EOL intelligent detection method and device for automatically driving vehicle to pass expressway section

Abstract

The invention provides an EOL intelligent detection method and device for automatically driving a vehicle to pass through a highway section, and the method comprises the steps: issuing a control task instruction containing a target driving speed and a lane center line reference path to an automatic driving system of a test vehicle after determining that the test vehicle enters a preset highway section; enabling the test vehicle to run along a preset path in an automatic driving mode; receiving whole vehicle sensing data of the tested vehicle in the automatic driving process in real time, and performing feature extraction processing on target sensing data related to the vehicle direction stability in the whole vehicle sensing data to obtain driving behavior features for representing the transverse dynamic state of the vehicle; and performing EOL intelligent detection on the test vehicle based on the driving behavior characteristics to obtain an EOL detection result. According to the method, the objectivity, the real-time performance and the intelligent level of dynamic detection are remarkably improved, the automation degree and the quality judgment efficiency of finished vehicle off-line testing are improved, and the efficient quality control requirement in a large-scale continuous production mode is met.

Description

Technical Field

[0001] This application relates to the field of vehicle intelligent manufacturing and quality inspection technology, and in particular to an intelligent EOL detection method and device for autonomous vehicles passing through highway sections. Background Technology

[0002] In existing vehicle end-of-life (EOL) testing, dynamic road testing relies on manual driving. Test drivers judge the vehicle's directional stability on highways based on their senses, which is highly subjective and lacks consistent evaluation standards. It is difficult to quantitatively identify anomalies such as steering wheel vibration, misalignment, and vehicle deviation during testing, and real-time data collection and analysis are impossible. The limited pace of manual driving affects overall production efficiency, and safety management is challenging in high-risk testing scenarios. Furthermore, test results are delayed, lacking a real-time feedback mechanism, making it impossible to promptly intercept defective vehicles, and failing to meet the objectivity, real-time nature, and systematic requirements of intelligent connected vehicles in testing. Summary of the Invention

[0003] The purpose of this application is to provide an intelligent EOL detection method and device for autonomous vehicles passing through highway sections, so as to alleviate the above-mentioned technical problems existing in the prior art.

[0004] In a first aspect, the present invention provides an intelligent EOL detection method for autonomous vehicles passing through highway sections, applied to a cloud subsystem, comprising: After confirming that the test vehicle has entered the preset highway section, a control task command containing the target driving speed and lane centerline reference path is issued to the test vehicle's autonomous driving system, so that the test vehicle can drive along the preset path in autonomous driving mode. The system receives real-time vehicle sensor data during the autonomous driving process of the test vehicle, extracts features from the target sensor data related to vehicle directional stability in the vehicle sensor data, and obtains driving behavior features that characterize the vehicle's lateral dynamics. Based on driving behavior characteristics, the test vehicle is subjected to intelligent EOL detection to obtain the EOL detection result.

[0005] In an optional implementation, after confirming that the test vehicle has entered a preset highway section, a control task command containing a target driving speed and a lane centerline reference path is issued to the test vehicle's autonomous driving system, causing the test vehicle to drive along the preset path in autonomous driving mode, including: Receive road segment status information sent by the roadside communication and sensing unit. The road segment status information is used to indicate whether the preset highway segment is occupied. Determine whether the current highway segment is in an idle state based on the road segment status information; When the system determines that the vehicle is in an idle state and the test vehicle has reached the entrance buffer zone, an entry authorization command is generated and sent to the test vehicle's autonomous driving system. After confirming that the test vehicle has actually entered the preset highway section, the control task command, which includes the target driving speed and the reference path of the lane center line, is sent to the autonomous driving system, so that the test vehicle can drive along the preset path in autonomous driving mode.

[0006] In an optional implementation, after confirming that the test vehicle has actually entered the preset highway section, a control task command including the target driving speed and lane centerline reference path is sent to the autonomous driving system, causing the test vehicle to drive along the preset path in autonomous driving mode, including: After confirming that the test vehicle has actually entered the preset highway section, spatial path points are generated based on the lane centerline reference path, and the desired motion trajectory is generated based on the spatial path points and the preset target driving speed. The aiming distance for lateral control is determined based on the desired motion trajectory, and a target tracking point is selected on the forward path based on the aiming distance. The front wheel steering angle is calculated based on the geometric relationship between the vehicle's current position and the target tracking point, and then converted into a steering control command. Generate longitudinal control commands based on the target driving speed and the actual vehicle speed; Based on steering control commands and longitudinal control commands, control task commands are determined and sent to the autonomous driving system of the test vehicle so that the test vehicle can perform tracking driving of the desired motion trajectory.

[0007] In an optional implementation, longitudinal control commands are generated based on the target driving speed and the actual vehicle speed, including: The system acquires the actual speed of the test vehicle while it is traveling on the highway in real time and calculates the speed deviation between the actual speed and the target speed. Error-driven adjustment is performed based on velocity deviation to generate an initial response quantity that is proportional to the current deviation amplitude; The speed deviation that persists within a preset segment is cumulatively calculated, and an error correction amount is generated based on the cumulative results to eliminate long-term deviations. A reverse adjustment amount is generated to suppress speed fluctuations based on the changing trend of the speed deviation; Based on the initial response, error correction, and reverse adjustment of the velocity deviation, corresponding longitudinal acceleration control parameters are generated, and longitudinal control commands are generated according to the longitudinal acceleration control parameters.

[0008] In an optional implementation, it further includes: During the test vehicle's autonomous driving process, the vehicle's positioning signal strength, control system operating status, and the results of perception of the road environment ahead are monitored in real time. When an abnormal event is detected, such as loss of vehicle positioning signal, no response to control commands, or the appearance of an unavoidable obstacle ahead, a minimum risk strategy command is generated. Send the minimum risk strategy instruction to the autonomous driving system of the test vehicle so that the test vehicle can decelerate and stop, activate the audible and visual alarms and remain parked. Log the abnormal event and suspend subsequent path planning tasks until a manual recovery instruction is received.

[0009] In an optional implementation, real-time vehicle sensor data during autonomous driving is received from the test vehicle. Feature extraction processing is performed on target sensor data related to vehicle directional stability within the vehicle sensor data to obtain driving behavior features characterizing the vehicle's lateral dynamics, including: Continuously receive message streams uploaded from the data acquisition device of the test vehicle. The message stream includes at least one or more signals such as steering wheel angle, steering torque, yaw rate, longitudinal acceleration, and left and right wheel speed difference. The steering wheel angle and steering torque are filtered to remove high-frequency noise interference. The standard deviation and instantaneous peak value of the filtered signal are calculated using a sliding time window to obtain the first behavioral feature used to characterize steering wheel vibration. While the vehicle is maintaining a straight line, the average value of the steering wheel angle over a preset time period is calculated to obtain a second behavioral feature used to characterize directional deviation. The yaw rate, longitudinal acceleration, and the speed difference between the left and right wheels are used as input parameters and input into a preset vehicle lateral kinematics model to calculate the estimated lateral displacement of the vehicle during driving. Based on the changing trend of the estimated lateral displacement, a third behavioral feature is determined to characterize the vehicle's deviation.

[0010] In an optional implementation, the test vehicle undergoes EOL intelligent detection based on driving behavior characteristics to obtain EOL detection results, including: The first behavioral feature is compared with the first type of judgment condition. If the standard deviation exceeds the allowable range in multiple consecutive time windows, it is determined that there is abnormal steering wheel vibration. The second behavioral feature is compared with the second type of judgment condition. If the average steering wheel angle deviates from the neutral position by more than the set tolerance, it is judged as incorrect direction. The third behavioral characteristic is compared with the third type of judgment condition. If the estimated lateral displacement continues to increase and exceeds the safety boundary, it is determined that the vehicle has deviated. When any judgment result is abnormal, an overall unqualified EOL test result is generated, and the specific abnormality type is marked.

[0011] Secondly, the present invention provides an EOL (End of Operation) intelligent detection device for autonomous vehicles passing through highway sections, applied to a cloud subsystem, comprising: The instruction issuing module is used to issue control task instructions containing the target driving speed and lane centerline reference path to the autonomous driving system of the test vehicle after confirming that the test vehicle has entered the preset highway section, so that the test vehicle can drive along the preset path in autonomous driving mode. The feature extraction module is used to receive the whole vehicle sensor data of the test vehicle in real time during the autonomous driving process, and to perform feature extraction processing on the target sensor data related to the vehicle's directional stability in the whole vehicle sensor data to obtain driving behavior features that characterize the vehicle's lateral dynamics. The intelligent detection module is used to perform EOL (End of Life) intelligent detection on the test vehicle based on driving behavior characteristics and obtain the EOL detection result.

[0012] Thirdly, the present invention provides an EOL (End of Operation) intelligent detection system for autonomous vehicles passing through highway sections, comprising: Vehicle terminal system, installed on the test vehicle; The roadside subsystem is deployed on dynamic test sections, including highway sections. The cloud subsystem is communicatively connected to the vehicle terminal subsystem and the roadside subsystem, wherein the cloud subsystem is used to execute the EOL intelligent detection method for autonomous driving of vehicles passing through highway sections as described in any of the foregoing embodiments.

[0013] Fourthly, the present invention provides an electronic device including a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the EOL intelligent detection method for autonomous driving of a vehicle passing through a highway section according to any of the foregoing embodiments.

[0014] The vehicle autonomous driving method and device provided in this application, which utilizes an intelligent EOL (End of Lane) detection system on highways, achieves standardized autonomous driving on highways by issuing control task commands containing target speed and lane centerline reference paths to the test vehicle. This eliminates operational differences caused by manual driving and improves test consistency and safety. Real-time reception of vehicle sensor data and extraction of driving behavior features related to directional stability enable objective quantitative analysis of dynamic performance issues such as steering wheel vibration, steering inaccuracy, and vehicle deviation, overcoming the subjective limitations of relying on sensory judgment. Based on driving behavior features, intelligent EOL detection enables real-time feedback and immediate anomaly identification during the testing process, effectively supporting closed-loop quality control. Overall, this method significantly improves the objectivity, real-time performance, and intelligence of dynamic detection, enhancing the automation and quality assessment efficiency of vehicle off-line testing, and meeting the high-efficiency quality control requirements of large-scale continuous production. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 A flowchart of an EOL (End of Hour) intelligent detection method for autonomous driving of a vehicle passing through a highway section, provided in an embodiment of this application; Figure 2 A flowchart illustrating an EOL (Electronic Occurrence Limit) intelligent detection method for autonomous driving of a pure electric vehicle passing through a highway section after it rolls off the production line, provided as an embodiment of this application. Figure 3 A structural diagram of an EOL (End of Hour) intelligent detection device for autonomous driving of a vehicle passing through a highway section, provided in an embodiment of this application; Figure 4 A structural diagram of an EOL (End of Hour) intelligent detection system for autonomous vehicles passing through highway sections, provided in an embodiment of this application; Figure 5 This is a structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0018] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0019] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0020] This application provides an intelligent EOL detection method for autonomous vehicles passing through highway sections, applied to a cloud subsystem. See [link to relevant documentation]. Figure 1As shown, the method mainly includes the following steps: S110, after confirming that the test vehicle has entered the preset highway section, issues a control task command containing the target driving speed and lane centerline reference path to the test vehicle's autonomous driving system, so that the test vehicle can drive along the preset path in autonomous driving mode.

[0021] The aforementioned control task instructions are structured operation commands generated in the cloud and sent to the vehicle's autonomous driving system. These instructions include specifying the target driving speed and lane centerline reference path that the vehicle should follow on highways. The target driving speed is a pre-set constant speed value used to ensure that all test vehicles complete the test under the same conditions. The lane centerline reference path is a trajectory line composed of a series of continuous spatial points constructed based on a high-precision map, representing the theoretical driving centerline that the vehicle should maintain within the lane.

[0022] By encapsulating the aforementioned parameters into executable instructions, standardized guidance for vehicle dynamics is achieved. In practice, when a vehicle approaches a highway entrance, the cloud verifies its location via roadside communication units or onboard positioning information. Once the entry conditions are met and the road segment is clear, a control task instruction is immediately sent to the vehicle's autonomous driving domain controller. Upon receiving the instruction, the autonomous driving system parses the path and speed information and invokes the lateral and longitudinal control systems to drive the vehicle autonomously along the specified path and speed, ensuring the consistency and controllability of the testing process.

[0023] S120 receives real-time vehicle sensor data from the test vehicle during autonomous driving, extracts features from target sensor data related to vehicle directional stability, and obtains driving behavior features that characterize the vehicle's lateral dynamics.

[0024] Vehicle sensor data consists of multi-source real-time signals acquired from the vehicle bus network by the onboard data acquisition unit, including steering wheel angle, steering torque, wheel speeds, inertial measurement unit output, and the internal state of the autonomous driving system. Target sensor data consists of physical quantities directly related to the vehicle's lateral motion stability, such as the trend of steering wheel angle changes, yaw rate fluctuations, and the difference in speed between the left and right wheels.

[0025] The aforementioned feature extraction process refers to filtering, aligning, and mathematically transforming the original signal to identify key indicators that reflect specific abnormal patterns. In practice, the data acquisition unit connects to CANFD or vehicle Ethernet via the OBD interface and continuously uploads message streams to the cloud. After time synchronization and noise filtering of the received data, the cloud uses a sliding window method to calculate statistical characteristics such as the standard deviation, mean shift, and frequency domain energy distribution of the signal, generating a behavioral feature set describing the vehicle's lateral dynamic response, providing a basis for subsequent intelligent judgment.

[0026] S130 performs EOL (End of Service) intelligent detection on test vehicles based on driving behavior characteristics, and obtains EOL detection results.

[0027] EOL (Effectiveness of Lift) intelligent detection based on driving behavior characteristics involves using a pre-set analysis model and extracted driving behavior features to automatically determine whether a vehicle has directional stability defects. The EOL detection results are structured outputs, including a pass / fail rating and a specific anomaly type identifier.

[0028] In practice, the cloud-deployed vehicle dynamic behavior analysis engine calls multiple parallel judgment modules: the steering wheel vibration analysis module determines whether there is resonance vibration by detecting the energy peak of the steering torque signal in a specific frequency band; the steering misalignment analysis module determines whether the neutral position is deviated based on the average deviation of the steering wheel angle during straight-line driving; and the vehicle pull-off analysis module integrates yaw rate, longitudinal speed, and wheel speed difference to estimate the lateral displacement deviation and determine whether it exceeds the tolerance threshold. Finally, the test decision module generates the final test result based on the logic that if any sub-item exceeds the standard, the entire system is deemed unqualified, and this result is fed back to the vehicle application interface in real time.

[0029] For ease of understanding, the EOL intelligent detection method for autonomous driving of vehicles passing through highway sections provided in the embodiments of this application will be described in detail below.

[0030] In one implementation, after confirming that the test vehicle has entered a preset highway section, a control task command containing the target driving speed and lane centerline reference path is issued to the test vehicle's autonomous driving system, causing the test vehicle to drive along the preset path in autonomous driving mode. In specific implementation, this may include the following steps 1.1 to 1.4: Step 1.1: Receive road segment status information sent by the roadside communication and sensing unit. The road segment status information is used to indicate whether the preset highway segment is occupied.

[0031] Roadside communication and sensing units are deployed at highway entrances or key locations. They possess communication and environmental sensing capabilities and can integrate sensors such as RFID readers, cameras, or LiDAR. The roadside communication and sensing units transmit road segment status information, which is data signals indicating whether a test vehicle is currently traveling on the highway segment, representing the occupancy or vacancy status of that segment. In practice, before a vehicle approaches the highway segment, the cloud receives status reports from the roadside units in real time via wired network or 5G / LTE communication to obtain the latest road segment usage information and determine whether the preset highway segment is occupied.

[0032] Step 1.2: Determine whether the current highway segment is in an idle state based on the road segment status information.

[0033] When determining whether a highway segment is currently idle, a logical judgment can be made based on the received segment status information to determine if no other test vehicles are currently traveling on that segment. This judgment process is executed by the highway segment scheduling module within the cloud server. If the segment status information shows that no vehicles are present, it is determined to be in an idle state; otherwise, it is marked as occupied, and a waiting instruction is maintained for requesting vehicles to ensure that no segment conflicts occur during multi-vehicle testing.

[0034] Step 1.3: When the test vehicle is determined to be in an idle state and has reached the entrance buffer zone, an entry authorization command is generated and sent to the test vehicle's autonomous driving system.

[0035] The entrance buffer zone is a waiting area located before the start of the highway section, used for temporary parking of vehicles preparing to enter for testing. The entry authorization command is a control signal generated by the scheduling module after the entry conditions are met, indicating that the vehicle is allowed to enter the highway section. In practice, the cloud platform, while confirming that the road section is clear, also needs to combine vehicle positioning data to determine whether it has arrived at the buffer zone. Only when both conditions are met simultaneously will an authorization command be sent to the corresponding vehicle's autonomous driving system, avoiding false triggering or premature entry, and ensuring the orderliness and safety of the testing process.

[0036] Step 1.4: After confirming that the test vehicle has actually entered the preset highway section, the control task command containing the target driving speed and the reference path of the lane center line is sent to the autonomous driving system, so that the test vehicle can drive along the preset path in autonomous driving mode.

[0037] The control task commands include the target driving speed (e.g., 50 km / h) and the lane centerline reference path (a desired trajectory composed of a series of high-precision spatial coordinate points). Confirmation that the test vehicle has actually entered the lane can be verified by either the location information uploaded by the onboard high-precision positioning system or the detection results from the roadside unit. Once entry is confirmed, the cloud immediately issues complete control commands, activating the lateral and longitudinal control functions of the autonomous driving system to accurately track the reference path and operate stably at the set speed, thus initiating the standardized EOL dynamic detection process.

[0038] The above approach, by introducing road segment status monitoring and a step-by-step authorization mechanism, achieves orderly scheduling and safe isolation of multi-vehicle testing on highways, avoiding delays and risks caused by manual coordination. Combining vehicle arrival status and road segment vacancy status for dual judgment improves the accuracy of scheduling decisions and the timeliness of response. Issuing control task instructions only after confirming actual entry ensures the precision of the timing of autonomous driving actions, ensuring that the testing process is controlled, consistent, and repeatable, effectively supporting a large number of vehicles to continuously and efficiently complete intelligent testing tasks.

[0039] Alternatively, the roadside units mentioned above can also employ a vehicle-to-vehicle communication scheme, where vehicles leaving the highway section can directly send a "roadway clear" broadcast to vehicles preparing to enter via C-V2X or DSRC technology. This scheme is more distributed, but it requires extremely high communication reliability.

[0040] Furthermore, after confirming that the test vehicle has actually entered the preset highway section, the control task command, including the target driving speed and the lane centerline reference path, is sent to the autonomous driving system, causing the test vehicle to drive along the preset path in autonomous driving mode. In specific implementation, this may include the following steps 2.1 to 2.5: Step 2.1: After confirming that the test vehicle has actually entered the preset highway section, generate spatial path points based on the lane centerline reference path, and generate the desired motion trajectory based on the spatial path points and the preset target driving speed.

[0041] The lane centerline reference path is an ideal driving path pre-planned and stored in a high-precision cloud map. It includes a series of continuous spatial coordinate points representing the vehicle's intended lane centering position. Spatial path points are path sampling points obtained by discretizing the reference path and are used in subsequent control algorithms. The desired trajectory is a dynamic driving target formed by overlaying time-dimensional information onto the spatial path points, including position, direction, and speed requirements at corresponding times.

[0042] In practice, after confirming that the vehicle has entered the highway section, the cloud service extracts the complete lane centerline reference path from the path planning module, and combines it with the set target driving speed (such as 50km / h) to optimize the density and time series allocation of the path points, generating the expected motion trajectory data that can be used for vehicle tracking.

[0043] Step 2.2: Determine the aiming distance for lateral control based on the desired motion trajectory, and select a target tracking point on the forward path based on the aiming distance.

[0044] The look-ahead distance is a defined distance value in the lateral control algorithm for autonomous driving. It represents the distance extended forward from the current rear axle center of the vehicle along the path ahead, used to determine the location of the next target to be tracked. The target tracking point is a specific spatial point located on the desired trajectory, and its position is determined by both the current vehicle pose and the look-ahead distance.

[0045] In practice, based on the vehicle's real-time uploaded position and heading information, the point closest to the current rear axle center and offset by the pre-aiming distance along the path direction can be searched on the desired trajectory as the target tracking point.

[0046] Step 2.3: Calculate the front wheel steering angle based on the geometric relationship between the vehicle's current position and the target tracking point, and convert the front wheel steering angle into a steering control command.

[0047] The geometric relationship between the vehicle's current position and the target tracking point is the angular deviation between the vehicle's current heading and the direction pointing towards the target tracking point, as well as the straight-line distance from the center of the vehicle's rear axle to the target tracking point. The front wheel steering angle is the ideal steering angle calculated based on this geometric relationship using a pure tracking algorithm or the Stanley algorithm, and is used to correct the vehicle's direction of travel to approach the target tracking point.

[0048] In practice, the cloud encapsulates the computational logic into executable instruction templates or transmits parameters to the vehicle's autonomous driving system to complete the calculations autonomously. The resulting front wheel steering angle is converted into steering control commands in standard CAN message format and sent to the electric power steering system via the vehicle communication interface to achieve lateral path tracking control.

[0049] Step 2.4: Generate longitudinal control commands based on the target driving speed and the actual vehicle speed.

[0050] The longitudinal control command is a control command used to adjust the vehicle's acceleration and deceleration behavior. This command is generated based on the difference between the target speed and the vehicle's current actual speed. In practice, a PID control strategy can be used for adjustment: the proportional part responds to the current speed error, the integral part eliminates long-term accumulated deviations, and the derivative part suppresses speed fluctuations. The controller outputs the desired acceleration or torque value and converts it into a braking or driving command that can be transmitted via the bus. By sending this longitudinal control command to the vehicle controller, which coordinates the motor and braking system to stabilize the vehicle's actual speed near the target value, ensuring consistency in test conditions.

[0051] Step 2.5: Determine the control task command based on the steering control command and the longitudinal control command, and send the control task command to the autonomous driving system of the test vehicle so that the test vehicle can perform the tracking driving of the desired motion trajectory.

[0052] The control task command is a complete set of driving action commands composed of steering control commands and longitudinal control commands, used to simultaneously control the vehicle's direction and speed. In practice, the cloud packages the two types of commands according to a predetermined communication protocol and transmits them with low latency to the autonomous driving domain controller of the test vehicle via the factory's WiFi network. After receiving the commands, the vehicle-side system parses the content and sends them to the electric power steering system and the vehicle controller, respectively, to achieve coordinated control of lateral path tracking and longitudinal speed stability. This enables the vehicle to strictly follow the desired trajectory to complete the autonomous driving test task on the highway.

[0053] The above method achieves refined remote guidance of vehicle autonomous driving behavior by decomposing the generation of expected motion trajectory, selection of preview points, and construction of steering and speed control commands. Combined with the lateral and longitudinal collaborative control mechanism, it ensures the trajectory accuracy and consistency of the test vehicle during high-speed driving. The entire process of generating and issuing control task commands is clear in structure and reliable in execution, effectively supporting the standardized collection and analysis of directional stability indicators in EOL testing, and improving the automation level and comparability of the testing process.

[0054] Furthermore, the aforementioned generation of longitudinal control commands based on the target driving speed and the actual vehicle speed may, in practice, include the following steps 3.1 to 3.5: Step 3.1: Obtain the actual speed of the test vehicle while it is traveling on the highway in real time, and calculate the speed deviation between the actual speed and the target speed.

[0055] The actual speed is the speed value collected and uploaded in real time by the vehicle's wheel speed sensors or inertial measurement unit, while the target speed is a pre-set standard test speed (e.g., 50 km / h). The difference between the two is the speed deviation, used to characterize whether the vehicle deviates from the ideal operating state. In practice, the cloud continuously acquires the actual speed data by parsing the vehicle's CAN messages and calculates the speed deviation between it and the target speed in real time, serving as the basic input for subsequent control adjustments.

[0056] Step 3.2: Based on the speed deviation, perform error-driven adjustment to generate an initial response quantity that is proportional to the current deviation amplitude.

[0057] The initial response corresponds to the proportional term output in PID control, and its magnitude is linearly related to the absolute value of the current speed deviation. When the actual vehicle speed is significantly lower or higher than the target value, the system immediately generates a corresponding acceleration or deceleration response to quickly approach the set speed. This mechanism improves the control system's responsiveness to instantaneous disturbances, ensuring that speed regulation has good dynamic response characteristics.

[0058] Step 3.3: Accumulate the speed deviation that persists within the preset segment, and generate an error correction amount to eliminate long-term deviations based on the accumulated results.

[0059] The error correction quantity corresponds to the integral term output in PID control. It is a compensation value generated by accumulating the speed deviation that has not been completely eliminated over a period of time. In practice, the system uses a sliding time window to perform integral calculations on historical deviations to identify steady-state errors caused by wind resistance, slope, or mechanical resistance. It then gradually applies additional adjustment to eventually bring the actual speed stably closer to the target value, avoiding long-term drift.

[0060] Step 3.4: Generate a reverse adjustment amount to suppress speed fluctuations based on the changing trend of the speed deviation.

[0061] The reverse adjustment quantity corresponds to the derivative control part in PID control, and it is based on how quickly the speed deviation changes over time. When a sharp fluctuation in speed is detected, the system generates an adjustment command in the opposite direction in advance, which acts as a damping effect, thereby improving the smoothness and stability of the control process and preventing overshoot or oscillation.

[0062] Step 3.5: Based on the initial response, error correction, and reverse adjustment of the velocity deviation, generate the corresponding longitudinal acceleration control parameters, and generate the longitudinal control command according to the longitudinal acceleration control parameters.

[0063] The above three items together constitute the complete PID control output, generating longitudinal acceleration control parameters that represent the desired acceleration or deceleration value of the vehicle. These parameters are encapsulated as control fields in a standardized communication protocol, transformed into braking or driving torque commands that can be executed via the vehicle network, and sent to the vehicle controller. The controller then coordinates the motor and braking system to work together to achieve precise speed tracking.

[0064] Furthermore, it also includes the following steps 4.1 to 4.4: Step 4.1: During the autonomous driving process of the test vehicle, monitor the vehicle positioning signal strength, the working status of the control system, and the perception results of the road environment ahead in real time.

[0065] Vehicle positioning signal strength refers to the signal quality of the high-precision GNSS and IMU integrated navigation system. The operating status of the control system includes feedback information from the autonomous driving domain controller, steering and braking actuators. The perception results of the road environment ahead come from the detection output of obstacles by sensors such as onboard cameras and millimeter-wave radar. In practice, the cloud continuously evaluates the system's operational health and external traffic conditions through diagnostic messages and perception data streams uploaded from the vehicle.

[0066] Step 4.2: When an abnormal event is detected, such as loss of vehicle positioning signal, no response to control commands, or the appearance of an unavoidable obstacle ahead, a minimum risk strategy command is generated.

[0067] The aforementioned minimum risk strategy instruction refers to a safety degradation command initiated when the system fails or a safety hazard exists. Once any of the above-mentioned abnormal situations is determined to be true, such as GNSS signal interruption exceeding the allowable duration, EPS failing to return execution confirmation, or forward perception confirming the presence of a stationary obstacle that cannot be bypassed, the instruction generation process is immediately triggered to ensure that the vehicle enters a controlled and safe state.

[0068] Step 4.3: Send the minimum risk strategy instruction to the autonomous driving system of the test vehicle so that the test vehicle can decelerate and stop, activate the audible and visual alarms, and remain in a parked state.

[0069] The minimum risk strategy comprises a series of orderly actions: first, the vehicle is smoothly decelerated until it comes to a stop; then, audible and visual warnings such as horn blaring or flashing lights are triggered to alert nearby personnel; and finally, the vehicle is locked via the electronic parking system to prevent it from rolling away. The entire process is completed autonomously by the vehicle-mounted system, ensuring operational reliability in emergency situations.

[0070] Step 4.4: Record the abnormal event and suspend subsequent path planning tasks until a manual recovery instruction is received.

[0071] Recording abnormal events involves logging all anomaly types, times, locations, and contextual data to the cloud-based log system for later traceability and analysis. Simultaneously, the route planning and service management module suspends task assignment to the vehicle to avoid further operations that could pose a greater risk. The testing process can only resume after the on-site operator confirms the problem has been resolved and manually issues a recovery command.

[0072] Furthermore, the aforementioned real-time reception of vehicle sensor data during autonomous driving involves extracting features from target sensor data related to vehicle directional stability to obtain driving behavior features characterizing the vehicle's lateral dynamics. In specific implementation, this may include the following steps 5.1 to 5.4: Step 5.1: Continuously receive message streams uploaded by the data acquisition device of the test vehicle. The message stream includes at least one or more of the following signals: steering wheel angle, steering torque, yaw rate, longitudinal acceleration, and left and right wheel speed difference.

[0073] The message stream is the raw communication data packet captured in real time from the vehicle's CANFD or in-vehicle Ethernet via the OBD-II interface, including status information from multiple key ECUs. These signals are the basic input source for analyzing the vehicle's lateral dynamic characteristics. In practice, the data acquisition device continuously uploads data at a high sampling frequency (e.g., above 100Hz) to ensure the capture of subtle dynamic changes.

[0074] Step 5.2: Filter the steering wheel angle and steering torque to remove high-frequency noise interference. Use a sliding time window to calculate the standard deviation and instantaneous peak value of the filtered signal to obtain the first behavioral feature used to characterize steering wheel vibration.

[0075] The sliding time window can be set to 1-3 seconds, and the standard deviation and maximum peak value of the signal within the window are calculated frame by frame. If the standard deviation exceeds the standard continuously or the energy is concentrated in a specific frequency band, it is identified as an abnormal jitter mode, and the first behavioral feature is generated to characterize whether there is resonance or instability caused by road excitation in the steering system.

[0076] Step 5.3: While the vehicle is maintaining a straight line, calculate the average value of the steering wheel angle over a preset time period to obtain the second behavioral feature used to characterize directional deviation.

[0077] The preset time period can cover the entire range of a vehicle traveling at a constant speed in a straight line on a highway (such as a distance of 100 meters or a duration of 10 seconds). During this period, the steering wheel should theoretically be in a near-neutral position. If the average steering angle deviates significantly from zero (e.g., more than 1.5 degrees), it indicates that although the vehicle can maintain straight-line driving, it needs to rely on continuous deflection angle compensation. This indicates that there is a problem with mechanical assembly deviation or inaccurate four-wheel alignment, thus obtaining the second behavioral characteristic.

[0078] Step 5.4: Input the yaw rate, longitudinal acceleration, and left and right wheel speed difference as input parameters into the preset vehicle lateral kinematics model to calculate the estimated value of the lateral displacement of the vehicle during driving, and determine the third behavioral feature used to characterize the vehicle deviation based on the changing trend of the estimated lateral displacement value.

[0079] The vehicle's lateral kinematics model is used to estimate the vehicle's actual lateral deviation relative to the lane centerline. The system integrates the yaw rate and longitudinal acceleration provided by the IMU, combines the speed difference between the left and right wheels to calculate the sideslip trend, and estimates the lateral displacement using an integral or Kalman filter algorithm. If this displacement continues to increase and exceeds a threshold (e.g., 0.2 meters), it is determined that the vehicle has exhibited unexpected deviation behavior, generating a third behavioral feature.

[0080] Furthermore, the above-mentioned intelligent EOL detection of the test vehicle based on driving behavior characteristics yields EOL detection results, including: (1) The first behavioral feature is compared with the first type of judgment condition. If the standard deviation in multiple consecutive time windows exceeds the allowable range, it is determined that there is abnormal steering wheel vibration. The first type of judgment condition is a pre-set vibration identification threshold rule. The system monitors the standard deviation of steering torque or angle signal in the sliding window. If the standard deviation exceeds the upper limit in three or more consecutive windows, it is considered that the vibration is persistent and significant, and an abnormal alarm is triggered.

[0081] (2) Compare the second behavioral feature with the second type of judgment condition. If the average steering wheel angle deviates from the neutral position by more than the set tolerance, it is judged as incorrect steering. The second type of judgment condition sets the quantitative boundary of incorrect steering (e.g., ±1.5 degrees). When the statistically obtained average steering wheel angle exceeds this range, it is marked as a defect in incorrect steering, indicating that the vehicle has a structural bias problem.

[0082] (3) Compare the third behavioral characteristics with the third type of judgment conditions. If the estimated lateral displacement continues to increase and exceeds the safety boundary, it is determined that the vehicle has veered off course. The third type of judgment conditions define the safety limit of the lateral displacement. The system judges whether the estimated lateral offset shows a monotonically increasing trend and exceeds the threshold. If it meets the criteria, the veergence behavior is confirmed, indicating that there may be a fault in the chassis or steering system.

[0083] When any judgment result is abnormal, an overall unqualified EOL test result is generated, and the specific abnormality type is marked. The final decision can adopt the "one-vote veto" principle, that is, as long as any of the steering wheel vibration, misalignment, or vehicle deviation is judged as abnormal, an unqualified conclusion is output, and the corresponding abnormality category (such as "misalignment" or "vehicle deviation") is marked in the result, which facilitates subsequent rework and quality traceability.

[0084] The above approach, by introducing a longitudinal speed control mechanism based on a PID structure, achieves precise and stable speed control of vehicles on highways, improving the consistency of test conditions. Combined with a safety monitoring system based on a minimum risk strategy, it enhances the reliability and emergency response capabilities of the autonomous driving testing process. By performing hierarchical feature extraction and classification of multi-source sensor data, an objective and quantifiable directional stability evaluation model is constructed, solving the subjective problem of traditional manual judgment. The resulting closed-loop detection logic supports real-time anomaly identification and immediate feedback, significantly improving the intelligence level and quality interception efficiency of EOL detection, and meeting the needs of efficient, safe, and accurate quality control in large-scale production environments.

[0085] Based on the above method, this application also provides a method for intelligent EOL detection of pure electric vehicles after they have rolled off the production line and are driving on highways. See [link to relevant documentation]. Figure 2 As shown, the method includes the following steps: Step S101: Test Initialization and Preparation. The vehicle is moved to workstation CP7, where the operator uses electrical testing tools to install the EOL application onto the central control screen. The vehicle automatically connects to the factory's WiFi. After completing the static test, the vehicle is moved to the starting point of the dynamic test area.

[0086] Step S102: Dynamic Road Test Initiation. The operator connects the data acquisition unit to the vehicle's OBD port. The operator selects "Dynamic Road Test" on the EOL application on the central control screen and clicks "Start Test". The EOL application sends a test start request to the cloud service management module.

[0087] Step S103: Cloud-based path planning and instruction issuance. The cloud-based path planning module generates an autonomous driving path for the vehicle that includes all dynamic test road segments, and issues the path instructions to the vehicle-side autonomous driving system.

[0088] Step S104: The vehicle autonomously navigates to the highway section. Based on the received path instructions, the vehicle's autonomous driving system controls the vehicle to sequentially traverse the previously tested road sections, including uneven and winding roads. The data acquisition device operates throughout the process, but the cloud may only perform in-depth analysis or simply record data from specific road sections.

[0089] Step S105: Request and Confirmation Before Entering the Highway Segment. When the vehicle's autonomous driving system detects an approach to the highway segment entrance through high-precision maps and positioning, it sends an entry request to the cloud-based highway segment scheduling module. The scheduling module queries the highway segment status; if it is available, it authorizes the vehicle to enter; if it is occupied, it instructs the vehicle to wait in the entrance buffer zone.

[0090] Step S106: Autonomous driving through the highway section. After obtaining authorization, the vehicle enters the highway section. The autonomous driving system controls the vehicle to accelerate to and maintain a stable speed of 50 km / h, driving in a straight line along the center line of the lane. This is the core step of the present invention to achieve autonomous driving. This process is not a simple cruise control, but a complex task that requires the integration of high-precision positioning, path tracking control, and status monitoring.

[0091] 1. Route Information Preloading: In step S103, the route instructions issued by the cloud include high-precision map data of the highway section. This data includes not only lane line geometry information, but also the specified driving speed (50km / h), the lane centerline reference path (represented by a series of dense waypoints (x_i, y_i)), and the identifiers for entering / exiting the area.

[0092] 2. Vehicle localization and perception: Comprehensive utilization of vehicle autonomous driving systems: (1) High-precision GNSS+IMU integrated navigation system: provides the vehicle's centimeter-level position (x, y) and heading angle φ in the current factory area coordinate system.

[0093] (2) Vehicle-mounted perception system: such as cameras and millimeter-wave radar, used to perceive lane lines, serve as redundancy verification and emergency safety assurance, and ensure that the vehicle always travels within the lane.

[0094] 3. Lateral control (path tracking): (1) Control objectives.

[0095] (2) Control algorithm: Use pure tracking algorithm or Stanley algorithm.

[0096] ① Pure tracking algorithm: This algorithm simulates human driving behavior, pre-aiming at a distance (pre-aiming distance L) in front of the vehicle and calculating the angle (i.e., steering angle) from the center of the rear axle of the vehicle to the pre-aiming point. Its core formula is: δ = arctan(2 * L * sin(α) / L) d ), where δ is the desired front wheel steering angle, L is the wheelbase, α is the angle between the vehicle's current heading and the direction of the aiming point, and L d δ represents the distance from the center of the vehicle's rear axle to the target point. By calculating δ in real time, the control system drives the electric power steering system to rotate the front wheels of the vehicle by a corresponding angle, thereby allowing the vehicle to continuously "chase" the target point ahead and achieve path tracking.

[0097] (3) Actuator: The autonomous driving domain controller calculates the target steering angle command and sends it to the electric power steering system via the CAN bus. The EPS then performs the steering action.

[0098] In addition to pure tracking and Stanley algorithm, lateral control can also employ more advanced control algorithms such as linear quadratic regulators and model predictive control to better handle vehicle nonlinear dynamics and constraints, and achieve smoother tracking results.

[0099] 4. Longitudinal control (speed control): (1) Control objective: To keep the actual speed v of the vehicle precisely stable at 50 km / h.

[0100] (2) Control algorithm: The classic PID control algorithm is adopted.

[0101] ① Proportional term: A basic control quantity is generated based on the speed error (target speed - actual speed).

[0102] ②Integral term: Accumulates historical speed error, used to eliminate steady-state error (such as overcoming speed decrease caused by wind resistance and rolling resistance).

[0103] ③ Differential term: Adjusted according to the rate of change of speed error to suppress speed fluctuations and make control smoother.

[0104] (3) Actuator: The output of the PID controller is the desired acceleration or torque. The autonomous driving domain controller sends this instruction to the vehicle controller via the CAN bus, and the VCU coordinates the motor controller and braking system to achieve precise acceleration and deceleration.

[0105] 5. Status Monitoring and Safety Redundancy: Throughout the autonomous driving process, the system continuously monitors its own status. If a positioning signal is lost, the control system malfunctions, or an insurmountable obstacle is detected ahead, the system will immediately execute a minimum risk strategy, such as smoothly decelerating, stopping, and triggering audible and visual alarms, while awaiting remote intervention.

[0106] Step S107: Real-time Data Acquisition and Upload. Throughout step S106, the data acquisition device collects relevant messages from the chassis, powertrain, and autonomous driving system at a high frequency and continuously streams them to the cloud via the factory's WiFi. This step is executed in parallel with S106 and S108.

[0107] Step S108: Real-time cloud-based analysis and anomaly detection. The cloud-based vehicle dynamic behavior analysis engine processes the uploaded data stream in real time, performing calculations and judgments on three models in parallel: steering wheel vibration, misalignment, and vehicle deviation.

[0108] 1. Data Preprocessing: After receiving the data stream, the cloud first performs data alignment, filtering, and invalid value removal. For example, the steering wheel torque signal is low-pass filtered to remove high-frequency noise and retain the frequency band related to jitter (such as 5-25Hz).

[0109] 2. Steering wheel vibration analysis: (1) Perform sliding window standardization on the filtered steering wheel torque signal.

[0110] (2) In the time domain, calculate the standard deviation of the signal within each time window. If the standard deviation continuously exceeds the threshold A, a jitter alarm is triggered.

[0111] (3) In the frequency domain, perform a Fast Fourier Transform on the signal within the window to convert the time-domain signal into a frequency-domain energy spectrum. Focus on monitoring the energy value of specific frequency bands (such as around 15Hz) related to the natural frequency of the steering system or road surface excitation. If the peak energy value of this frequency band exceeds the threshold B, it is determined that there is resonance vibration.

[0112] 3. Analysis of Incorrect Direction: (1) During the controlled straight-line driving of the vehicle, continuously read the value of the steering wheel angle sensor.

[0113] (2) Calculate the average value θ_avg of the steering wheel angle over a sufficiently long time period (e.g., driving through a 100-meter section of road).

[0114] (3) If |θ_avg|>θ_threshold (e.g., exceeding 1.5 degrees), then the vehicle is determined to be out of sync. This indicates that even if the vehicle can maintain a straight line through automatic control, there is a physical deviation in the neutral position of its steering wheel.

[0115] 4. Vehicle pull-off analysis: (1) Kinematics-based method: Given that the vehicle is initially located at the center of the lane, the lateral displacement of the vehicle is estimated by performing a second integral on the lateral acceleration provided by the IMU. This method is simple but the error will accumulate.

[0116] (2) The method based on the dynamic model (preferred): A two-degree-of-freedom dynamic model of the vehicle, including yaw rate and lateral acceleration, is established. Using a Kalman filter, the yaw rate, longitudinal velocity, and left and right wheel speed difference of the IMU are fused to make an optimal estimate of the actual lateral velocity and position of the vehicle. By comparing the estimated lateral position with the expected value (0), if its absolute value continuously exceeds the threshold (e.g., 0.2 meters), it is determined that the vehicle is veering off course.

[0117] Ultimately, the cloud-based decision-making module will set rules, such as: "If any of the three analysis items—steering wheel vibration, incorrect steering, or deviation—is judged as abnormal, then the overall result of this highway section test is unqualified." Furthermore, if the factory network conditions are excellent and the edge computing capabilities are strong enough, the vehicle dynamic behavior analysis engine can be deployed to the factory's edge computing nodes. This can further reduce analysis latency, alleviate the load on the cloud central server, and achieve a more refined real-time response.

[0118] Step S109: Exiting the highway section and test termination. The vehicle, using high-precision maps and positioning to sense that it has exited the highway section, reports a "exited" signal to the cloud. Upon receiving this signal, the cloud marks the section as idle and triggers the test decision module to make a final decision on the highway section test.

[0119] Step S110: Test Result Distribution and Display. The cloud-based test decision module will immediately distribute the test results, including "Pass" or "Fail" (and specific failure items), to the vehicle's EOL application. The EOL application will then prominently display the result on the vehicle's central control screen, informing the onboard operator or site monitoring personnel.

[0120] Step S111: Proceed to the next test section. The vehicle's autonomous driving system receives the next section route instructions from the cloud and continues to subsequent test sections such as the ABS testing area and the wading area.

[0121] Optionally, for test vehicles lacking advanced autonomous driving capabilities, a "guided vehicle" mode can be designed. In this mode, the lead vehicle is driven by a test driver, and subsequent test vehicles automatically follow the lead vehicle to complete dynamic road tests via vehicle-to-vehicle communication and vision-based following technology. In this case, the test vehicles still need to possess basic longitudinal and lateral automatic control capabilities.

[0122] In summary, this application combines autonomous driving technology with real-time cloud data analysis to achieve fully unmanned EOL (End-of-Life) testing of pure electric vehicles on highways after they roll off the production line, significantly improving testing safety and efficiency. The system utilizes vehicle-side data collection, roadside collaborative perception, and cloud-based intelligent analysis to objectively quantify key indicators such as steering wheel vibration, misalignment, and vehicle deviation, replacing the subjective evaluation of traditional manual testing and ensuring consistent testing standards and traceable results. Simultaneously, it achieves real-time closed-loop control of "collection-analysis-decision-feedback," enabling immediate identification and interception of potential quality defects, effectively preventing problematic vehicles from entering subsequent stages, and significantly improving the overall vehicle quality control level and production line cycle time.

[0123] Based on the above method embodiments, this application also provides an EOL (End of Hour) intelligent detection device for autonomous vehicles passing through highway sections, applied to a cloud subsystem, see [link to relevant documentation]. Figure 3 As shown, the device mainly includes the following parts: The instruction issuing module 310 is used to issue a control task instruction containing the target driving speed and the lane center line reference path to the autonomous driving system of the test vehicle after confirming that the test vehicle has entered the preset highway section, so that the test vehicle can drive along the preset path in autonomous driving mode. The feature extraction module 320 is used to receive the whole vehicle sensing data of the test vehicle in real time during the autonomous driving process, and to perform feature extraction processing on the target sensing data related to the vehicle's directional stability in the whole vehicle sensing data to obtain driving behavior features that characterize the vehicle's lateral dynamics. The intelligent detection module 330 is used to perform EOL intelligent detection on the test vehicle based on driving behavior characteristics and obtain the EOL detection result.

[0124] In one feasible implementation, the instruction issuing module 310 is specifically used for: Receive road segment status information sent by the roadside communication and sensing unit. The road segment status information is used to indicate whether the preset highway segment is occupied. Determine whether the current highway segment is in an idle state based on the road segment status information; When the system determines that the vehicle is in an idle state and the test vehicle has reached the entrance buffer zone, an entry authorization command is generated and sent to the test vehicle's autonomous driving system. After confirming that the test vehicle has actually entered the preset highway section, the control task command, which includes the target driving speed and the reference path of the lane center line, is sent to the autonomous driving system, so that the test vehicle can drive along the preset path in autonomous driving mode.

[0125] In one feasible implementation, the instruction issuing module 310 is specifically used for: After confirming that the test vehicle has actually entered the preset highway section, spatial path points are generated based on the lane centerline reference path, and the desired motion trajectory is generated based on the spatial path points and the preset target driving speed. The aiming distance for lateral control is determined based on the desired motion trajectory, and a target tracking point is selected on the forward path based on the aiming distance. The front wheel steering angle is calculated based on the geometric relationship between the vehicle's current position and the target tracking point, and then converted into a steering control command. Generate longitudinal control commands based on the target driving speed and the actual vehicle speed; Based on steering control commands and longitudinal control commands, control task commands are determined and sent to the autonomous driving system of the test vehicle so that the test vehicle can perform tracking driving of the desired motion trajectory.

[0126] In one feasible implementation, the instruction issuing module 310 is specifically used for: The system acquires the actual speed of the test vehicle while it is traveling on the highway in real time and calculates the speed deviation between the actual speed and the target speed. Error-driven adjustment is performed based on velocity deviation to generate an initial response quantity that is proportional to the current deviation amplitude; The speed deviation that persists within a preset segment is cumulatively calculated, and an error correction amount is generated based on the cumulative results to eliminate long-term deviations. A reverse adjustment amount is generated to suppress speed fluctuations based on the changing trend of the speed deviation; Based on the initial response, error correction, and reverse adjustment of the velocity deviation, corresponding longitudinal acceleration control parameters are generated, and longitudinal control commands are generated according to the longitudinal acceleration control parameters.

[0127] In one feasible implementation, the device further includes an anomaly detection and recording module for: During the test vehicle's autonomous driving process, the vehicle's positioning signal strength, control system operating status, and the results of perception of the road environment ahead are monitored in real time. When an abnormal event is detected, such as loss of vehicle positioning signal, no response to control commands, or the appearance of an unavoidable obstacle ahead, a minimum risk strategy command is generated. Send the minimum risk strategy instruction to the autonomous driving system of the test vehicle so that the test vehicle can decelerate and stop, activate the audible and visual alarms and remain parked. Log the abnormal event and suspend subsequent path planning tasks until a manual recovery instruction is received.

[0128] In one feasible implementation, the feature extraction module 320 is specifically used for: Continuously receive message streams uploaded from the data acquisition device of the test vehicle. The message stream includes at least one or more signals such as steering wheel angle, steering torque, yaw rate, longitudinal acceleration, and left and right wheel speed difference. The steering wheel angle and steering torque are filtered to remove high-frequency noise interference. The standard deviation and instantaneous peak value of the filtered signal are calculated using a sliding time window to obtain the first behavioral feature used to characterize steering wheel vibration. While the vehicle is maintaining a straight line, the average value of the steering wheel angle over a preset time period is calculated to obtain a second behavioral feature used to characterize directional deviation. The yaw rate, longitudinal acceleration, and the speed difference between the left and right wheels are used as input parameters and input into a preset vehicle lateral kinematics model to calculate the estimated lateral displacement of the vehicle during driving. Based on the changing trend of the estimated lateral displacement, a third behavioral feature is determined to characterize the vehicle's deviation.

[0129] In one feasible implementation, the aforementioned intelligent detection module 330 is specifically used for: The first behavioral feature is compared with the first type of judgment condition. If the standard deviation exceeds the allowable range in multiple consecutive time windows, it is determined that there is abnormal steering wheel vibration. The second behavioral feature is compared with the second type of judgment condition. If the average steering wheel angle deviates from the neutral position by more than the set tolerance, it is judged as incorrect direction. The third behavioral characteristic is compared with the third type of judgment condition. If the estimated lateral displacement continues to increase and exceeds the safety boundary, it is determined that the vehicle has deviated. When any judgment result is abnormal, an overall unqualified EOL test result is generated, and the specific abnormality type is marked.

[0130] The EOL intelligent detection device for autonomous vehicles passing through highways provided in this application has the same implementation principle and technical effects as the aforementioned method embodiments. For the sake of brevity, any parts not mentioned in the embodiments of the EOL intelligent detection device for autonomous vehicles passing through highways can be referred to the corresponding content in the aforementioned embodiments of the EOL intelligent detection method for autonomous vehicles passing through highways.

[0131] Based on the above method embodiments, this application also provides an EOL (Effective Occurrence Limit) intelligent detection system for autonomous vehicles passing through highway sections. See [link to relevant documentation]. Figure 4 As shown, the system includes: A vehicle terminal system is installed on the test vehicle. Specifically, it includes: (1) EOL Application APP: Deployed on the vehicle's central control screen as a human-machine interface. It is responsible for receiving operator instructions, displaying test items and execution instructions sent from the cloud, and showing the final test results.

[0132] (2) Data Acquisition Unit: Connects to the vehicle network (such as CAN FD, vehicle Ethernet) via the OBD-II interface or a dedicated vehicle data port. It is responsible for real-time acquisition of all relevant messages on the vehicle bus, including but not limited to data from ECUs such as the steering wheel angle sensor, torque sensor, wheel speed sensors, inertial measurement unit, vehicle stability control system, electric power steering system, and autonomous driving domain controller. This device has a built-in factory WiFi module, which can automatically connect to the network and stream data to the cloud.

[0133] (3) Autonomous driving system: As a core functional module of the vehicle, it plays the role of the actuator in this invention. It receives path planning instructions from the cloud or local storage, controls the vehicle's drive, braking and steering systems, and accurately completes the autonomous driving task of dynamic road test.

[0134] The roadside subsystem is deployed on dynamic test sections, including highway sections. Specifically, it includes: (1) Factory WiFi coverage: The entire dynamic test area, especially each test section, achieves seamless and high-speed factory WiFi network coverage to ensure the continuity and low latency of communication between the vehicle and the cloud.

[0135] (2) Roadside Communication and Sensing Unit: Deployed at key road entrances / exits, such as highway sections. This unit can integrate RFID readers, cameras, or LiDAR to detect the presence of vehicles within the road section. It communicates with a cloud server via a wired network or 5G / LTE to report the road section status.

[0136] A cloud-based subsystem communicates with both the vehicle-mounted terminal system and the roadside subsystem. This cloud-based subsystem is used to execute the EOL (End of Hour) intelligent detection method for autonomous vehicle passage through highway sections, as described in any of the foregoing embodiments. Specifically, it includes: (1) Path planning and service management module: Plans the optimal path for dynamic road testing for each test vehicle and manages its test status (pending testing, testing in progress, completed).

[0137] (2) Expressway Segment Dispatch Module: Receives road segment status information reported by roadside units and maintains an "Expressway Segment Occupancy Status Table". When a vehicle requests entry, dispatch authorization is granted according to the "one vehicle, one road segment" principle.

[0138] (3) Vehicle Dynamic Behavior Analysis Engine: The core algorithm module of this invention. It receives the message stream uploaded by the data acquisition device in real time and runs the following analysis model: ① Steering wheel vibration analysis model: Perform time-domain (e.g., standard deviation, peak detection) and frequency-domain (fast Fourier transform) analysis on steering wheel torque or angle signals to find high-frequency abnormal fluctuations related to road excitation or system resonance.

[0139] ② Directional Misalignment Analysis Model: On highways where vehicles should maintain straight-line driving, the deviation of the steering wheel center position is continuously monitored. The average steering wheel angle over a period of time is calculated; if its absolute value exceeds a preset threshold, it is determined to be directional misalignment.

[0140] ③ Vehicle drift analysis model: Combining the yaw rate, longitudinal acceleration, and left and right wheel speed difference of the IMU, the model uses Kalman filtering based on the dynamic model or directly calculates the lateral offset of the lane centerline to determine whether the vehicle has experienced unexpected lateral movement.

[0141] ④ Test Decision and Result Distribution Module: Integrates the output of the analysis engine, determines the test results of the highway section according to preset logic (such as failure if any indicator exceeds the standard), and immediately distributes the results to the corresponding vehicle's EOL application via the network.

[0142] This system achieves fully intelligent dynamic EOL (End of Hour) detection through collaboration between the vehicle, roadside, and cloud terminals. The vehicle-mounted autonomous driving system executes testing tasks, eliminating the subjectivity and safety risks of human driving; the roadside unit monitors road conditions in real time, ensuring the safety of multi-vehicle collaboration; and the cloud centrally performs path planning, scheduling management, and big data analysis, ensuring an orderly and efficient testing process and enabling objective quantitative judgment of key indicators such as steering wheel vibration, misalignment, and deviation. This three-terminal linkage constructs a closed loop of "perception-decision-control-feedback," significantly improving detection efficiency, consistency, and traceability, and providing a scalable technical framework for quality control of intelligent connected vehicles.

[0143] This application also provides an electronic device, such as... Figure 5 The diagram shows the structure of the electronic device 100, which includes a processor 51 and a memory 50. The memory 50 stores computer-executable instructions that can be executed by the processor 51. The processor 51 executes the computer-executable instructions to implement any of the above-mentioned intelligent EOL detection methods for autonomous driving of vehicles passing through highway sections.

[0144] exist Figure 5 In the illustrated embodiment, the electronic device further includes a bus 52 and a communication interface 53, wherein the processor 51, the communication interface 53, and the memory 50 are connected via the bus 52.

[0145] The memory 50 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 53 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 52 may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 52 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0146] The processor 51 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 51 or by instructions in software form. The processor 51 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or can be executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in the memory. The processor 51 reads the information in the memory and, in conjunction with its hardware, completes the steps of the EOL intelligent detection method for autonomous driving of vehicles passing through highway sections as described in the aforementioned embodiment.

[0147] This application also provides a computer-readable storage medium storing computer-executable instructions. When these computer-executable instructions are called and executed by a processor, they cause the processor to implement the aforementioned intelligent EOL detection method for autonomous driving of vehicles passing through highway sections. For specific implementation details, please refer to the foregoing method embodiments, which will not be repeated here.

[0148] The computer program product of the vehicle autonomous driving passage through highway section EOL intelligent detection method and device provided in this application includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation, please refer to the method embodiments, which will not be repeated here.

[0149] Unless otherwise specifically stated, the relative steps, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0150] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0151] In the description of this application, it should be noted that the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0152] In the description of this application, it should also be noted that, unless otherwise expressly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0153] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for intelligent detection of end-of-life (EOL) of a vehicle during autonomous driving on a highway, characterized in that, Applied to cloud subsystems, including: After confirming that the test vehicle has entered the preset highway section, a control task command containing the target driving speed and lane centerline reference path is issued to the test vehicle's autonomous driving system, so that the test vehicle can drive along the preset path in autonomous driving mode. The system receives real-time vehicle sensor data during the autonomous driving process of the test vehicle, performs feature extraction processing on the target sensor data related to vehicle directional stability in the vehicle sensor data, and obtains driving behavior features that characterize the lateral dynamics of the vehicle. Based on the driving behavior characteristics, the test vehicle is subjected to EOL intelligent detection to obtain the EOL detection result.

2. The intelligent EOL detection method for autonomous vehicles passing through highway sections according to claim 1, characterized in that, After confirming that the test vehicle has entered the preset highway section, a control task command containing the target driving speed and lane centerline reference path is issued to the test vehicle's autonomous driving system, causing the test vehicle to drive along the preset path in autonomous driving mode, including: Receive road segment status information sent by the roadside communication and sensing unit, the road segment status information being used to indicate whether a preset highway segment is occupied; Based on the road segment status information, determine whether the current highway segment is in an idle state; When the system determines that the vehicle is in an idle state and the test vehicle has reached the entrance buffer zone, an entry authorization command is generated and sent to the test vehicle's autonomous driving system. After confirming that the test vehicle has actually entered the preset highway section, a control task command containing the target driving speed and lane centerline reference path is sent to the autonomous driving system, so that the test vehicle can drive along the preset path in autonomous driving mode.

3. The intelligent EOL detection method for autonomous vehicles passing through highway sections according to claim 2, characterized in that, After confirming that the test vehicle has actually entered the preset highway section, a control task command containing the target driving speed and lane centerline reference path is issued to the autonomous driving system, causing the test vehicle to drive along the preset path in autonomous driving mode, including: After confirming that the test vehicle has actually entered the preset highway section, spatial path points are generated based on the lane centerline reference path, and the desired motion trajectory is generated according to the spatial path points and the preset target driving speed. The aiming distance for lateral control is determined based on the desired motion trajectory, and a target tracking point is selected on the forward path based on the aiming distance. The front wheel steering angle is calculated based on the geometric relationship between the vehicle's current position and the target tracking point, and the front wheel steering angle is converted into a steering control command; Generate longitudinal control commands based on the target driving speed and the actual vehicle speed; Based on the steering control command and longitudinal control command, a control task command is determined and sent to the autonomous driving system of the test vehicle so that the test vehicle can perform tracking driving of the desired motion trajectory.

4. The intelligent EOL detection method for autonomous vehicles passing through highway sections according to claim 3, characterized in that, Generate longitudinal control commands based on the target driving speed and the actual vehicle speed, including: The actual speed of the test vehicle during its journey on the highway is acquired in real time, and the speed deviation between the actual speed and the target speed is calculated. Error-driven adjustment is performed based on the speed deviation to generate an initial response quantity that is proportional to the current deviation amplitude; The speed deviation that persists within a preset segment is cumulatively calculated, and an error correction amount is generated based on the cumulative results to eliminate long-term deviations. A reverse adjustment amount is generated to suppress speed fluctuations based on the changing trend of the speed deviation; Based on the initial response, error correction, and reverse adjustment of the speed deviation, corresponding longitudinal acceleration control parameters are generated, and longitudinal control commands are generated according to the longitudinal acceleration control parameters.

5. The intelligent EOL detection method for autonomous vehicles passing through highway sections according to claim 3, characterized in that, Also includes: During the autonomous driving process of the test vehicle, the vehicle positioning signal strength, the working status of the control system and the perception results of the road environment ahead are monitored in real time. When an abnormal event is detected, such as loss of vehicle positioning signal, no response to control command, or the appearance of an unavoidable obstacle in front, a minimum risk strategy command is generated. The minimum risk strategy instruction is sent to the autonomous driving system of the test vehicle to cause the test vehicle to decelerate and stop, activate the audible and visual alarms, and remain in a parked state. The abnormal event will be recorded, and subsequent path planning tasks will be suspended until a manual recovery instruction is received.

6. The intelligent EOL detection method for autonomous vehicles passing through highway sections according to claim 1, characterized in that, The system receives real-time vehicle sensor data during autonomous driving, extracts features from target sensor data related to vehicle directional stability, and obtains driving behavior features characterizing the vehicle's lateral dynamics, including: The system continuously receives message streams uploaded from the data acquisition device of the test vehicle. The message streams include at least one or more signals such as steering wheel angle, steering torque, yaw rate, longitudinal acceleration, and left and right wheel speed difference. The steering wheel angle and steering torque are filtered to remove high-frequency noise interference. The standard deviation and instantaneous peak value of the filtered signal are calculated using a sliding time window to obtain the first behavioral feature used to characterize steering wheel vibration. While the vehicle is maintaining a straight line, the average value of the steering wheel angle over a preset time period is calculated to obtain a second behavioral feature used to characterize directional deviation. The yaw rate, longitudinal acceleration, and left and right wheel speed difference are input as parameters to a preset vehicle lateral kinematics model to calculate the estimated lateral displacement of the vehicle during driving, and a third behavioral feature to characterize vehicle deviation is determined based on the changing trend of the estimated lateral displacement.

7. The intelligent EOL detection method for autonomous vehicles passing through highway sections according to claim 6, characterized in that, Based on the driving behavior characteristics, the test vehicle undergoes EOL intelligent detection to obtain EOL detection results, including: The first behavioral feature is compared with the first type of judgment condition. If the standard deviation exceeds the allowable range within multiple consecutive time windows, it is determined that there is abnormal steering wheel vibration. The second behavioral feature is compared with the second type of judgment condition. If the average steering wheel angle deviates from the neutral position by more than the set tolerance, it is judged as incorrect direction. The third behavioral feature is compared with the third type of judgment condition. If the estimated lateral displacement continues to increase and exceeds the safety boundary, it is determined that the vehicle has deviated. When any judgment result is abnormal, an overall unqualified EOL test result is generated, and the specific abnormality type is marked.

8. An intelligent EOL detection device for autonomous vehicles passing through highway sections, characterized in that, Applied to cloud subsystems, including: The instruction issuing module is used to issue a control task instruction containing the target driving speed and the lane centerline reference path to the autonomous driving system of the test vehicle after confirming that the test vehicle has entered the preset highway section, so that the test vehicle can drive along the preset path in autonomous driving mode. The feature extraction module is used to receive the whole vehicle sensor data of the test vehicle in real time during the autonomous driving process, and to perform feature extraction processing on the target sensor data related to the vehicle's directional stability in the whole vehicle sensor data to obtain driving behavior features that characterize the vehicle's lateral dynamics. The intelligent detection module is used to perform EOL intelligent detection on the test vehicle based on the driving behavior characteristics and obtain the EOL detection result.

9. An intelligent EOL detection system for autonomous vehicles passing through highway sections, characterized in that, include: Vehicle terminal system, installed on the test vehicle; The roadside subsystem is deployed on dynamic test sections, including highway sections; A cloud subsystem is communicatively connected to the vehicle terminal subsystem and the roadside subsystem, wherein the cloud subsystem is used to execute the EOL intelligent detection method for autonomous driving of vehicles passing through highway sections as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, The system includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the EOL intelligent detection method for autonomous driving of a vehicle through a highway section as described in any one of claims 1 to 7.

Images