A power-on self-test method of an autonomous driving system

Abstract

This invention relates to a power-on self-test method for an autonomous driving system. The system includes: a controller detection module, used to acquire key HAD parameters, compare them with set thresholds, and determine whether the HAD status can support the stable operation of the autonomous driving system. Key parameters include CPU temperature < threshold, CPU load rate < threshold, CPU disk space > threshold, and CPU memory read / write error rate < threshold; and a communication detection module, used to send specific messages to acquire key network parameters, compare them with thresholds, and determine whether the network status is available. Key parameters include CAN / LIN. This invention not only enables comprehensive self-testing during the autonomous driving startup phase, ensuring vehicle driving safety, but also comprehensively and accurately detects the status of each module, promptly identifying potential faults or anomalies, thus effectively improving the system's safety and reliability.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a power-on self-test method for an autonomous driving system. Background Technology

[0002] With the rapid development of autonomous driving technology, Level 3 autonomous driving systems have gradually become a research hotspot. Level 3 autonomous driving technology mainly relies on three core components—perception, decision-making, and control—to achieve automated vehicle operation. The perception component acquires environmental information, including vehicle position, obstacle distance, and road conditions, through sensors such as LiDAR, millimeter-wave radar, and cameras. The decision-making component, based on Advanced Driver Assistance Systems (ADAS) and real-time data, performs path planning and behavioral decisions. The control component is responsible for executing specific vehicle operations, such as acceleration, braking, and steering. However, in existing technologies, the startup phase of autonomous driving systems often lacks a comprehensive self-checking mechanism, leading to the failure to detect potential faults in a timely manner, thus affecting the safety and stability of system operation. Summary of the Invention

[0003] In view of the above problems, the present invention provides a power-on self-test method for an autonomous driving system, which can not only perform a comprehensive self-test during the autonomous driving startup phase to ensure the driving safety of the vehicle, but also comprehensively and accurately detect the status of each module and promptly detect potential faults or anomalies, thereby effectively improving the safety and reliability of the system.

[0004] To achieve the above and other related objectives, the present invention provides the following technical solution: A power-on self-test method for an autonomous driving system, the system comprising: The controller detection module is used to acquire key HAD parameters, compare them with set thresholds, and determine whether the HAD state can support the stable operation of the autonomous driving system. Key parameters include CPU temperature < threshold, CPU load rate < threshold, CPU disk space > threshold, and CPU memory read / write error rate < threshold. The communication detection module is used to send specific messages to obtain key network parameters, compare them with thresholds, and determine whether the network status is available. Key parameters include: CAN / LIN: bus baud rate error < threshold, error frame rate < threshold, load rate < threshold; Ethernet: packet loss rate < threshold, latency < threshold, error rate < threshold, line utilization < threshold; middleware: specific message transmission and reception test to determine whether normal transmission and feedback are achieved, and to confirm the middleware status. The algorithm startup detection module is used to verify the integrity of model files and parameter files, check the self-check status of perception, decision-making, control, mapping, positioning, driving, and communication, and record and diagnose the self-check status of the log system. The sensor system detection module includes a sensor self-test unit and a sensor data verification unit. The sensor data verification unit is used to detect the raw data through the sensor driver and, following the principle of downstream detection of upstream, determine the sensor data frequency, timestamp, effective range of key values, general parameters of delay, and CRC-specific verification parameters in some protocols to determine the sensor data verification result.

[0005] Furthermore, the sensor self-test unit includes a camera self-test node, a lidar self-test node, a millimeter-wave radar self-test node, an ultrasonic radar self-test node, a combined navigation self-test node, and a chassis system self-test node.

[0006] Furthermore, the camera self-test node is used to collect images and send them back to the HAD according to the self-test command issued by the HAD. The HAD performs quality analysis on the image data, including image clarity, color reproduction, and whether there are any abnormalities such as blur, black screen, or screen distortion, thereby determining whether the camera is abnormal.

[0007] Furthermore, the lidar self-test node is used to perform a complete scan of the lidar according to the self-test command issued by the HAD and send the point cloud data back to the HAD. The HAD uses diagnostic algorithms to analyze whether there are any abnormalities such as data gaps or error points, thereby determining whether the lidar is abnormal.

[0008] Furthermore, the diagnostic algorithm includes: U1. Obtain the point cloud data information of the complete scan by the LiDAR; U2. Based on the point cloud data from the complete scan of the lidar, establish the probability density function Q of the lidar point cloud. , Where X represents the point cloud data information of the complete scan of the lidar, and α, β and γ are weighting coefficients that characterize the probability density value of the lidar point cloud to obtain the probability density value data information of the lidar point cloud. U3. Based on the probability density value of the point cloud of the lidar, a preset threshold is set. If the probability density value of the point cloud of the lidar is less than the preset threshold, there is an abnormal situation of data gaps or error points. This is used to determine whether the lidar is abnormal. If the probability density value of the point cloud of the lidar is greater than the preset threshold, the lidar is operating normally.

[0009] Furthermore, the weighting coefficients α, β, and γ are, , , , Where X represents the point cloud data from a complete scan by the lidar.

[0010] Furthermore, the millimeter-wave radar self-test node transmits its own operating status data back to the HAD according to the self-test command issued by the HAD. The HAD analyzes whether the millimeter-wave radar is abnormal by analyzing the transmission frequency, received signal strength, beam pointing, whether its own input signal is complete, and self-diagnostic status.

[0011] Furthermore, the ultrasonic radar self-test node is used to send self-test commands according to the HAD, and the ultrasonic radar returns its own working status data to the HAD. The HAD confirms the working status of the ultrasonic radar by analyzing key parameters such as transmission frequency, gain, and threshold, as well as the self-diagnostic status. The integrated navigation self-test node is used to send self-test commands according to the HAD, and the integrated navigation returns real-time working status data. The HAD confirms whether the integrated navigation is usable by analyzing key parameters such as convergence status, residual, state estimation error, and calibration status. The chassis system self-test node is used to simultaneously send small-angle steering simulation signals and braking simulation signals according to the self-test commands sent by the HAD. Through feedback, the steering status is judged by comparing the steering angle error and the power assist torque fluctuation value, and the braking status is judged by the hydraulic pressure deviation. The chassis self-test can only pass if both the braking and steering status are normal.

[0012] Furthermore, the system also includes a time synchronization detection module and a redundancy system detection module. The time synchronization detection module is used to obtain the system time of each system and the timestamp of each sensor data, judge the difference, and judge all time errors under the consideration of the error of the sensor data acquisition cycle. The maximum error is compared with a threshold, and the time of the P-Box is used as the truth value to judge whether it is consistent with the real time. When both errors are lower than the threshold, the time synchronization is considered to be normal. The redundancy system detection module is used to simulate sending a failure signal of the main system through HAD to confirm whether the switch to the redundant system is normal. It also confirms the data consistency between the main system and the redundant system. After the HAD simulates sending the failure signal of the main system, it calculates whether the time to switch to the redundant system is less than the threshold and whether the control error after the switch is less than the threshold. When both are less than the corresponding threshold, the redundancy system delay is considered to meet the system requirements.

[0013] Furthermore, the system also includes a security mechanism detection module, which sends a self-test signal based on HAD, simulates network attacks using the 5G module, determines whether network key and authentication detection is effectively blocked, ensures network security, and simulates risk signals based on HAD to determine whether the AEB function is executed normally, so as to ensure that the last line of defense can take effect in an emergency.

[0014] The present invention has the following positive effects: Through testing and comparison, this invention has found that the probability of non-algorithmic takeover in the later stages of the system that performs power-on self-test is significantly reduced. Furthermore, the more extreme the relevant threshold settings are, the lower the takeover rate, resulting in higher system stability. At the same time, it can comprehensively and accurately detect the status of each module and promptly detect potential faults or anomalies, thereby effectively improving the system's safety and reliability. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the system judgment process of the present invention; Figure 2 This is a flowchart illustrating the diagnostic algorithm of the present invention; Figure 3 This is a schematic diagram of the system framework of the present invention. Detailed Implementation

[0016] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0017] Example 1: As Figure 1 or Figure 3 As shown, a power-on self-test method for an autonomous driving system, the system comprising: The controller detection module is used to acquire key HAD parameters, compare them with set thresholds, and determine whether the HAD state can support the stable operation of the autonomous driving system. Key parameters include CPU temperature < threshold, CPU load rate < threshold, CPU disk space > threshold, and CPU memory read / write error rate < threshold. The communication detection module is used to send specific messages to obtain key network parameters, compare them with thresholds, and determine whether the network status is available. Key parameters include: CAN / LIN: bus baud rate error < threshold, error frame rate < threshold, load rate < threshold; Ethernet: packet loss rate < threshold, latency < threshold, error rate < threshold, line utilization < threshold; middleware: specific message transmission and reception test to determine whether normal transmission and feedback are achieved, and to confirm the middleware status. The algorithm startup detection module is used to verify the integrity of model files and parameter files, check the self-check status of perception, decision-making, control, mapping, positioning, driving, and communication, and record and diagnose the self-check status of the log system. The sensor system detection module includes a sensor self-test unit and a sensor data verification unit. The sensor data verification unit is used to detect the raw data through the sensor driver and, following the principle of downstream detection of upstream, determine the sensor data frequency, timestamp, effective range of key values, general parameters of delay, and CRC-specific verification parameters in some protocols to determine the sensor data verification result.

[0018] In this embodiment, the sensor self-test unit includes a camera self-test node, a lidar self-test node, a millimeter-wave radar self-test node, an ultrasonic radar self-test node, a combined navigation self-test node, and a chassis system self-test node.

[0019] In this embodiment, the camera self-test node is used to collect images and send them back to the HAD according to the self-test command issued by the HAD. The HAD performs quality analysis on the image data, including image clarity, color reproduction, and whether there are any abnormalities such as blur, black screen, or distorted screen, thereby determining whether the camera is abnormal.

[0020] In this embodiment, the lidar self-test node is used to perform a complete scan of the lidar according to the self-test command issued by the HAD and send the point cloud data back to the HAD. The HAD uses diagnostic algorithms to analyze whether there are any abnormalities such as data gaps or error points, thereby determining whether the lidar is abnormal.

[0021] In this embodiment, as Figure 2 As shown, the diagnostic algorithm includes: U1. Obtain the point cloud data information of the complete scan by the LiDAR; U2. Based on the point cloud data from the complete scan of the lidar, establish the probability density function Q of the lidar point cloud. , Where X represents the point cloud data information of the complete scan of the lidar, and α, β and γ are weighting coefficients that characterize the probability density value of the lidar point cloud to obtain the probability density value data information of the lidar point cloud. U3. Based on the probability density value of the point cloud of the lidar, a preset threshold is set. If the probability density value of the point cloud of the lidar is less than the preset threshold, there is an abnormal situation of data gaps or error points. This is used to determine whether the lidar is abnormal. If the probability density value of the point cloud of the lidar is greater than the preset threshold, the lidar is operating normally.

[0022] In this embodiment, the weighting coefficients α, β, and γ are, , , , Where X represents the point cloud data from a complete scan by the lidar.

[0023] Example 2: Based on the power-on self-test method of an autonomous driving system in Example 1, the present invention will be further explained and described below.

[0024] like Figure 1 or Figure 3 As shown, a power-on self-test method for an autonomous driving system, the system comprising: The controller detection module is used to acquire key HAD parameters, compare them with set thresholds, and determine whether the HAD state can support the stable operation of the autonomous driving system. Key parameters include CPU temperature < threshold, CPU load rate < threshold, CPU disk space > threshold, and CPU memory read / write error rate < threshold. The communication detection module is used to send specific messages to obtain key network parameters, compare them with thresholds, and determine whether the network status is available. Key parameters include: CAN / LIN: bus baud rate error < threshold, error frame rate < threshold, load rate < threshold; Ethernet: packet loss rate < threshold, latency < threshold, error rate < threshold, line utilization < threshold; middleware: specific message transmission and reception test to determine whether normal transmission and feedback are achieved, and to confirm the middleware status. The algorithm startup detection module is used to verify the integrity of model files and parameter files, check the self-check status of perception, decision-making, control, mapping, positioning, driving, and communication, and record and diagnose the self-check status of the log system. The sensor system detection module includes a sensor self-test unit and a sensor data verification unit. The sensor data verification unit is used to detect the raw data through the sensor driver and, following the principle of downstream detection of upstream, determine the sensor data frequency, timestamp, effective range of key values, general parameters of delay, and CRC-specific verification parameters in some protocols to determine the sensor data verification result.

[0025] In this embodiment, the millimeter-wave radar self-test node transmits its own operating status data back to the HAD according to the self-test command issued by the HAD. The HAD analyzes whether the millimeter-wave radar is abnormal by analyzing the transmission frequency, received signal strength, beam pointing, whether its own input signal is complete, and self-diagnostic status.

[0026] In this embodiment, the ultrasonic radar self-test node is used to send a self-test command according to the HAD. The ultrasonic radar returns its own working status data to the HAD. The HAD confirms the working status of the ultrasonic radar by analyzing key parameters such as transmission frequency, gain, and threshold, as well as the self-diagnostic status. The integrated navigation self-test node is used to send a self-test command according to the HAD. The integrated navigation returns real-time working status data. The HAD confirms whether the integrated navigation is usable by analyzing key parameters such as convergence status, residual, state estimation error, and calibration status. The chassis system self-test node is used to send a small-angle steering simulation signal and a braking simulation signal simultaneously according to the self-test command sent by the HAD. Through feedback, the steering status is judged by comparing the steering angle error and the power assist torque fluctuation value. The braking status is judged by the hydraulic pressure deviation. The chassis self-test can only pass if both the braking and steering status are normal.

[0027] In this embodiment, the system further includes a time synchronization detection module and a redundancy system detection module. The time synchronization detection module is used to obtain the system time of each system and the timestamp of each sensor data, determine the difference, and, considering the error of the sensor data acquisition cycle, determine all time errors, take the maximum error and compare it with a threshold. At the same time, using the P-Box time as the true value, it determines whether it is consistent with the real time. When both errors are lower than the threshold, the time synchronization is considered normal. The redundancy system detection module is used to simulate sending a main system failure signal through HAD to confirm whether the switch to the redundant system is normal, and to confirm the data consistency between the main system and the redundant system. After simulating sending the main system failure signal through HAD, it calculates whether the time to switch to the redundant system is less than the threshold, and whether the control error after the switch is less than the threshold. When both are less than the corresponding threshold, the redundancy system delay is considered to meet the system requirements.

[0028] In this embodiment, the system further includes a security mechanism detection module, which is used to send a self-test signal based on HAD, simulate network attacks by the 5G module, determine whether the network key and authentication detection is effectively blocked, ensure network security, and simulate risk signals based on HAD to determine whether the AEB function is executed normally, so as to ensure that the last line of defense can take effect in an emergency.

[0029] After the system starts up, the controller detection module first collects various key parameters of the CPU and compares them with preset thresholds. If all parameters meet the requirements, the process proceeds to the next step; otherwise, an alarm is triggered and the startup process is terminated.

[0030] The communication detection module sends specific messages to obtain key network parameters and compares them with thresholds. For example, in a CAN bus, if the baud rate error is less than 1%, the error frame rate is less than 0.1%, and the load rate does not exceed 70%, then the communication status is considered available.

[0031] The algorithm's startup detection module loads the model and parameter files and verifies their integrity. Simultaneously, it performs self-checks on the perception, decision-making, and control modules, logging the results for subsequent analysis.

[0032] The sensor system's detection module acquires raw data through the driver program and verifies the data's validity based on general parameters and protocol-specific verification parameters. For example, the frequency of a certain lidar should be maintained between 10Hz and 20Hz, and the timestamp deviation should not exceed 5ms.

[0033] This invention has been successfully applied to a Level 3 autonomous driving system. In real-world road tests, the system demonstrated excellent stability and reliability. For example, in a test under complex road conditions, the system quickly identified a sensor's timestamp deviation exceeding the allowable range through a power-on self-test and promptly issued a warning, preventing potential safety hazards. Furthermore, through continuous monitoring of the communication link, the system maintained low-latency and high-accuracy data transmission even under high load conditions, fully validating the practical application value of this invention.

[0034] In summary, this invention not only enables comprehensive self-checks during the autonomous driving startup phase to ensure vehicle driving safety, but also comprehensively and accurately detects the status of each module, promptly identifying potential faults or anomalies, thereby effectively improving the safety and reliability of the system.

[0035] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A power-on self-test method for an autonomous driving system, characterized in that, The method includes: The controller detection module is used to acquire key HAD parameters, compare them with set thresholds, and determine whether the HAD state can support the stable operation of the autonomous driving system. Key parameters include CPU temperature < threshold, CPU load rate < threshold, CPU disk space > threshold, and CPU memory read / write error rate < threshold. The communication detection module is used to send specific messages to obtain key network parameters, compare them with thresholds, and determine whether the network status is available. Key parameters include: CAN / LIN: bus baud rate error < threshold, error frame rate < threshold, load rate < threshold; Ethernet: packet loss rate < threshold, latency < threshold, error rate < threshold, line utilization < threshold; middleware: specific message transmission and reception test to determine whether normal transmission and feedback are achieved, and to confirm the middleware status. The algorithm startup detection module is used to verify the integrity of model files and parameter files, check the self-check status of perception, decision-making, control, mapping, positioning, driving, and communication, and record and diagnose the self-check status of the log system. The sensor system detection module includes a sensor self-test unit and a sensor data verification unit. The sensor data verification unit is used to detect the raw data through the sensor driver and, following the principle of downstream detection of upstream, determine the sensor data frequency, timestamp, effective range of key values, general parameters of delay, and CRC-specific verification parameters in some protocols to determine the sensor data verification result. The sensor self-test unit includes a camera self-test node, a lidar self-test node, a millimeter-wave radar self-test node, an ultrasonic radar self-test node, a combined navigation self-test node, and a chassis system self-test node. The self-test node of the lidar is used to perform a complete scan of the lidar according to the self-test command issued by the HAD and send the point cloud data back to the HAD. The HAD uses diagnostic algorithms to analyze whether there are abnormalities such as data holes or error points, thereby determining whether the lidar is abnormal. The diagnostic algorithm includes: U1. Obtain the point cloud data information of the complete scan by the LiDAR; U2. Based on the point cloud data from the complete scan of the lidar, establish the probability density function Q of the lidar point cloud. ， Where X represents the point cloud data information of the complete scan of the lidar, and α, β and γ are weighting coefficients that characterize the probability density value of the lidar point cloud to obtain the probability density value data information of the lidar point cloud. U3. Based on the probability density value of the point cloud of the lidar, a preset threshold is set. If the probability density value of the point cloud of the lidar is less than the preset threshold, there is an abnormal situation of data gaps or error points. This is used to determine whether the lidar is abnormal. If the probability density value of the point cloud of the lidar is greater than the preset threshold, the lidar is operating normally. The weighting coefficients α, β, and γ are, ， ， ， Where X represents the point cloud data from a complete scan by the lidar.

2. The power-on self-test method for an autonomous driving system according to claim 1, characterized in that: The camera self-test node is used to collect images and send them back to the HAD according to the self-test command issued by the HAD. The HAD performs quality analysis on the image data, including image clarity, color reproduction, and whether there are any abnormalities such as blur, black screen, or screen distortion, thereby determining whether the camera is abnormal.

3. The power-on self-test method for an autonomous driving system according to claim 1, characterized in that: The millimeter-wave radar self-test node transmits its own operating status data back to the HAD according to the self-test command issued by the HAD. The HAD analyzes the transmission frequency, received signal strength, beam direction, whether its own input signal is complete, and self-diagnostic status to determine whether the millimeter-wave radar is abnormal.

4. The power-on self-test method for an autonomous driving system according to claim 1, characterized in that: The ultrasonic radar self-test node is used to send self-test commands according to the HAD. The ultrasonic radar returns its own working status data to the HAD. The HAD confirms the working status of the ultrasonic radar by analyzing key parameters such as transmission frequency, gain, and threshold, as well as the self-diagnostic status. The integrated navigation self-test node is used to send self-test commands according to the HAD. The integrated navigation returns real-time working status data. The HAD confirms whether the integrated navigation is usable by analyzing key parameters such as convergence status, residual, state estimation error, and calibration status. The chassis system self-test node is used to send small-angle steering simulation signals and braking simulation signals simultaneously according to the self-test commands sent by the HAD. Through feedback, the steering status is judged by comparing the steering angle error and the power assist torque fluctuation value. The braking status is judged by the hydraulic pressure deviation. The chassis self-test can only pass if both the braking and steering status are normal.

5. The power-on self-test method for an autonomous driving system according to claim 1, characterized in that, The system further includes a time synchronization detection module and a redundancy system detection module. The time synchronization detection module is used to obtain the system time of each system and the timestamp of each sensor data, judge the difference, and judge all time errors under the consideration of the error of the sensor data acquisition cycle. The maximum error is compared with a threshold. At the same time, the time of the P-Box is used as the truth value to judge whether it is consistent with the real time. When both errors are lower than the threshold, the time synchronization is considered to be normal. The redundancy system detection module is used to confirm whether the switch to the redundant system is normal by simulating the sending of the main system failure signal through HAD. After confirming the data consistency between the main system and the redundant system and after simulating the sending of the main system failure signal through HAD, it calculates whether the switching time of the redundant system is less than the threshold and whether the control error after the switch is less than the threshold. When both are less than the corresponding threshold, the redundancy system delay is considered to meet the system requirements.

6. The power-on self-test method for an autonomous driving system according to claim 1, characterized in that, The system also includes a security mechanism detection module, which sends a self-test signal based on HAD, simulates network attacks using the 5G module, and determines whether the network key and authentication detection is effectively blocked to ensure network security. It also simulates risk signals based on HAD to determine whether the AEB function is executed normally, so as to ensure that the last line of defense can take effect in an emergency.

Images