically connected to the throttle pedal through a push rod, and the second communication module is used for receiving the instructions of the first control system.

[0116] The above technical solution not only guarantees the reliability of the triggering signal, but also isolates the VUT and the target vehicle control risk through the split architecture, focuses on perception decision (calculation module) on the VUT side, and focuses on execution (motor mechanism) on the target vehicle side, so as to avoid system crash caused by single point failure; and reduces the end-to-end delay, compresses the whole link delay of instruction transmission-analysis-execution to within 200 ms, and meets the response time requirement of 50-200 ms. Moreover, the target vehicle only needs to be equipped with a general execution mechanism, supports rapid deployment of multiple target vehicle sequences, and reduces the hardware cost.

[0117] As shown in Figure 3 In one of the technical solutions, the throttle speed execution mechanism comprises:

[0118] An electric travel push rod 2, one end of which is rotatably connected with the vehicle body 1 of the target vehicle;

[0119] A throttle pedal 3, one end of which is rotatably connected to the other end of the electric stroke push rod, and the other end is rotatably connected to the target vehicle body 1;

[0120] The VT control module controls the stroke slope of the electric stroke push rod 2 according to the trigger signal.

[0121] The above technical solution discloses a preferred mechanical structure of the throttle speed execution structure, which realizes precise control of acceleration through a double-degree-of-freedom rotating structure, including an electric stroke push rod 2 connected to the target vehicle body 1 through a first rotating body 4 and a throttle pedal 3 connected to the electric stroke push rod 2 through a second rotating body 5. The first rotating body 4 and the second rotating body 5 are both preferably pin hole bodies. The assembly position of the electric stroke push rod 2 is the front firewall of the engine compartment, and the assembly position of the throttle pedal 3 is the bottom plate of the cockpit. The VT control module maps the target acceleration required by the trigger signal to the stroke slope of the electric stroke push rod 2, and adjusts the motor speed of the electric stroke push rod through a PWM signal.

[0122] The above technical solution realizes linear control of acceleration, and the double rotating structure eliminates the lateral load of the push rod, so that the thrust is completely converted into pedal torque, which, together with the precise control of the slope, ensures that the target vehicle moves according to the preset acceleration curve. The use of a standardized electric stroke push rod instead of a servo hydraulic system greatly reduces the hardware cost.

[0123] In one of the technical solutions, the load rate of the calculation module of the first control system is monitored in real time. When the load rate is greater than 90% for 3-5 seconds, the real-time TTC trigger channel is automatically closed, a pure pre-trigger position trigger mode is adopted, and a hardware expansion warning is generated and a degradation event is recorded. By automatically closing the real-time time-to-collision (TTC) trigger channel and degrading to a pure pre-trigger position mode by monitoring the load rate of the calculation module in real time, system overload collapse is prevented; a hardware expansion warning is generated and a degradation event is recorded, which ensures the continuity of the test while providing operation and maintenance decision basis, and improves the system availability.

[0124] In one of the technical solutions, the low-cost dynamic trigger test method for automatic driving multi-target vehicle collision sequence in a closed field further comprises establishing a multi-target vehicle trigger collision arbitration mechanism, which specifically performs the following steps:

[0125] The pre-trigger position influence area is a circular area with a pre-trigger position Ai as the center and a radius of 4-6 m.

[0126] According to the current driving direction of the VUT, the target vehicles are given priority according to the following rules:

[0127] First level: target vehicles corresponding to longitudinal conflict points coinciding with the trajectory extension line of the VUT;

[0128] Second level: target vehicle corresponding to the lateral conflict point orthogonal to the VUT trajectory;

[0129] Third level: target vehicle corresponding to the oblique conflict point with an acute angle to the VUT trajectory;

[0130] If multiple target vehicle pre-trigger position influence areas are detected to overlap, the low-priority target vehicle trigger sequence is forced to be delayed until the high-priority target vehicle completes the conflict;

[0131] For the target vehicle triggered with delay, the calibrated motion time ti is reduced by 10% for each priority delay, and the compressed calibrated motion time shall not be lower than 80% of the time required for the target vehicle to accelerate from rest to the corresponding conflict point with maximum acceleration, otherwise the trigger of the target vehicle is cancelled.

[0132] The above technical solution solves the conflict problem caused by the overlap of multiple target vehicle trigger areas through a spatial priority arbitration mechanism. By partitioning the electronic fence, a circular influence area with a radius of 4-6 m is constructed with each target vehicle pre-trigger position (Ai) as the center, forming a pre-trigger position influence area. At the same time, three levels of dynamic priority are set:

[0133] First level: longitudinal conflict target vehicle (coinciding with the VUT trajectory extension line, such as rear-end scenario), highest priority;

[0134] Second level: lateral conflict target vehicle (orthogonal to the VUT trajectory, such as intersection cut-in), medium priority;

[0135] Third level: oblique conflict target vehicle (with an acute angle to the trajectory, such as confluence area conflict), lowest priority.

[0136] When multiple influence areas are detected to overlap, the trigger of the low-priority target vehicle is forced to be delayed until the high-priority target vehicle completes the conflict; when triggered with delay, the calibrated motion time ti is reduced by 10% for each priority delay, and the compressed ti is greater than or equal to 80% of the time required for maximum acceleration (lower limit protection).

[0137] The above technical solution guarantees the timing authenticity of high-value scenarios, accurately quantifies the influence area overlap through the electronic fence (radius 4-6 m), and combines the three-level priority forced scheduling to avoid the risk of physical collision caused by multiple target vehicles triggered simultaneously, thereby reducing the accident rate; the ti compression mechanism of the target vehicle triggered with delay compensates for the time delay, and the lower limit protection strictly controls the time and space error in the delay scenario, ensuring the effectiveness of the test. The priority focuses on longitudinal / lateral core scenarios (accounting for 80% of real accidents), avoiding oblique conflict interference; a single test can handle multiple overlapping target vehicles, further improving the utilization rate of the site.

[0138] The application further claims the protection of a test system for the low-cost dynamic trigger test method of the automatic multi-target vehicle collision sequence in the closed field, comprising:

[0139] A first control system installed on the vehicle under test VUT, comprising:

[0140] An inertial navigation positioning module comprising at least an on-board inertial measurement unit and a wheel speed sensor, for collecting VUT position, speed and deceleration data in real time;

[0141] A calculation module for performing:

[0142] Generating a basic speed-time curve based on the predefined VUT driving route and the collision point sequence when there is no target vehicle for the first time;

[0143] When i=1, calculating the pre-trigger position of the first collision point according to the basic speed-time curve and the nominal motion time ti of VTi reaching the corresponding collision point, and when i≥2, dynamically calculating the pre-trigger position Ai of each target vehicle according to the actual speed-time curve after the previous collision avoidance and the nominal motion time ti of VTi reaching the corresponding collision point;

[0144] Real-time calculating the real-time distance L and time margin TTC of VUT reaching the ith collision point;

[0145] A first communication module for sending trigger instructions;

[0146] A second control system installed on each target vehicle, comprising:

[0147] A VT control module for receiving trigger instructions and generating control signals;

[0148] A throttle speed execution mechanism for driving the target vehicle to move according to the preset trajectory according to the control signals;

[0149] A second communication module for establishing a communication link with the first communication module;

[0150] A dynamic trigger module integrated in the calculation module for:

[0151] When VUT reaches the pre-trigger position Ai or TTC≤the preset threshold Ti of the corresponding collision point, sending trigger instructions to the corresponding target vehicle VTi; if the pre-trigger position and TTC conditions are activated at the same time, the TTC trigger is forced to be selected and the other channel is locked;

[0152] Wherein, the nominal motion time ti is obtained by empty-load dynamics calibration, and the calibration method comprises: measuring the average acceleration of the target vehicle VTi from static acceleration to the collision point, and calculating the nominal motion time of VTi reaching the corresponding collision point based on the acceleration and the measured distance between VTi and the corresponding collision point.

[0153] The application further claims to protect a test system based on a low-cost trigger test mode of an automatic driving multi-target conflict sequence in a closed field, which realizes full-process closed-loop control through an integrated system architecture - the inertial navigation positioning module and the calculation module of the first control system work together to accurately execute the generation of basic curves, dynamic calculation of pre-trigger positions, and real-time time-to-collision (TTC) monitoring; the throttle speed execution mechanism and the VT control module of the second control system ensure that the target vehicle accurately responds to the instructions, the double systems interact in real time through a communication link, the end-to-end delay is compressed to within 200 ms, and the synchronization of the multi-target vehicle trigger timing is ensured. And the dynamic trigger module adopts a double-condition judgment mechanism (pre-trigger position Ai or TTC≤T) and forcibly executes TTC trigger, combined with the channel locking function, a continuous conflict sequence can be reliably constructed in a single test, the system false trigger rate is reduced to below 1%, and the test efficiency is significantly improved.

[0154] In one of the technical solutions, the calculation module further executes:

[0155] Performing risk avoidance intensity monitoring: when the measured deceleration of the current sequence conflict point exceeds the limit, a margin compensation value is generated and added to the preset threshold Ti of all subsequent conflict points;

[0156] Load monitoring: when the load rate of the calculation module is continuously greater than 90% for 3-5s, the TTC trigger channel is closed, a hardware expansion warning is generated, and a degradation event is recorded.

[0157] In the above technical solution, through the risk avoidance intensity monitoring mechanism of the calculation module, the measured deceleration exceeding the limit of the previous conflict point can be detected in real time, and a margin compensation value is automatically generated and added to the preset threshold Ti of all subsequent conflict points, so as to dynamically compensate the speed deviation caused by the risk avoidance behavior of the VUT, significantly improve the accuracy of the continuous conflict sequence trigger and the authenticity of the test scene, and avoid the problem of false trigger or missed trigger of subsequent conflict points caused by excessive risk avoidance intensity. In addition, through the load monitoring function, when the load rate of the calculation module continuously exceeds the limit, the real-time TTC trigger channel is automatically closed and degraded to a pure pre-trigger position mode, a hardware expansion warning is generated, and a degradation event is recorded, effectively preventing the calculation delay or trigger failure caused by system overload, ensuring the stability and reliability of the test process, providing data support for hardware optimization, and enhancing the overall robustness.

[0158] Although the embodiments of the present application have been disclosed as above, they are not limited to the application and implementation listed in the specification and embodiments, and can be fully applied to various fields suitable for the present application, and additional modifications can be easily realized by those skilled in the art, therefore the present application is not limited to specific details and the figures shown and described herein without departing from the general concept defined by the claims and the equivalent scope.

Claims

1. A low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving, characterized in that, include: S1. In a closed test area, predefine the driving route of the vehicle under test (VUT) and multiple conflict points. Each conflict point corresponds to the intervention position of a target vehicle (VTi), i=1,2,3..., and execute the following data acquisition loop: Initial trial run: Control the VUT to drive the entire process in a scenario without a target vehicle, and generate basic speed-time curves for each location point; S2. For the i-th VTi, when i=1, based on the basic velocity-time curve and the calibration motion time ti of VTi reaching the corresponding conflict point, the pre-trigger position Ai at the moment ti before VUT reaches the conflict point is calculated in reverse; when i≥2, after VUT has passed the (i-1)-th conflict point avoidance behavior, the actual velocity-time curve is generated based on the actual velocity and time data of its velocity stability segment, and the pre-trigger position Ai is recalculated. Simultaneously, the position and velocity of the VUT are acquired in real time, and the real-time distance L and time margin TTC of the VUT to the conflict point are calculated in real time, where TTC = L / VUT real-time velocity; S3. When VUT reaches the pre-trigger position Ai or real-time TTC ≤ the preset threshold Ti of the conflict point, a trigger command is sent to the corresponding target vehicle VTi. VTi starts from a standstill and drives to the conflict point according to the preset trajectory, forming a preset conflict scenario with VUT. S4. According to the VUT driving sequence, each VTi is dynamically triggered in turn to form a continuous conflict sequence, and multiple scenarios are verified in a single test. The method for obtaining the calibration motion time ti of VTi reaching the corresponding conflict point is as follows: Before the test, the no-load dynamics calibration of any target vehicle VTi was performed. The average acceleration of VTi from a standstill to the collision point was measured. Based on the acceleration and the measured distance between VTi and the corresponding collision point, the calibration motion time of VTi to the corresponding collision point was calculated. The preset threshold Ti = VTi response time + ti; the VTi response time is 50~200ms.

2. The low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving as described in claim 1, characterized in that, In step S2, during the recalculation of the pre-trigger position Ai, a risk avoidance behavior intensity monitoring and compensation mechanism is implemented. Specifically, this includes: real-time acquisition of the braking deceleration value of the VUT at the preceding conflict point through the fusion of the onboard inertial measurement unit and wheel speed sensors; when the measured deceleration value at any preceding conflict point exceeds the limit of 3.5 m / s², the risk avoidance behavior intensity monitoring and compensation mechanism is implemented. 2 When this happens, the compensation mechanism is activated: Calculate the arithmetic mean of all over-limit decelerations at the preceding conflict points, and generate the margin compensation value according to the following rules: 3.5m / s 2 ≤The arithmetic mean of all out-of-limit decelerations<4.5m / s² 2 At that time, the margin compensation value is 100ms; 4.5m / s 2 ≤The arithmetic mean of all out-of-limit decelerations <5.5m / s² 2 At that time, the margin compensation value is 150ms; The arithmetic mean of all out-of-limit decelerations is >5.5 m / s². 2 At that time, the margin compensation value is 300ms; The generated margin compensation value is added to the preset threshold Ti of all subsequent conflict points, and the compensation effect continues until the end of the current test cycle.

3. The low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving as described in claim 2, characterized in that, When step S3 is executed, if the pre-trigger position condition and the real-time TTC condition are activated at the same time, the real-time TTC trigger command is forcibly selected as the valid trigger signal, and the other trigger channel is immediately locked until the current conflict scenario ends.

4. The low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving as described in claim 3, characterized in that, A first control system is installed on the vehicle under test (VUT). The first control system includes an inertial navigation positioning module, a calculation module, and a first communication module installed on the VUT. The inertial navigation positioning module includes at least one on-board inertial measurement unit and one wheel speed sensor. The calculation module is communicatively connected to the inertial navigation positioning module and the first communication module. A second control system is installed on any target vehicle. The second control system includes a throttle speed actuator, a VT control module, and a second communication module. The VT control module is communicatively connected to the throttle speed actuator and the second communication module. The first communication module is communicatively connected to the second communication module.

5. The low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving as described in claim 4, characterized in that, The throttle speed actuator includes: An electric stroke actuator, one end of which is rotatably connected to the target vehicle body; An accelerator pedal, one end of which is rotatably connected to the other end of the electric travel push rod, and the other end of which is rotatably connected to the target vehicle body; The VT control module controls the stroke slope of the electric stroke push rod according to the trigger signal.

6. The low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving as described in claim 5, characterized in that, The load rate of the computing module of the first control system is monitored in real time. When the load rate is >90% for 3~5 seconds, the real-time TTC trigger channel is automatically shut down, a pure pre-trigger position trigger mode is adopted, and a hardware expansion alarm is generated and a degradation event is recorded.

7. The low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving as described in claim 6, characterized in that, This also includes establishing a multi-target vehicle-triggered conflict arbitration mechanism, specifically implementing the following steps: The electronic fence technology is used to divide the pre-triggering position influence zone associated with each conflict point. The pre-triggering position influence zone is a circular area with a radius of 4 to 6 m centered on the pre-triggering position Ai. Based on the current direction of travel of the VUT, the target vehicle is assigned priority according to the following rules: Level 1: Target vehicles corresponding to longitudinal conflict points that coincide with the VUT trajectory extension line; Level 2: Target vehicles corresponding to lateral collision points orthogonal to the VUT trajectory; Level 3: Target vehicles corresponding to oblique collision points that form an acute angle with the VUT trajectory; If the influence areas of multiple target vehicles' pre-trigger positions are detected to overlap, the triggering order of low-priority target vehicles will be forcibly postponed until the high-priority target vehicles have completed their conflict. For target vehicles that are delayed in triggering, their calibrated motion time ti is reduced by 10% for each priority level of delay. The compressed calibrated motion time must not be less than 80% of the time required for the target vehicle to accelerate from a standstill to the corresponding conflict point at maximum acceleration. Otherwise, the triggering of the target vehicle will be cancelled.

8. A test system for a low-cost dynamic triggering test method for multi-target vehicle conflict sequences in closed-field autonomous driving according to any one of claims 1 to 7, characterized in that, include: The first control system, installed on the vehicle under test (VUT), includes: The inertial navigation positioning module includes at least an onboard inertial measurement unit and wheel speed sensors, used to collect VUT position, velocity and deceleration data in real time; The calculation module is used to perform: Based on the predefined VUT driving route and conflict point sequence, a basic speed-time curve is generated during the first run without a target vehicle. When i=1, the pre-trigger position of the first conflict point is calculated based on the basic speed-time curve and the calibration motion time ti of VTi to reach the corresponding conflict point. When i≥2, the pre-trigger position Ai of each target vehicle is dynamically calculated based on the actual speed-time curve after the previous conflict avoidance and the calibration motion time ti of VTi to reach the corresponding conflict point. Calculate the real-time distance L and time margin TTC from the VUT to the i-th conflict point; The first communication module is used to send trigger commands; The second control system, installed on each target vehicle, includes: The VT control module is used to receive trigger commands and generate control signals; The throttle speed actuator drives the target vehicle to move along a preset trajectory according to the control signal. The second communication module establishes a communication link with the first communication module; The dynamic triggering module, integrated into the calculation module, is used for: When VUT reaches the pre-trigger position Ai or TTC ≤ the preset threshold Ti of the corresponding conflict point, a trigger command is sent to the corresponding target vehicle VTi; if the pre-trigger position and TTC conditions are activated at the same time, TTC trigger is forcibly selected and another channel is locked. The calibration motion time ti is obtained through no-load dynamic calibration. The calibration method includes: measuring the average acceleration of the target vehicle VTi from a standstill to the conflict point, and calculating the calibration motion time of VTi to the corresponding conflict point based on the acceleration and the measured distance between VTi and the corresponding conflict point.

9. The testing system as described in claim 8, characterized in that, The calculation module also performs: Perform risk avoidance intensity monitoring: When the measured deceleration of the preceding conflict point exceeds the limit, generate a margin compensation value and add it to the preset threshold Ti of all subsequent conflict points; Load monitoring: However, if the load rate of the computing module remains above 90% for 3-5 seconds, the TTC trigger channel will be closed and a hardware expansion alarm will be generated, and the degradation event will be recorded.

Images