Method for reducing loss of collision test equipment of automatic driving vehicle

Abstract

The invention discloses a method for reducing the loss of an automatic driving vehicle collision test device, and the method comprises the following steps: obtaining preset test vehicle parameters which comprise the maximum deceleration amax of an automatic emergency braking system, the test vehicle speed vtest and the communication system delay time t delay; when an emergency braking system of the test vehicle is triggered, emergency avoidance triggering time t triggering is calculated through a formula t triggering = v test / amax + t delay; after position data of the test vehicle and the background target object are collected in real time, the real-time relative distance D is calculated, and the collision time t collision is calculated through a formula t collision = D / v test; and when t collision is less than or equal to t trigger, driving the background target object to move away from the collision direction through the mechanical execution mechanism. Before collision, the target object is actively driven to be away from the collision direction, and structural damage caused by direct collision can be avoided.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving. More specifically, the present invention relates to a method for reducing the loss of collision test equipment for autonomous vehicles. Background Art

[0002] With the rapid development of autonomous driving technology, the frequency and complexity of collision tests for autonomous vehicles have increased significantly. In traditional collision tests, the collision between the test vehicle and background targets (such as static or dynamic targets simulating pedestrians and other vehicles) usually relies on the structural design of the targets themselves (such as flying out or scattering after collision) to consume the collision kinetic energy to protect the test vehicle. However, this passive protection method has significant defects: Severe equipment loss: The background targets are prone to structural damage or even scrapping after multiple collisions, and high-precision test targets (such as dynamic targets that can simulate complex movements) are costly, and frequent replacement leads to a significant increase in test costs; Limited test efficiency: The repair or replacement of the targets requires interrupting the test, affecting the continuity of the test process; Insufficient safety: Passive protection cannot actively avoid collisions, which may cause the fragments of the targets to fly or get out of control, posing potential risks to the personnel and equipment on the test site.

[0003] The prior art lacks an active protection mechanism for background targets and is difficult to reduce losses by dynamically adjusting the position of the targets before a collision occurs. Therefore, there is an urgent need for a method that can accurately predict the collision risk and trigger the avoidance of the targets to improve the test safety and economy. Summary of the Invention

[0004] The object of the present invention is to provide a method for reducing the loss of collision test equipment for autonomous vehicles. By real-time monitoring of the relative distance, speed, and deceleration of the test vehicle and the background targets, accurately judging the collision risk, and actively driving the targets away from the collision direction before a collision occurs, it is possible to avoid or reduce the structural damage caused by direct collisions and extend the service life of the targets; To achieve these and other advantages of the present invention, a method for reducing the loss of collision test equipment for autonomous vehicles is provided, including the following steps: Obtain the pre-set test vehicle parameters, where the test vehicle parameters include the maximum deceleration of the automatic emergency braking system a max , the test vehicle speed v 试验 and the communication system delay time t 延迟 ; When the emergency braking system of the test vehicle is triggered, through the formula t 触发 = v试验 / a max + t 延迟 Calculate the emergency avoidance trigger time t 触发 ; After collecting the position data of the test vehicle and the background target in real time, calculate the real-time relative distance D , and through the formula t 碰撞 = D / v 试验 Calculate the collision time t 碰撞 ; When t 碰撞 ≤ t 触发 Drive the stationary background target to move away from the collision direction through the mechanical actuator

[0005] Preferably, in the method for reducing the loss of the collision test equipment for autonomous vehicles, the following steps are further included: Use the GPS positioning device to obtain the GPS information and acceleration data of the test vehicle and the background target; respectively set the communication system delay monitoring device on the platforms of the test vehicle and the background target t 延迟 is the sum of the delay time measured by the communication system delay monitoring device at the vehicle end and the delay time measured by the communication system delay monitoring device at the target end When obtaining the pre-set test vehicle parameters, obtain the maximum deceleration of the test vehicle a 测试 ; At consecutive time points t 1 and t 2 record the GPS coordinates of the test vehicle and the background target, and calculate the relative displacement △ t 1, t 2 at the moment of the coordinate calculation relative displacement △ s 定位 ; Through the formula v 相对 = △ s 定位 / ( t 1- t 2 ) calculate the relative speed based on GPS information v 相对 ; According to the real-time relative distance D and v相对 , combined with a 测试 , through the displacement formula of uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 测试 t 2 , recalculate the collision time to obtain a new collision time t 新 .

[0006] Preferably, in the method for reducing the loss of the collision test equipment of the autonomous driving vehicle, the actual deceleration is obtained through the acceleration sensor built in the GPS positioning device a 实际 , and compare it with a 测试 : If a 实际 = a 测试 , the collision time is t 新 ; If a 实际 < a 测试 , combined with the real-time relative distance D and v 相对 , substitute a 实际 back into the displacement formula of uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 实际 t 2 , recalculate the collision time to obtain an adjusted collision time t 调整 ; If a 实际 > a 测试 , perform a calibration check on the sensor data. If after calibration a 实际 > a 测试 , analyze whether there is a system anomaly. If there is an anomaly, issue an alarm and stop the test. If there is no anomaly, combined with the real-time relative distance D and relative speed v 相对 , substitutea 实际 Substitute back into the displacement formula for uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 实际 t 2 , recalculate the collision time to obtain the adjusted collision time t 调整 .

[0007] Preferably, in the method for reducing the loss of the collision test equipment for autonomous vehicles, the following steps are further included: Install a lidar and multiple vision sensors on the test vehicle, and install a lidar on the background target; set a slope sensor in the test site and use a laser profilometer; the lidars on the test vehicle and the background target respectively measure the distance to each other, and the vision sensors collect images of the surrounding environment; the slope sensor obtains the slope information of the test site in real time, and the laser profilometer measures the ground flatness to obtain high-precision map data; Calculate the relative distance between the test vehicle and the background target using the GPS positioning device information D 定位 ; Process and analyze the lidar measurement data to obtain the relative distance based on the lidar D 雷达 ; Process the images collected by the vision sensors to obtain the relative distance based on the vision sensors D 视觉 ; Match the position information of the test vehicle and the background target with the high-precision map to obtain the relative distance based on the high-precision map D 地图 ; Adopt a data fusion algorithm to fuse D 定位 , D 雷达 , D 视觉 and D 地图 to obtain the fused relative distance D 融合 , according to the slope angle obtained by the slope sensor θ , 0° ≤ θ < 15°, if the vehicle and the target are on the same slope and the slope is a straight line, through the formula D 修正 = D 融合 / cosθ Correct D 融合 to obtain the distance after slope correction D 修正 , if the vehicle and the target are not on the same straight-line slope, skip the correction and directly use D 融合 as D 修正 ; The laser profilometer divides the ground into multiple small areas and measures the undulation height of each small area h i and length l i , i = 1, 2, …, n , i , where represents the number of areas, calculate the actual distance increase of each small area , accumulate the actual distance increases of all small areas to obtain the final and more accurate real-time relative distance D = D 修正 + △ L .

[0008] Preferably, in the method for reducing the loss of the automatic driving vehicle collision test equipment, it further includes: After the emergency braking system of the test vehicle is triggered and reaches the maximum deceleration of the automatic emergency braking system a max , based on the maximum deceleration a max of the automatic emergency braking system and the real-time relative speed v 相对 , calculate the theoretical shortest braking distance through the kinematic formula s 最短 = v 2 相对 / (2 a max ); if the remaining real-time relative distance during the braking phase s 最短 ≤ D ≤ s 最短 , it is determined that the collision cannot be avoided only by braking; When and only when the following conditions are simultaneously satisfied, the calculation module sends a trigger signal to the background target carrying system through the communication link. After receiving the trigger signal, the background target carrying system starts the mechanical actuator to drive the background target away from the collision direction: 1) From the start time of the testt 0 = Current time starting from 0s t Meet t 制动 ≤ t ≤ t 制动 + Collision time, where t 制动 Is the triggering moment of the emergency braking system; 2) The emergency braking system of the test vehicle has been triggered, and the deceleration reaches the maximum deceleration of the automatic emergency braking system a max ; 3) The collision time is less than or equal to the triggering time; If a 实际 = a 测试 The collision time is t 新 ; If a 实际 ≠ a 测试 The collision time is t 调整 .

[0009] Preferably, in the method for reducing the loss of the collision test equipment of the autonomous driving vehicle, after the background target carrying system receives the trigger signal, by calculating the collision risk coefficient R = D / D 参考 × v 相对 Determine the moving method of the background target, where R The dimension of is m / s, D 参考 Is a reference distance preset according to the performance of the test vehicle and the background target; The method for determining the moving method of the background target according to the collision risk coefficient is: If R < R 低 The mechanical actuator drives the background target with 30% of its maximum power, and the moving distance L 1= k 1×( D-D 安全 ), t 移动 < t 碰撞 , t 移动 = L 1 / v 驱动 ; Where kThe value range of 1 is 1.1 - 1.3, t 移动 is the moving time, v 驱动 is the driving speed of the actuator, D 安全 is the safety distance threshold; and k 1 is dynamically selected according to the relative speed v 相对 and the relative distance D : When v 相对 ≤ 5m / s, and D > 10m, k 1 = 1.1; When v 相对 ≤ 5m / s, and 5m < D ≤ 10m, k 1 = 1.2; When v 相对 ≤ 5m / s, and D ≤ 5m, k 1 = 1.3; If R 低 ≤ R < R 高 , the mechanical actuator drives the background target with 60% of its maximum power, and the moving distance L 2 = k 2 × ( D - D 安全 ), t 移动 < t 碰撞 , t 移动 = L 2 / v 驱动 ; where k the value range of 2 is 1.3 - 1.5; and k 2 is dynamically selected according to the relative speed v 相对 and the relative distance D : When 5m / s < v 相对 ≤ 15m / s, and D > 8m, k 2 = 1.3; When 5m / s < v 相对 ≤ 15m / s, and 5m <D When it is ≤ 8m, k 2 = 1.4; When 5m / s < v 相对 ≤ 15m / s, and D ≤ 5m, k 2 = 1.5; If the collision risk coefficient R ≥ R 高 , regardless of v 相对 whether it is greater than 15m / s or not, the mechanical actuator immediately drives the background target with maximum power and moves it to the nearest safe area along the shortest straight path. The moving distance L 3 = L 初始 + L 缓冲 , t 移动 < t 碰撞 , t 移动 = L 3 / v 驱动 , v 驱动 is the driving speed of the actuator; where L 初始 is the straight-line distance from the current position to the center of the safe area, L 缓冲 is the buffer distance of 5 - 10m.

[0010] Preferably, in the method for reducing the loss of the collision test equipment of the autonomous driving vehicle, R 低 The value range of is 1.2 - 2.5; R 高 The value range of is 2.5 - 4.0; D 参考 = k × s 最短 ; k, D 参考 Dynamically match according to the following rules according to the test scenario: Urban road test: k = 1.5 - 2.0, D 参考 = 15 - 25m; Highway test: k = 2.0 - 3.0, D 参考 ≥ 100m; Extreme condition test: k = 2.5 - 3.5, D 参考 = 5 - 10m; D 安全 The value range is 5.56 - 50m, and it satisfies D ≤ D 安全 ≤ D 参考 , and D 安全 = max(2 × L 车辆 , 1.2 × s 最短 , v 相对 × t 响应 ), L 车辆 is the total length of the test vehicle, t 响应 is the total time required for the mechanical actuator to start driving the background object to move from the system detecting a collision risk, specifically including: sensor data processing time t 数据处理 、computing module decision-making time t 计算 、 t 延迟 、mechanical actuator startup time t 执行 ; Urban road test: t 响应 = 0.3 - 0.5s; Highway test: t 响应 = 0.2 - 0.3s; Extreme condition test: t 响应 = 0.1 - 0.2s; v 相对 in m / s; t 响应 in s; L 车辆 、 s 最短 、 D 安全 are all in m.

[0011] Preferably, in the method for reducing the loss of the automatic driving vehicle collision test equipment, the data fusion algorithm adopts a dynamic weight adjustment mechanism, which specifically includes the following steps: Real-time evaluation of the confidence metrics of each sensor's data, including: Calculating the confidence based on the measurement noise level of the lidar C 雷达 , where C 雷达 = 1 / ( σ 雷达 2 + ε ), σ 雷达 is the standard deviation of the current frame data of the lidar, ε is a very small constant to prevent the denominator from being zero; Calculating the confidence based on the image sharpness and feature matching degree of the vision sensor C 视觉 , where C 视觉 = S 视觉 × M 匹配 , S 视觉 is the image sharpness score, M 匹配 is the matching accuracy of the target feature points; Calculating the confidence based on the number of satellites and signal strength of the GPS positioning device C 定位 , where C 定位 = N 卫星 × I 信号 , N 卫星 is the number of visible satellites, I 信号 is the signal strength normalization value; Calculating the confidence based on the update timeliness and matching error of the high-precision map C 地图 , where C 地图 = 1 / (△ t 更新 + E 匹配 ), △ t 更新 is the time difference of map data update, E 匹配 is the position matching error; For D 定位 , D 雷达 , D视觉 and D 地图 are respectively given dynamic weights W 定位 、 W 雷达 、 W 视觉 and W 地图 , and the weight calculation formula is: W j = C j / (Σ C j ) wherein, j ∈ {positioning, radar, vision, map}; C j is the confidence index of each sensor, C j including C 雷达 、 C 视觉 、 C 定位 、 C 地图 ; The relative distance after fusion is calculated through the weighted fusion formula: D 融合 = W 定位 × D 定位 + W 雷达 × D 雷达 + W 视觉 × D 视觉 + W 地图 × D 地图 .

[0012] The present invention has at least the following beneficial effects: By calculating the time to collision t 碰撞 and the emergency avoidance trigger time t 触发 in real time, the background target is driven away from the collision direction before the collision occurs, avoiding structural damage caused by direct collision, and prolonging the service life of high-value targets (such as dynamic simulation dummies, high-precision vehicle models), which is especially suitable for scenarios that require repeated tests.

[0013] The present invention dynamically adjusts the power of the mechanical actuator (30%, 60%, maximum power) according to the collision risk coefficient R to reduce energy consumption in low-risk scenarios and respond quickly in high-risk scenarios, reducing unnecessary losses of the actuator while ensuring safety and balancing the protection effect and equipment life.

[0014] The present invention integrates GPS, lidar, vision sensors, and terrain data (slope, ground flatness), and through data fusion algorithms and formula corrections (such as D 修正 = D 融合 / cosθ ), eliminates the errors of single sensors and the influence of terrain, improves the calculation accuracy of relative distance and collision time, and prevents misjudgment or missed judgment.

[0015] The present invention monitors the actual deceleration in real time a 实际 and compares it with the design value a 测试 to automatically adjust the collision time t 新 or t 调整 , adapts to the actual performance of the braking system, ensures that the trigger time matches the vehicle's dynamic performance, and avoids avoidance failure caused by braking deviation.

[0016] The present invention can prevent the target object from being damaged due to collision and requiring repair or replacement, realizes the continuity of the test process, is especially suitable for high-frequency and long-cycle test scenarios, and significantly improves the test efficiency.

[0017] The present invention dynamically adjusts the reference distance D 参考 , safety distance D 安全 and response time t 响应 according to the test scenario (urban road, highway, extreme working conditions), is compatible with different vehicle models and complex environments (such as slopes, slippery roads), and ensures reliable operation under non-ideal conditions.

[0018] For extreme scenarios with short distances and high speeds (such as pedestrians crossing the road, high-speed rear-end collisions), the present invention completes avoidance in an extremely short time through the shortest path planning and maximum power drive, meeting the safety requirements of high-risk scenarios.

[0019] The complete closed-loop of "real-time data acquisition - collision risk calculation - active control execution" of the present invention realizes the automation from risk identification to action execution through algorithm optimization and hardware cooperation, providing an intelligent protection solution for autonomous driving collision tests.

[0020] The present invention supports integration with vehicle ECUs, high-precision maps, and multiple types of sensors, can be flexibly adapted to different test platforms, and provides a general framework for the sustainable development of future autonomous driving test technologies.

[0021] Through core advantages such as active protection, precise calculation, cost optimization, and scenario adaptation, the present invention effectively solves the pain points of equipment loss in traditional collision tests, and has technological innovation, engineering practicability, and economic rationality, providing an efficient, safe, and sustainable solution for the field of autonomous driving testing.

[0022] Other advantages, objectives, and features of the present invention will be partially reflected by the following description, and will also be understood by those skilled in the art through the research and practice of the present invention. Detailed Description of the Invention

[0023] The following further elaborates on the present invention in conjunction with embodiments, enabling those skilled in the art to implement it with reference to the text of the specification.

[0024] The present invention provides a method for reducing equipment loss in autonomous vehicle collision tests, including the following steps: Obtain preset test vehicle parameters, where the test vehicle parameters include the maximum deceleration of the automatic emergency braking system a max , test vehicle speed v 试验 , and communication system delay time t 延迟 ; When the emergency braking system of the test vehicle is triggered, calculate the emergency avoidance trigger time t 触发 = v 试验 / a max + t 延迟 through the formula t 触发 ; After real-time collecting the position data of the test vehicle and the background target, calculate the real-time relative distance D , and calculate the collision time t 碰撞 = D / v 试验 through the formula t 碰撞 ; When t 碰撞 ≤ t 触发 , drive the stationary background target to move along the track away from the collision direction through a mechanical actuator.

[0025] When obtaining the pre-set test vehicle parameters, the test vehicle speed can be selected as 8.33 - 16.67 m / s, the maximum deceleration of the automatic emergency braking system is 3 - 5 m / s², and the communication system delay time is 0.1 - 0.5 s. These parameters can be determined through the vehicle design manual or historical test data. The communication system delay monitoring device used can be an existing wireless signal delay tester on the market, installed on the carrying platforms of the test vehicle and the background target.

[0026] When the emergency braking system of the test vehicle is triggered, through the calculation module built into the on-vehicle central processing unit, using the formula t 触发 = v 试验 / a max + t 延迟 calculate the emergency avoidance trigger time. Among them, the test vehicle speed v 试验 is obtained in real time from the vehicle dashboard, the maximum deceleration of the automatic emergency braking system a max is the pre-set value when the vehicle leaves the factory, and the communication system delay time t 延迟 is fed back in real time by the real-time monitoring device.

[0027] When collecting the position data of the test vehicle and the background target in real time, a GPS positioning device with centimeter-level accuracy can be used, installed on the top of the vehicle and the center position of the target respectively. Calculate the real-time relative distance D by parsing the GPS coordinates, and use the formula t 碰撞 = D / v 试验 to calculate the collision time. When t 碰撞 ≤ t 触发 , the calculation module sends a trigger signal to the background target to activate its mechanical actuator, driving the target to move away from the collision direction along the preset track.

[0028] The background target is stationary during the test.

[0029] Before the test, perform dynamic parameter adjustment according to relevant parameters such as the maximum deceleration, test vehicle speed, and communication system delay time of the vehicle design.

[0030] Emergency avoidance module trigger time = test vehicle speed / maximum deceleration of the automatic emergency braking system + communication system delay time; In the test collision experiment, when the emergency braking system of the test vehicle is triggered, the calculation module calculates the GPS information of the test vehicle and the background target (calculates the relative distance between the test vehicle and the background target, and calculates in real time through the relative distance t 碰撞 the relative relationship with the triggering time of the emergency avoidance module. When calculating in real time the relative distance t 碰撞 is less than or equal to the triggering time of the emergency avoidance module, a triggering signal is sent). When the emergency braking system of the test vehicle cannot avoid a collision even when it reaches the maximum deceleration, when the preset time for triggering the emergency avoidance module is reached, the calculation module triggers the emergency avoidance module of the background target carrying system, so that the background target moves away from the collision direction, thereby reducing the possible damage to the background target.

[0031] In another solution, in the method for reducing the loss of the automatic driving vehicle collision test equipment, the following steps are included: Use a GPS positioning device with centimeter-level positioning ability and built-in acceleration sensor and gyroscope to obtain the GPS information and acceleration data of the test vehicle and the background target. The GPS positioning device can be installed at the center of gravity positions of the vehicle and the target. Communication system delay monitoring devices are respectively set on the test vehicle and the background target carrying platform. t 延迟 is the sum of the delay time measured by the communication system delay monitoring device at the vehicle end and the delay time measured by the communication system delay monitoring device at the target end. The real-time delay time is usually obtained by the system through a real-time measurement mechanism. For example: 1. Round-trip time measurement: By sending a specific test signal and recording the time difference between the signal sending and receiving, the current delay is directly calculated. 2. System built-in monitoring module: Deploy dedicated monitoring nodes or algorithms in the communication link to collect and feedback the current delay data in real time. 3. Protocol layer feedback: Some communication protocols (such as TCP, Real-Time Transport Protocol RTP) have built-in delay statistics functions, and report the delay information in real time through protocol interaction.

[0032] These real-time measurement results can be directly used for system control (such as adjusting control strategies, compensating for the impact of delays) to ensure that the real-time requirements can still be met when the delay changes; When obtaining the preset test vehicle parameters, the maximum deceleration of the test vehicle is also obtained a 测试 ; a 测试 is the maximum deceleration value that the vehicle can withstand during braking, which is determined by comprehensively considering various factors such as the vehicle body structure, braking system reliability, and tire grip during the vehicle design stage, and can be obtained from the vehicle braking system technical document as a preset parameter in the design stage.

[0033] At consecutive time points t 1 and t 2 ( t both 1 and t 2 are counted from the start time of the test t 0 = 0 s, and the time interval can be set to 0.1 s), record the GPS coordinates of the test vehicle and the background target. Calculate the relative displacement △ t 1,[[]] t 2 at the coordinates at time s 定位 ; Through the formula v 相对 = △ s 定位 / ([[]] t 1 - t 2) calculate the relative speed v 相对 ; The calculation of the relative speed can help judge the relative motion state between the vehicle and the target, providing an important basis for the subsequent trigger time calculation and collision risk assessment.

[0034] According to the real-time relative distance D and v 相对 , combined with a 测试 , through the displacement formula of uniformly variable rectilinear motion D =[[]] v 相对 t - 1 / 2 a 测试 t 2 , recalculate the collision time to obtain the new collision time t 新 .

[0035] In another scheme, in the method for reducing the loss of the collision test equipment of the autonomous driving vehicle, obtain the actual deceleration a 实际 through the acceleration sensor built in the GPS positioning device, and compare it with a 测试 : If a 实际 =[[]] a 测试 , the collision time is t 新 ; If a 实际 < a 测试, combined with the real-time relative distance D and v 相对 , substitute a 实际 back into the displacement formula for uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 实际 t 2 , recalculate the collision time to obtain the adjusted collision time t 调整 ; When the actual deceleration is less than the designed maximum deceleration, it indicates that there is still a certain margin in the braking ability of the vehicle. The collision time can be recalculated according to the actual deceleration to further optimize the avoidance strategy.

[0036] If a 实际 > a 测试 , the system automatically triggers the sensor calibration program to check the sensor data. If after calibration a 实际 > a 测试 , analyze whether there is a system anomaly. If there is an anomaly, issue an alarm and stop the test. If there is no anomaly, combined with the real-time relative distance D and the relative speed v 相对 , substitute a 实际 back into the displacement formula for uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 实际 t 2 , recalculate the collision time to obtain the adjusted collision time t 调整 . When the actual deceleration is greater than the designed maximum deceleration, there may be a sensor failure or system anomaly, and calibration and inspection are required to ensure the safety and reliability of the test.

[0037] In another solution, in the method for reducing the loss of the collision test equipment for autonomous vehicles, the following steps are further included: Install a lidar and multiple vision sensors on the test vehicle, and install a lidar on the background target; set up a slope sensor in the test site and use a laser profilometer; the lidars on the test vehicle and the background target measure the distances to each other respectively, and the vision sensors collect images of the surrounding environment; the slope sensor obtains the slope information of the test site in real time, and the laser profilometer measures the ground flatness to obtain high-precision map data; the lidar can select a commercial model with 16 lines or 32 lines, and is installed on the front bumper of the vehicle and the side of the target. The vision sensor can select an industrial camera with a resolution of 1080p and is installed on both sides of the vehicle windshield. The purpose of the multi-sensor layout is to obtain test-related data from multiple angles and improve the accuracy and reliability of the data. Different types of sensors can provide complementary information. For example, a GPS positioning device can provide the position information of the vehicle and the target, the lidar can accurately measure the distance, the vision sensor can collect environmental images, and the slope sensor and the laser profilometer can obtain the terrain information of the site. These data will be used for subsequent calculations of relative distance and speed and optimization of trigger time.

[0038] Calculate the relative distance between the test vehicle and the background target using the GPS positioning device information D 定位 ; The GPS positioning device can provide the accurate position information of the vehicle and the target. By calculating the position difference between the two, the relative distance can be obtained.

[0039] Process and analyze the lidar measurement data to obtain the relative distance based on the lidar D 雷达 ; The lidar calculates the distance by emitting laser beams and measuring the time of the reflected light, and has the characteristics of high precision.

[0040] Apply computer vision technology to process the images collected by the vision sensors to obtain the relative distance based on the vision sensors D 视觉 ; Computer vision technology can estimate the distance by analyzing the object features and proportional relationships in the images.

[0041] Match the position information of the test vehicle and the background target with the high-precision map to obtain the relative distance based on the high-precision map D 地图 ; The high-precision map can provide detailed geographical information. By matching the positions of the vehicle and the target with the information on the map, a more accurate relative distance can be obtained.

[0042] Adopt data fusion algorithms, such as weighted average or Kalman filter algorithms, to D 定位 、 D 雷达 、D 视觉 and D 地图 fuse them to obtain the fused relative distance D 融合 , data fusion can integrate data from different sensors to improve the accuracy and reliability of relative distance calculation. According to the slope angle obtained by the slope sensor θ , 0° ≤ θ < 15°, if the vehicle and the target are on the same slope and the slope is a straight line, through the formula D 修正 = D 融合 / cosθ correct D 融合 to obtain the distance after slope correction D 修正 , if the vehicle and the target are not on the same straight-line slope, skip the correction and directly use D 融合 as D 修正 ; Considering the influence of the slope can calculate the actual distance between the vehicle and the target more accurately and avoid distance calculation errors caused by the slope.

[0043] The laser profilometer divides the ground into multiple small areas, such as small areas of 1m × 1m, and measures the undulation height h i and length l i , i = 1, 2,..., n , i represents the number of areas, calculates the actual distance increment of each small area , accumulates the actual distance increments of all small areas to obtain the final more accurate real-time relative distance D = D 修正 + △ L .

[0044] Considering the influence of ground flatness can further improve the accuracy of relative distance calculation and ensure the reliability of subsequent calculations.

[0045] In another solution, in the method for reducing the loss of the autonomous driving vehicle collision test equipment, it further includes: With the help of a high-precision clock module inside the equipment that is synchronized with the vehicle electronic control unit (ECU), continuously and real-time obtain the time from the start of the test at a set frequency tThe current time starting from 0 = 0s is compared with the triggering time of the emergency avoidance module. Real-time monitoring of the time can ensure that the emergency avoidance module is triggered at an appropriate time, avoiding collision accidents caused by time errors.

[0046] Using the acceleration sensor built into the GPS positioning device, the change in the vehicle's deceleration during braking is fed back in real time. After the emergency braking system is activated, the vehicle's central processing unit (CPU) tracks the change in deceleration to determine whether it reaches the maximum deceleration of the preset automatic emergency braking system. a max The GPS positioning device performs redundant verification on the acceleration data through the built-in multi-sensor fusion technology, and takes the average value after removing outliers. Accurately monitoring the status of the braking system can ensure that the vehicle brakes at the expected deceleration during braking, improving the accuracy and safety of the test.

[0047] When the emergency braking system of the test vehicle is triggered and reaches the maximum deceleration of the automatic emergency braking system a max After that, based on the maximum deceleration of the automatic emergency braking system a max And the real-time relative speed v 相对 , through the kinematic formula s 最短 = v 2 相对 / (2 a max ) Calculate the theoretical shortest braking distance s 最短 ; s 最短 is the shortest stopping distance that the vehicle can reach when braking with the maximum braking ability (i.e., a max deceleration), which is used to determine whether a collision will still occur even if the vehicle brakes in the best state.

[0048] If the remaining real-time relative distance during the braking phase D ≤ s 最短 , it is determined that braking alone cannot avoid a collision; When and only when the following conditions are simultaneously met, the calculation module sends a trigger signal to the background target carrying system through the communication link. After receiving the trigger signal, the background target carrying system activates the mechanical actuator to drive the background target away from the collision direction: 1) The current time starting from the test start time t 0 = 0s t meets t 制动 ≤t ≤ t 制动 + collision time, where t 制动 is the triggering moment of the emergency braking system; 2) The emergency braking system of the test vehicle has been triggered and the deceleration reaches the maximum deceleration of the automatic emergency braking system a max ; 3) The collision time is less than or equal to the triggering time; If a 实际 = a 测试 , the collision time is t 新 ; If a 实际 ≠ a 测试 , the collision time is t 调整 . Efficient triggering of emergency avoidance can start the avoidance mechanism of the background target in time when a collision is about to occur, reducing the possibility of collision and equipment loss.

[0049] In another solution, in the method for reducing the loss of the collision test equipment of the autonomous vehicle, after receiving the trigger signal, the background target carrying system calculates the collision risk coefficient R = D / D 参考 × v 相对 to determine the moving method of the background target, where R The dimension of is m / s, and the collision risk coefficient R Combines the ratio of the relative distance to the reference distance with the relative speed to comprehensively quantify the avoidance speed that the target needs to reach. The larger its value, the higher the power required to drive the target to move to avoid collision. D 参考 is the reference distance preset according to the performance of the test vehicle and the background target; The method for determining the moving method of the background target according to the collision risk coefficient is: If R < R 低 ( R 低 Determined by statistical analysis of historical collision data), the mechanical actuator drives the background target with 30% of its maximum power. In low-risk situations, using less power to drive the background target to move can reduce energy consumption and equipment wear. The moving distance L 1= k 1× ([[]] D - D安全 ), t 移动 < t 碰撞 , t 移动 = L 1 / v 驱动 ; where k The value range of 1 is 1.1 - 1.3, t 移动 is the moving time, v 驱动 is the driving speed of the actuator, D 安全 is the safety distance threshold, D ≤ D 安全 ; and k 1 is dynamically selected according to the relative speed v 相对 and the relative distance D : When v 相对 ≤ 5m / s, and D > 10m, k 1 = 1.1; When v 相对 ≤ 5m / s, and 5m < D ≤ 10m, k 1 = 1.2; When v 相对 ≤ 5m / s, and D ≤ 5m, k 1 = 1.3; If R 低 ≤ R < R 高 ( R 高 determined by statistical analysis of historical collision data), the mechanical actuator drives the background object with 60% of its maximum power. In medium-risk situations, appropriately increasing the power and the moving distance can improve the avoidance effect. The moving distance L 2 = k 2 × ([[]] D - D 安全 ), t 移动 < t 碰撞 , t 移动 = L 2 / v驱动 ; where k the value range of 2 is 1.3 - 1.5; and k 2 is dynamically selected according to the relative speed v 相对 and the relative distance D : When 5m / s < v 相对 ≤ 15m / s, and D > 8m, k 2 = 1.3; When 5m / s < v 相对 ≤ 15m / s, and 5m < D ≤ 8m, k 2 = 1.4; When 5m / s < v 相对 ≤ 15m / s, and D ≤ 5m, k 2 = 1.5; If the collision risk coefficient R ≥ R 高 , regardless of v 相对 whether it is greater than 15m / s or not, the mechanical actuator immediately drives the background target with the maximum power and moves it to the nearest safe area along the shortest straight-line path. The safe area is pre-marked in the test site, for example, a circular area with a radius of 5m is delimited by a reflective road sign. The moving distance L 3 = L 初始 + L 缓冲 , t 移动 < t 碰撞 , t 移动 = L 3 / v 驱动 , v 驱动 is the driving speed of the actuator; where, L 初始 is the straight-line distance from the current position to the center of the safe area, L 缓冲 is a buffer distance of 5 - 10m. In high-risk situations, moving the background target with the maximum power and the shortest path can minimize the possibility of collision.

[0050] In another solution, in the method for reducing the loss of the collision test equipment of the autonomous driving vehicle,R 低 The value range is 1.2 - 2.5; R 高 The value range is 2.5 - 4.0; D 参考 = k × s 最短 ; k, D 参考 Dynamically match according to the test scenario according to the following rules: Urban road test: k = 1.5 - 2.0, D 参考 = 15 - 25m; Highway test: k = 2.0 - 3.0, D 参考 ≥ 100m; Extreme working condition test: k = 2.5 - 3.5, D 参考 = 5 - 10m D 安全 The value range is 5.56 - 50m, and it satisfies D ≤ D 安全 ≤ D 参考 , and D 安全 = max(2 × L 车辆 , 1.2 × s 最短 , v 相对 × t 响应 ). L 车辆 is the total length of the test vehicle, t 响应 is the total time required for the mechanical actuator to start driving the background object to move from the time when the system detects a collision risk, specifically including: sensor data processing time t 数据处理 , calculation module decision time t 计算 , t 延迟 , mechanical actuator start time t 执行 ; Urban road test: t 响应 = 0.3 - 0.5s; Highway Test: t 响应 = 0.2 - 0.3 s; Extreme Condition Test: t 响应 = 0.1 - 0.2 s.

[0051] v 相对 in m / s; t 响应 in s; L 车辆 , s 最短 , D 安全 All in m. The speed unit of this application is m / s, the time unit is s, and the length unit is m.

[0052] Extreme conditions usually refer to test scenarios with high dynamics, short distances, and high collision risks, such as a target object suddenly cutting into the lane (e.g., a pedestrian crossing, a vehicle suddenly braking), a test vehicle approaching a stationary target object at high speed (e.g., a breakdown vehicle on the highway), and emergency avoidance in complex terrains (e.g., slopes, slippery roads).

[0053] Core Requirement: Collision prediction and avoidance actions need to be completed in an extremely short time; otherwise, the probability of avoidance failure will increase sharply.

[0054] Response Time t 响应 refers to the total time from when the system detects a collision risk to when the mechanical actuator starts to drive the background target object to move, specifically including the following stages: 1. Sensor Data Processing Time t 数据处理 : The time for GPS, lidar, vision sensors, etc. to collect data and perform fusion (typical value: 50 - 100 ms).

[0055] 2. Computational Module Decision Time t 计算 : The time to calculate the collision time t 碰撞 , risk coefficient R (typical value: 20 - 50 ms).

[0056] 3. Communication System Delay Time t 延迟 : The time from when the computational module sends a trigger signal to when the background target object carrier system receives the signal.

[0057] 4. Mechanical Actuator Startup Time t执行 : The time from when the actuator receives a signal to when it starts to act (typical value: 100 - 200 ms).

[0058] D 安全 = max(2 × L 车辆 , 1.2 × s 最短 , v 相对 × t 响应 ), taking the maximum value among the three results.

[0059] Urban road test: Typical v 相对 ≤ 5 m / s, t 响应 ≤ 0.5 s, and the safety distance is mainly determined by 2 × L 车辆 or 1.2 × s 最短 .

[0060] Highway test: High v 相对 (such as 30 m / s) and short t 响应 (such as 0.2 s), v 相对 × t 响应 = 6 m, which may be less than 2 × L 车辆 .

[0061] Extreme working condition test If v 相对 = 40 m / s, t 响应 = 0.1 s, then v 相对 × t 响应 = 4 m, and the static parameters are still dominant.

[0062] In another solution, in the method for reducing the loss of the automatic driving vehicle collision test equipment, the data fusion algorithm adopts a dynamic weight adjustment mechanism, which specifically includes the following steps: Evaluate the confidence index of each sensor data in real time, including: Calculate the confidence based on the measurement noise level of the lidar C 雷达 , where,C 雷达 = 1 / ( σ 雷达 2 + ε ), σ 雷达 is the standard deviation of the current frame data of the lidar, ε is a very small constant to prevent the denominator from being zero; Calculate the confidence based on the image sharpness and feature matching degree of the vision sensor C 视觉 , where, C 视觉 = S 视觉 × M 匹配 , S 视觉 is the image sharpness score, M 匹配 is the matching accuracy of the target feature points; Calculate the confidence based on the number of satellites and signal strength of the GPS positioning device C 定位 , where, C 定位 = N 卫星 × I 信号 , N 卫星 is the number of visible satellites, I 信号 is the signal strength normalization value; Calculate the confidence based on the update timeliness and matching error of the high-precision map C 地图 , where, C 地图 = 1 / (△ t 更新 + E 匹配 ), △ t 更新 is the map data update time difference, E 匹配 is the position matching error; Assign dynamic weights to D 定位 、 D 雷达 、 D 视觉 and D 地图 respectively, W 定位 、 W 雷达 、W 视觉 and W 地图 , the weight calculation formula is: W j = C j / (Σ C j ) Wherein, j ∈ {positioning, radar, vision, map}; j is the index variable of the sensor type, C j is the confidence index of each sensor, C j includes C 雷达 , C 视觉 , C 定位 , C 地图 ; Σ C j is the sum of the confidence of all sensors; Calculate the fused relative distance through the weighted fusion formula: D 融合 = W 定位 × D 定位 + W 雷达 × D 雷达 + W 视觉 × D 视觉 + W 地图 × D 地图 .

[0063] The confidence C 雷达 can be calculated based on the measurement noise level of the lidar, σ 雷达 where the standard deviation ε is set to 1×10 -6 . The image sharpness score S 视觉 of the vision sensor ranges from 0.8 - 1.0, and the feature point matching accuracy M 匹配 is 0.7 - 0.95. The number of visible satellites of the GPS positioning deviceN 卫星 There should be ≥ 4 pieces, and the signal strength normalization value I 信 ≥ 0.7. The update time difference △ of the high-precision map t 更新 ≤ 5s, and the position matching error E 匹配 ≤ 0.5m. The lidar can be installed in the center of the front bumper of the test vehicle, the vision sensor can be arranged on both sides of the vehicle windshield, the GPS module can be fixed in the center of the roof, and the high-precision map data can be stored in the in-vehicle computing unit. During operation, the lidar collects point cloud data in real time and calculates the standard deviation, and the vision sensor evaluates the sharpness and matching degree through image processing algorithms.

[0064] The dynamic weight calculation formula is W j = C j / (Σ C j ), where C j is the confidence of each sensor. For example, if C 雷达 = 0.4, C 视觉 = 0.3, C 定位 = 0.2, C 地图 = 0.1, then the weights are 0.4, 0.3, 0.2, and 0.1 respectively. If the confidence of a certain sensor is lower than the threshold C 阈值 = 0.2 for 3 consecutive frames, it can be determined that it fails and triggers the compensation mechanism. The weight calculation module can be integrated into the in-vehicle computing unit, and the real-time update frequency is 100ms. During operation, the system dynamically adjusts the weights according to the sensor status. For example, when the GPS signal is lost, it reduces W 定位 , and increases W 雷达 and W 视觉 .

[0065] The weighted fusion formula is D 融合 = W 定位 × D 定位 + W 雷达 × D 雷达 + W 视觉 × D视觉 + W 地图 × D 地图 For example, D 定位 = 10.0m, D 雷达 = 9.8m, D 视觉 = 10.2m, D 地图 = 10.1m, when the weights are 0.2, 0.5, 0.2, and 0.1 respectively, D 融合 = 9.89m. The fused data can be further adjusted by the slope correction formula D 修正 = D 融合 / cosθ (slope angle θ ≤ 15°). The calculation module can select an embedded processor that supports parallel operations, and the data output frequency is 10Hz. During the working process, the fusion result is transmitted to the avoidance decision-making module in real time to trigger the action of the mechanical actuator.

[0066] Through dynamic weight allocation and multi-sensor data fusion, this solution can effectively suppress the measurement deviation caused by the noise of a single sensor or environmental interference (such as rain, fog, and strong light), and improve the relative distance calculation accuracy. When part of the sensor fails, the system can still maintain a reliable output through weight adjustment, reduce false triggering or missed triggering caused by data errors, and enhance the stability and safety of the test process.

[0067] Based on the existing data fusion, this solution innovatively introduces a dynamic weight adjustment mechanism. By quantifying the confidence of each sensor data in real time (such as lidar noise, visual image quality, GPS signal strength, and map timeliness), the fusion weights are dynamically allocated. Compared with traditional fixed weight or simple filtering methods, this technology can adaptively suppress noise interference and compensate for environmental mutations (such as GPS signal loss and visual occlusion), significantly improving the fusion accuracy. Especially when there is a conflict in multi-sensor data (such as a sudden increase in lidar noise on a rainy and foggy day), it can automatically reduce the weight of the low-confidence sensor (such as C 雷达 ), avoiding error accumulation and solving the fusion deviation problem caused by the failure of a single sensor in complex scenarios, further enhancing the robustness and adaptability of the system.

[0068] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those skilled in the art, additional modifications can be easily made. Therefore, without departing from the general concept defined by the claims and the equivalent scope, the present invention is not limited to the specific details and the embodiments shown and described herein.

Claims

1. A method for reducing the loss of collision test equipment for autonomous vehicles, characterized in that, Including the following steps: Obtain preset test vehicle parameters, where the test vehicle parameters include the maximum deceleration of the automatic emergency braking system a max , test vehicle speed v 试验 and the communication system delay time t 延迟 ; When the emergency braking system of the test vehicle is triggered, the emergency avoidance trigger time is calculated through the formula t 触发 = v 试验 / a max + t 延迟 t 触发 ;​ After collecting the position data of the test vehicle and the background target in real time, calculate the real-time relative distance D , and through the formula t 碰撞 = D / v 试验 Calculate the collision time t 碰撞 ; When t 碰撞 ≤ t 触发 , the stationary background target is driven by a mechanical actuator to move away from the collision direction.

2. The method for reducing the loss of the collision test equipment of an autonomous vehicle according to claim 1, characterized in that, Also including the following steps: Use a GPS positioning device to obtain the GPS information and acceleration data of the test vehicle and the background target; respectively set up communication system delay monitoring devices on the platforms of the test vehicle and the background target. t 延迟 is the sum of the delay time measured by the communication system delay monitoring device at the vehicle end and the delay time measured by the communication system delay monitoring device at the target end; When obtaining the pre-set test vehicle parameters, obtain the maximum deceleration of the test vehicle a 测试 ; At consecutive time points t 1 and t 2 record the GPS coordinates of the test vehicle and the background target, and calculate the relative displacement △ t 1,[[]] t 2 at the coordinates of the moment s 定位 ; Through the formula v 相对 = △ s 定位 / ( t 1 - t 2 )calculate the relative speed based on GPS information v 相对 ; According to the real-time relative distance D and v 相对 , combined with a 测试 , through the displacement formula of uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 测试 t 2 , recalculate the collision time to obtain the new collision time t 新 .

3. The method for reducing the loss of the automatic driving vehicle collision test equipment according to claim 2, characterized in that Obtain the actual deceleration through the acceleration sensor built in the GPS positioning device a 实际 , and compare it with a 测试 : If a 实际 = a 测试 , the collision time is t 新 ; If a 实际 < a 测试 , combined with the real-time relative distance D and v 相对 , substitute a 实际 back into the displacement formula of uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 实际 t 2 , recalculate the collision time to obtain the adjusted collision time t 调整 ; If a 实际 > a 测试 , perform a calibration check on the sensor data. If after calibration a 实际 > a 测试 , analyze whether there is a system anomaly. If there is an anomaly, issue an alarm and stop the test. If there is no anomaly, combine the real-time relative distance D and relative speed v 相对 , and substitute a 实际 back into the displacement formula for uniformly variable rectilinear motion D = v 相对 t - 1 / 2 a 实际 t 2 , recalculate the collision time to obtain the adjusted collision time t 调整 .

4. The method for reducing the loss of the collision test equipment for autonomous vehicles according to claim 2, characterized in that, Also including the following steps: Install a lidar and a plurality of vision sensors on the test vehicle, and install a lidar on the background target; set a slope sensor in the test site and use a laser profilometer; the lidars on the test vehicle and the background target respectively measure the distance to each other, and the vision sensors collect the surrounding environment images; the slope sensor obtains the slope information of the test site in real time, and the laser profilometer measures the ground flatness to obtain high-precision map data; Calculate the relative distance between the test vehicle and the background target using the GPS positioning device information D 定位 ; Process and analyze the lidar measurement data to obtain the relative distance based on the lidar D 雷达 ; Process the images collected by the vision sensor to obtain the relative distance based on the vision sensor D 视觉 ; Match the position information of the test vehicle and the background target with the high-precision map to obtain the relative distance based on the high-precision map D 地图 ; Adopt a data fusion algorithm to fuse D 定位 , D 雷达 , D 视觉 and D 地图 to obtain the fused relative distance D 融合 . According to the slope angle θ obtained by the slope sensor, where 0° ≤ θ < 15°. If the vehicle and the target are on the same straight slope surface, use the formula D 修正 = D 融合 / cosθ to correct D 融合 and obtain the distance D 修正 after slope correction. If the vehicle and the target are not on the same straight slope surface, skip the correction and directly use D 融合 as D 修正 .; The laser profilometer divides the ground into multiple small areas and measures the undulation height of each small area h i and length l i , i = 1, 2, …, n , i indicating the number of areas, calculates the actual distance increment of each small area , and accumulates the actual distance increments of all small areas to obtain the final and more accurate real-time relative distance D = D 修正 + △ L .

5. The method for reducing the loss of an autonomous vehicle collision test device according to claim 3, characterized in that, Also including: When the emergency braking system of the test vehicle is triggered and reaches the maximum deceleration of the automatic emergency braking system a max After that, based on the maximum deceleration of the automatic emergency braking system a max And the real-time relative speed v 相对 , through the kinematic formula s 最短 = v 2 相对 / (2 a max ) Calculate the theoretical shortest braking distance s 最短 ; If the remaining real-time relative distance in the braking phase D ≤ s 最短 , it is determined that braking alone cannot avoid a collision; When and only when the following conditions are simultaneously satisfied, the calculation module sends a trigger signal to the background target carrying system through the communication link. After receiving the trigger signal, the background target carrying system starts the mechanical actuator to drive the background target away from the collision direction: 1) Starting from the test start time t The current time counted from 0 = 0s t Meet t 制动 ≤ t ≤ t 制动 + Collision time, where t 制动 Is the triggering time of the emergency braking system; 2) The emergency braking system of the test vehicle has been triggered, and the deceleration has reached the maximum deceleration of the automatic emergency braking system a max ; 3) The collision time is less than or equal to the trigger time; If a 实际 = a 测试 , the collision time is t 新 ; If a 实际 ≠ a 测试 , the collision time is t 调整 .

6. The method for reducing the loss of the automatic driving vehicle collision test equipment according to claim 5, characterized in that After the background target object carrying system receives the trigger signal, it calculates the collision risk coefficient R = D / D 参考 × v 相对 to determine the moving method of the background target object, where R has the dimension of m / s, D 参考 is a reference distance preset according to the performance of the test vehicle and the background target object; the method for determining the moving method of the background target object based on the collision risk coefficient is as follows: If R < R 低 , the mechanical actuator drives the background target with 30% of its maximum power, and the moving distance L 1 = k 1 × ( D - D 安全 ), t 移动 < t 碰撞 , t 移动 = L 1 / v 驱动 ; where k The value range of 1 is 1.1 - 1.3, t 移动 is the moving time, v 驱动 is the driving speed of the actuator, D 安全 is the safety distance threshold; and k 1 is dynamically selected according to the relative speed v 相对 and the relative distance D : When v 相对 ≤ 5m / s, and D > 10m, k 1 = 1.1; When v 相对 ≤ 5 m / s, and 5 m < D ≤ 10 m, k 1 = 1.2; When v 相对 ≤ 5 m / s, and D ≤ 5 m, k 1 = 1.3; If R 低 ≤ R < R 高 , the mechanical actuator drives the background target at 60% of its maximum power, and the moving distance L 2 = k 2 × ([[]]END]] D - D 安全 ), t 移动 < t 碰撞 , t 移动 = L 2 / v 驱动 ; where k The value range of 2 is 1.3 - 1.5; and k 2 is dynamically selected according to the relative speed v 相对 and the relative distance D : When 5 m / s < v 相对 ≤ 15 m / s, and D > 8 m, k 2 = 1.3; When 5 m / s < v 相对 ≤ 15 m / s, and 5 m < D ≤ 8 m, k 2 = 1.4; When 5 m / s < v 相对 ≤ 15 m / s, and D ≤ 5 m, k 2 = 1.5; If the collision risk coefficient R ≥ R 高 , regardless of v 相对 whether it is greater than 15 m / s, the mechanical actuator immediately drives the background target with maximum power and moves it to the nearest safe area along the shortest straight-line path. The moving distance L 3 = L 初始 + L 缓冲 , t 移动 < t 碰撞 , t 移动 = L 3 / v 驱动 , v 驱动 is the driving speed of the actuator; where L 初始 is the straight-line distance from the current position to the center of the safe area, L 缓冲 is a buffer distance of 5 - 10 m.

7. The method for reducing the loss of the automatic driving vehicle collision test equipment according to claim 6, characterized in that R 低 The value range is 1.2 - 2.5; R 高 The value range is 2.5 - 4.0; D 参考 = k × s 最短 ; k, D 参考 Dynamically match according to the following rules based on the test scenario: Urban road test: k = 1.5 - 2.0, D 参考 = 15 - 25 m; Highway test: k = 2.0 - 3.0, D 参考 ≥ 100m; Extreme condition test: k = 2.5 - 3.5, D 参考 = 5 - 10m; D 安全 The value range is 5.56 - 50 m and satisfies D ≤ D 安全 ≤ D 参考 and D 安全 = max(2 × L 车辆 1.2 × s 最短 , v 相对 × t 响应 ), L 车辆 is the total length of the test vehicle, t 响应 is the total time required from the system detecting a collision risk to the mechanical actuator starting to drive the background object to move, specifically including: sensor data processing time t 数据处理 , calculation module decision-making time t 计算 , t 延迟 , mechanical actuator startup time t 执行 ; Urban road test: t 响应 = 0.3 - 0.5 s; Highway test: t 响应 = 0.2 - 0.3 s; Extreme condition test: t 响应 = 0.1 - 0.2 s; v 相对 in m / s t 响应 in s L 车辆 and s 最短 and D 安全 all in m 8. The method for reducing the loss of the collision test equipment of an autonomous vehicle according to claim 4, wherein The data fusion algorithm adopts a dynamic weight adjustment mechanism, specifically including the following steps: Evaluate the confidence index of each sensor data in real time, including: Confidence calculation for the measurement noise level based on lidar C 雷达 , where C 雷达 = 1 / ( σ 雷达 2 + ε ), σ 雷达 is the standard deviation of the current frame data of the lidar, ε is a very small constant to prevent the denominator from being zero; Calculating confidence based on image sharpness and feature matching degree of visual sensor C 视觉 , where C 视觉 = S 视觉 × M 匹配 , S 视觉 is the image sharpness score, M 匹配 is the matching accuracy of target feature points; Calculating Confidence Based on the Number of Satellites and Signal Strength of a GPS Positioning Device C 定位 , where C 定位 = N 卫星 × I 信号 , N 卫星 is the number of visible satellites, I 信号 is the signal strength normalization value; Calculating confidence based on the timeliness of high-precision map updates and matching errors C 地图 , where C 地图 = 1 / (△ t 更新 + E 匹配 ), △ t 更新 is the time difference of map data updates, E 匹配 is the position matching error; Pair D 定位 、 D 雷达 、 D 视觉 and D 地图 are respectively assigned dynamic weights W 定位 、 W 雷达 、 W 视觉 and W 地图 , and the weight calculation formula is: W j = C j / (Σ C j ) Among them, j ∈ {positioning, radar, vision, map}; C j is the confidence index of each sensor, C j including C 雷达 , C 视觉 , C 定位 , C 地图 ; Calculate the fused relative distance through the weighted fusion formula: D 融合 = W 定位 × D 定位 + W 雷达 × D 雷达 + W 视觉 × D 视觉 + W 地图 × D 地图 。