Cleaning method, system and equipment for automatic driving sensor and storage medium

Abstract

The invention relates to the technical field of automatic driving, discloses a cleaning method, system and equipment for an automatic driving sensor and a storage medium, and is used for effectively reducing vibration and noise of a vehicle. The cleaning method of the automatic driving sensor comprises the steps of detecting whether the external environment of an automatic driving vehicle rains or not; if it is detected that the external environment is raining, a corresponding cleaning mode is selected according to the rainfall and the type of the sensor; if it is detected that the external environment does not rain, whether the surface of the sensor is fuzzy is detected; if it is detected that the surface of the sensor is fuzzy, a corresponding cleaning mode is selected according to the fuzzy degree and type of the sensor; and cleaning the sensor according to the selected cleaning mode, wherein the cleaning mode is fan cleaning or windscreen wiper cleaning.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a cleaning method, system, equipment and storage medium for an autonomous driving sensor. Background Art

[0002] In the autonomous driving industry, sensors are key components for sensing the external environment, and their cleanliness is directly related to the accuracy and reliability of the autonomous driving system. Currently, the industry generally uses air pumps as a high-pressure air source to clean sensors. However, this cleaning method has many drawbacks. First, the air pump generates significant vibration during operation, which can not only affect the normal operation of other vehicle components but also reduce passenger comfort. Second, the air pump's high noise level can disrupt the interior environment and affect the driving experience. Furthermore, the procurement, installation, and subsequent maintenance costs of air pump equipment are relatively high, increasing the overall cost of autonomous vehicles and hindering the promotion and application of the technology. Summary of the Invention

[0003] The present invention provides a cleaning method, system, equipment and storage medium for an autonomous driving sensor to solve the problem of high vibration and noise caused by the use of an air pump for cleaning in the prior art.

[0004] A first aspect of the present invention provides a method for cleaning a dynamic driving sensor, comprising: detecting whether it is raining in the external environment of an autonomous driving vehicle; if it is detected that it is raining, selecting a corresponding cleaning method based on the amount of rain and the type of sensor; if it is detected that it is not raining in the external environment, detecting whether the sensor surface is blurred; if it is detected that the sensor surface is blurred, selecting a corresponding cleaning method based on the degree of blur and type of the sensor; and cleaning the sensor according to the selected cleaning method, wherein the cleaning method is fan cleaning or wiper cleaning.

[0005] In a feasible embodiment, the detecting whether it is raining in the external environment of the autonomous driving vehicle includes: collecting image data or point cloud data of the external environment of the vehicle through a camera or a lidar; detecting whether there are raindrops, water mist or road surface reflection features based on the image data; if the detected raindrops, water mist or road surface reflection features are greater than a preset threshold, it is determined that the external environment is raining; based on the point cloud data, detecting whether there is raindrop reflection noise or point cloud density anomaly caused by ambient humidity; if the detected number of noise points exceeds a dynamic threshold or the point cloud scattering features conform to a rainy day model, it is determined that the external environment is raining.

[0006] In a feasible embodiment, the method for determining the amount of rainfall includes: when determining rain using image data collected by a camera, counting the number of raindrops in the image per unit area, if the number of raindrops is not greater than a first preset number threshold, determining it as light rain; if the number of raindrops is greater than the first preset number threshold and not greater than a second preset number threshold, determining it as moderate rain; if the number of raindrops is greater than the second preset number threshold, determining it as heavy rain; when determining rain using point cloud data collected by a lidar, counting the number of raindrop reflection noise points in the point cloud per unit volume, if the number of noise points is not greater than a first preset noise threshold, determining it as light rain; if the number of noise points is greater than the first preset noise threshold and not greater than the second preset noise threshold, determining it as moderate rain; if the number of noise points is greater than the second preset noise threshold, determining it as heavy rain.

[0007] In a feasible embodiment, detecting whether the sensor surface is blurred includes: collecting image data or point cloud data through a sensor, wherein the sensor is a camera or a lidar; analyzing a first fuzzy feature in the image data or a second fuzzy feature in the point cloud data; and judging whether the corresponding sensor surface is blurred based on the first fuzzy feature or the second fuzzy feature.

[0008] In a feasible embodiment, if the presence of sensor surface blur is detected, a corresponding cleaning method is selected according to the blur degree of the corresponding sensor, including: if the sensor surface blur is detected, calculating the image clarity index based on the image data, or calculating the point cloud density attenuation rate based on the point cloud data; determining the blur degree of the sensor based on the image clarity index or the point cloud density attenuation rate, the blur degree including light blur, moderate blur or heavy blur; selecting a corresponding cleaning method according to the blur degree and type of the sensor.

[0009] In a feasible implementation, the cleaning of the sensor according to the selected cleaning method includes: sending a cleaning instruction to the selected fan or wiper according to the selected cleaning method to start the cleaning operation; and performing fault diagnosis on the cleaning process and cleaning results.

[0010] In a feasible embodiment, the fault diagnosis of the cleaning process and cleaning results includes: monitoring the working status of the fan or wiper, and if an abnormal working status of the fan or wiper is detected during the cleaning process, immediately generating a fault report and stopping the cleaning operation; after the cleaning is completed, re-detecting whether the sensor surface is blurred; if the sensor surface is still blurred, determining that the corresponding sensor cleaning is invalid.

[0011] The second aspect of the present invention provides a cleaning system for an autonomous driving sensor, comprising: a sensor module for acquiring external environment data and fuzzy detection data; an autonomous driving controller for receiving the detection data of the sensor module, and selecting a corresponding cleaning method according to the amount of rainfall or the degree of fuzziness of the corresponding sensor, and the type of sensor, and generating corresponding cleaning control instructions; a control IO module for transmitting the cleaning control instructions of the autonomous driving controller to a fan or wiper; a fan for starting the fan according to the cleaning control instructions to blow clean the sensor surface; and a wiper for starting the wiper according to the control instructions to clean the sensor surface.

[0012] A third aspect of the present invention provides an electronic device comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the electronic device executes the above-mentioned method for cleaning the autonomous driving sensor.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium, which stores instructions that, when executed on a computer, enable the computer to execute the above-mentioned method for cleaning the autonomous driving sensor.

[0014] In the technical solution provided by the present invention, whether it is raining in the external environment of the autonomous driving vehicle is detected; if it is detected that it is raining in the external environment, a corresponding cleaning method is selected based on the amount of rain and the type of sensor; if it is detected that it is not raining in the external environment, whether the sensor surface is blurred is detected; if it is detected that the sensor surface is blurred, a corresponding cleaning method is selected based on the degree of blur and type of the sensor; the sensor is cleaned according to the selected cleaning method, which is fan cleaning or wiper cleaning. In an embodiment of the present invention, by detecting whether it is raining in the external environment of the autonomous driving vehicle and the degree of blur on the sensor surface, and flexibly selecting a corresponding low-vibration, low-noise cleaning method such as a fan or wiper according to the detection results and the type of sensor, the vibration and noise caused by traditional air pump cleaning are effectively reduced, ride comfort is improved, and equipment procurement and maintenance costs are reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of an embodiment of a cleaning method for an autonomous driving sensor according to an embodiment of the present invention;

[0016] Figure 2 This is a schematic diagram of another embodiment of a cleaning method for an autonomous driving sensor according to an embodiment of the present invention;

[0017] Figure 3 A schematic diagram of a system fault diagnosis flow chart in an embodiment of the present invention;

[0018] Figure 4 This is a schematic diagram of an embodiment of a cleaning system for an autonomous driving sensor according to an embodiment of the present invention;

[0019] Figure 5 FIG. 1 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] Embodiments of the present invention provide a cleaning method, system, device, and storage medium for an autonomous driving sensor, which reduces vehicle vibration and noise by flexibly selecting corresponding cleaning methods such as fan cleaning or wiper cleaning.

[0021] The terms "first," "second," "third," "fourth," and the like (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

[0022] It is understandable that the execution subject of the present invention may be a cleaning system for an autonomous driving sensor. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0023] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 An embodiment of a method for cleaning an autonomous driving sensor according to an embodiment of the present invention includes:

[0024] 101. Detect whether it is raining in the external environment of the autonomous driving vehicle;

[0025] The data obtained by the sensors of the autonomous vehicle can be used to determine whether it is raining in the external environment. The sensors of the autonomous vehicle include cameras and lidar.

[0026] When the sensor is a camera, image data of the autonomous vehicle's external environment is acquired through the sensor, analyzed, and a determination is made based on the analysis results regarding whether it is raining. Specifically, a pre-trained deep learning model is used to analyze parameters such as raindrop characteristics, sky brightness, and ground reflections in the image. Combined with meteorological data such as humidity and temperature acquired by the sensor, a comprehensive determination is made regarding whether it is raining. If the model outputs a probability of rain exceeding a set threshold, it is determined that it is raining. The deep learning model may be a weather recognition model based on a convolutional neural network.

[0027] When the sensor is a lidar, three-dimensional point cloud data is collected in real time through the sensor, and the point cloud data is preprocessed to filter out static background noise; the point cloud data is detected to see whether there are a large number of short-lived tiny scattering points, namely raindrop feature points, and the density and distribution pattern of high-dynamic noise points within the unit volume are counted. When the density exceeds the preset first threshold and the spatial distribution conforms to the raindrop falling trajectory model, the primary rainfall judgment is triggered; the laser echo intensity attenuation rate is calculated. Rain will cause the laser beam to be scattered more, and by comparing the echo intensity average of the current frame with the historical dry environment, the secondary rainfall judgment is triggered when the attenuation amplitude exceeds the preset second threshold; the point cloud motion characteristics are analyzed, and the rainfall motion vector field is constructed using continuous multi-frame point clouds to verify whether the noise points conform to the vertical motion pattern under gravitational acceleration. Finally, the confidence score of the three indicators is integrated. When the comprehensive score exceeds the set threshold and lasts for N scanning cycles, it is determined that the external environment of the autonomous driving vehicle is raining, and N is determined according to the actual situation, for example, N = 5.

[0028] 102. If it is detected that the external environment is raining, the corresponding cleaning method is selected according to the amount of rainfall and the type of sensor;

[0029] If it is detected that the external environment is raining, the rainfall is divided into different levels such as light rain, moderate rain and heavy rain according to the pre-set rainfall level standard, and the type of sensor is determined. The sensors are divided into windshield sensors and non-windshield sensors. According to the determined rainfall amount and sensor type, the corresponding cleaning method is selected. For example, if it is determined to be light rain and it is a windshield sensor, the first cleaning method is selected. The first cleaning method is to turn on the wiper at a low speed and wipe slowly at a preset low frequency to prevent rainwater from accumulating on the windshield; if it is determined to be moderate rain and it is a windshield sensor, the second cleaning method is selected. The second cleaning method is to adjust the wiper to a medium speed and wipe at a preset medium frequency to clear more rainwater in time; if it is determined to be heavy rain and it is a windshield sensor, the third cleaning method is selected. The third cleaning method is to adjust the wiper to a high speed and wipe at a high frequency to forcefully scrape off a large amount of rainwater. If it is determined to be light rain and the sensor is not a windshield type, the fourth cleaning method is selected. The fourth cleaning method is to turn on the fan and run it at a lower air volume to generate a gentle airflow to blow across the sensor surface and remove attached water droplets and a small amount of dust; if it is determined to be moderate rain and the sensor is not a windshield type, the fifth cleaning method is selected. The fifth cleaning method is to adjust the fan air volume to medium to enhance the airflow's drying and cleaning effect on the sensor; if it is determined to be heavy rain and the sensor is not a windshield type, the sixth cleaning method is selected. The sixth cleaning method is to adjust the fan to the maximum air volume to form a strong airflow to quickly blow away a large amount of rainwater on the sensor surface.

[0030] 103. If it is detected that the external environment is not raining, then the sensor surface is detected to see if it is blurred;

[0031] If the sensor is a camera, the image data is preprocessed, including denoising and color correction; the multi-dimensional features of the pre-processed image are extracted and analyzed, and the sensor surface is judged whether it is blurred based on the analysis results. Specifically: the energy proportion of the high-frequency component of the image is calculated, and the image is converted to the frequency domain through fast Fourier transform. When the high-frequency energy proportion is lower than the preset third threshold, the primary blur judgment is triggered; the image edge sharpness is analyzed, the Sobel operator is used to detect the edge, and the average gradient amplitude of the edge pixels is calculated. When the average gradient amplitude is lower than the preset fourth threshold, the secondary blur judgment is triggered; the local contrast of the image is detected, the image is divided into several blocks, and the Weber contrast of each block is calculated. When the proportion of low-contrast blocks exceeds the preset fifth threshold, the third blur judgment is triggered; at the same time, it is also possible to detect whether there are typical water stain features such as annular spots or striped refraction patterns in the image. Finally, the confidence scores of all features are combined. When the weighted score exceeds the set threshold, it is determined that the sensor surface is blurred.

[0032] If the sensor is a lidar, point cloud data is acquired through the lidar. The lidar emits a laser beam into the surrounding environment and receives the reflected signal, thereby acquiring point cloud data of the vehicle's surroundings. The point cloud data is analyzed and processed, and a specific algorithm is used to detect whether there are any abnormalities in the laser signal reflected from the sensor surface, such as uneven signal intensity or abnormal distribution of reflection points, to determine whether the sensor surface is blurry. Specifically, the signal intensity of each point in the point cloud is calculated, and the distribution of the signal intensity is statistically analyzed. If the signal intensity fluctuation range exceeds a preset threshold, the signal intensity is determined to be uneven. Based on this, the distribution density and regularity of the reflection points on the sensor surface are analyzed. If there are local areas with excessively high or low density or irregular distribution, the sensor surface is determined to be blurry.

[0033] 104. If the sensor surface is detected to be blurry, a corresponding cleaning method is selected according to the blur level and type of the sensor;

[0034] If camera surface blur is detected, the camera's blur level is determined based on the high-frequency component energy ratio, the mean gradient amplitude, and the proportion of low-contrast areas. For example, the ratio of the high-frequency energy ratio to a baseline value is divided into three intervals: 60-80% for light blur, 40-60% for moderate blur, and less than 40% for heavy blur. The mean gradient amplitude is graded based on the dry environment baseline value: 50-70% for light, 30-50% for moderate, and less than 30% for heavy blur. The proportion of low-contrast areas is categorized based on image segmentation statistics: 20-35% for light, 35-50% for moderate, and greater than 50% for heavy blur. A weighted fusion algorithm is used to calculate a comprehensive blur index, with weights of 0.4, 0.3, and 0.3, respectively. An index in the range [0.4, 0.6) is considered light blur, [0.6, 0.8) for moderate blur, and ≥0.8 for heavy blur. For special scenarios, such as nighttime or backlighting, the illumination compensation coefficient is dynamically introduced to adaptively adjust the thresholds. For example, each threshold is lowered by 15% at night. At the same time, the spatial distribution pattern of water stain characteristics is combined for final level calibration.

[0035] If the LiDAR surface is detected to be blurry, analysis is performed based on the point cloud data to determine the degree of blur on the LiDAR surface and the corresponding cleaning method is selected based on the blur level and type of LiDAR surface.

[0036] The point cloud area corresponding to the lidar surface is divided into multiple small analysis units, and the average signal strength of the reflection points in each unit is calculated, and the standard deviation of the signal strength is counted. If the standard deviation exceeds the preset range, it indicates that the signal strength of the unit fluctuates greatly. At the same time, the number of reflection points in each unit is counted, the density of reflection points is calculated, and the density distribution is analyzed. If there are units whose density deviates significantly from the overall average level, the spatial statistical analysis method is used to examine the distribution pattern of reflection points in the analysis unit, and determine the areas with irregular distribution, aggregation or discrete anomalies. The degree of blur is divided into three levels: mild, moderate and severe, based on indicators such as signal strength fluctuation, abnormal reflection point density and abnormal distribution pattern.

[0037] After determining the blur level of the camera or lidar, you also need to determine the corresponding type of camera or lidar. The sensor types here are divided into windshield type and non-windshield type. Windshield type sensors are sensors installed inside the windshield, and non-windshield type sensors are sensors installed outside the vehicle.

[0038] For light blur and the type is windshield, select the seventh cleaning method, which is to start the low-speed wiper for a single wipe; for moderate blur and the type is windshield, select the eighth cleaning method, which is to start the medium-speed wiper for two consecutive wipes; for heavy blur and the type is windshield, select the ninth cleaning method, which is to start the high-speed wiper for multiple wipes; for light blur and the type is non-windshield, select the tenth cleaning method, which is to run the fan at low speed; for moderate blur and the type is non-windshield, select the eleventh cleaning method, which is to run the fan at full power; for heavy blur and the type is non-windshield, select the twelfth cleaning method, which is to run the fan in overpressure mode.

[0039] 105. Clean the sensor according to the selected cleaning method, which can be fan cleaning or wiper cleaning.

[0040] A control instruction is generated according to the selected cleaning method, and based on the control instruction, a fan or wiper is controlled to clean the sensor, wherein cleaning the sensor specifically involves cleaning the windshield outer layer of the sensor.

[0041] For example, if the first cleaning mode is selected, a control instruction is generated to turn on the wiper at a low speed and wipe slowly at a preset low-speed frequency; if the second cleaning mode is selected, a control instruction is generated to adjust the wiper to a medium speed and wipe at a preset medium-speed frequency; if the third cleaning mode is selected, a control instruction is generated to adjust the wiper to a high speed and wipe at a high-speed frequency to forcefully scrape off a large amount of rainwater; if the fourth cleaning mode is selected, a control instruction is generated to turn on the fan at a lower air volume to generate a gentle airflow to sweep the sensor surface to remove attached water droplets and a small amount of dust; if the fifth cleaning mode is selected, a control instruction is generated to adjust the fan air volume to a medium level to enhance the airflow's drying and cleaning effect on the sensor; if the sixth cleaning mode is selected If the cleaning mode is selected, a control instruction is generated to adjust the fan to the maximum air volume to form a strong airflow to quickly blow away a large amount of rainwater on the sensor surface; if the seventh cleaning mode is selected, a control instruction is generated to start the low-speed wiper for a single wipe; if the eighth cleaning mode is selected, a control instruction is generated to make the medium-speed wiper wipe twice in a row; if the ninth cleaning mode is selected, a control instruction is generated to make the high-speed wiper wipe multiple times; if the tenth cleaning mode is selected, a control instruction is generated to make the fan run at low speed; if the eleventh cleaning mode is selected, a control instruction is generated to make the fan run at full power; if the twelfth cleaning mode is selected, a control instruction is generated to make the fan run in overpressure mode, so as to ensure that the sensor can be effectively cleaned under different conditions.

[0042] In an embodiment of the present invention, by detecting whether it is raining in the external environment of the autonomous driving vehicle, if it is detected that it is raining, a corresponding cleaning method is selected according to the amount of rainfall and the type of sensor; if it is detected that it is not raining, whether the surface of the sensor is blurred is detected; if it is detected that the surface of the sensor is blurred, a corresponding cleaning method is selected according to the degree of blur and type of the sensor; the sensor is cleaned according to the selected cleaning method, which is fan cleaning or wiper cleaning, thereby effectively reducing the vibration and noise of the vehicle, improving ride comfort, and reducing equipment procurement and maintenance costs.

[0043] See also Figure 2 Another embodiment of the cleaning method of the autonomous driving sensor in the embodiment of the present invention includes:

[0044] 201. Detect whether it is raining in the external environment of the autonomous driving vehicle;

[0045] Image data or point cloud data of the vehicle's external environment is collected through a camera or lidar; the presence of raindrops, water mist or road reflective features is detected based on the image data; if the detected raindrops, water mist or road reflective features are greater than a preset threshold, it is determined that the external environment is raining; based on the point cloud data, the presence of raindrop reflection noise or point cloud density anomalies caused by ambient humidity is detected; if the number of detected noise points exceeds a dynamic threshold or the point cloud scattering characteristics meet the rainy day model, it is determined that the external environment is raining.

[0046] When the sensor is a camera, the camera collects image data of the vehicle's external environment and determines whether it is raining. Specifically, a deep learning-based target detection algorithm can be used to extract and classify features of raindrops, water mist, and road surface reflective areas in the image. By calculating the pixel ratio and grayscale gradient change value of the feature area, the raindrop feature threshold is set to meet the requirements that the number of circular bright spots detected in the image exceeds the preset number threshold and the average grayscale value is lower than the preset grayscale threshold. The water mist feature threshold requires that the ratio of pixels with regional transparency lower than the preset transparency threshold exceeds the preset ratio threshold. The road surface reflective feature threshold must also meet the requirements that the ratio of the highlight area exceeds the preset area threshold and the brightness standard deviation is greater than the preset discreteness threshold. When any feature detection result exceeds its corresponding judgment threshold, it is determined that the current environment is in a rainy state.

[0047] When the sensor is a lidar, point cloud data of the vehicle's external environment is collected through the lidar, and the point cloud data is preprocessed to filter out static points of the vehicle itself and fixed obstacles. The number of high-reflectivity discrete noise points in the remaining point cloud is counted. These noise points are usually formed by laser reflection from raindrops. At the same time, the density distribution characteristics of the point cloud in a specific distance range are analyzed. In a rainy environment, the density of medium and long-distance point clouds will be significantly reduced and unevenly distributed due to suspended water droplets in the air. The number of noise points in the current frame is compared with the threshold dynamically adjusted according to the vehicle speed and ambient brightness. At the same time, the point cloud density characteristics are compared with the preset rainy day model. If the number of noise points exceeds the preset noise threshold or the point cloud density characteristics meet the rainy day model, it is comprehensively determined that the external environment is raining.

[0048] 202. If it is detected that the external environment is raining, a corresponding cleaning method is selected according to the amount of rainfall and the type of sensor;

[0049] The method for determining rainfall amount is as follows: when determining rain using image data collected by a camera, the number of raindrops per unit area of ​​the image is counted. If the number of raindrops is not greater than a first preset number threshold, it is determined to be light rain; if the number of raindrops is greater than the first preset number threshold and not greater than a second preset number threshold, it is determined to be moderate rain; if the number of raindrops is greater than the second preset number threshold, it is determined to be heavy rain. When determining rain using point cloud data collected by a lidar, the number of raindrop reflection noise points per unit volume of the point cloud is counted. If the number of noise points is not greater than a first preset noise threshold, it is determined to be light rain; if the number of noise points is greater than the first preset noise threshold and not greater than a second preset noise threshold, it is determined to be moderate rain; if the number of noise points is greater than the second preset noise threshold, it is determined to be heavy rain.

[0050] Sensor types are divided into windshield type and non-windshield type. The execution steps of selecting the corresponding cleaning method according to the amount of rainfall and sensor type are similar to the corresponding execution process in the above step 102 and will not be repeated here.

[0051] 203. If it is detected that the external environment is not raining, then whether the sensor surface is blurred is detected;

[0052] Image data or point cloud data is collected through a sensor, where the sensor is a camera or a lidar; a first fuzzy feature in the image data or a second fuzzy feature in the point cloud data is analyzed; and whether the corresponding sensor surface is blurred is determined based on the first fuzzy feature or the second fuzzy feature.

[0053] An image clarity evaluation algorithm based on gradient amplitude is used to calculate the horizontal and vertical gradient values ​​of each pixel in the image, and the distribution of the overall gradient amplitude is statistically analyzed. The clarity threshold is set to when the average gradient amplitude is less than 10 and the standard deviation of the gradient amplitude is greater than 5. The image is judged to be blurred. The distribution characteristics of the blurred area are further analyzed. If the blurred area is mainly concentrated at the edge of the image and shows a circular diffusion trend, and combined with historical data comparison, it is found that the blurred feature is highly matched with the water stains or dust pollution features on the camera surface, then the camera surface is judged to be blurred.

[0054] A spatial density clustering algorithm is used to analyze the distribution characteristics of effective reflection points on the target's reflective surface in the point cloud. The number of reflection points per unit area is calculated and their standard deviation is statistically analyzed. A fuzzy threshold is set such that when the reflection point density falls below a preset 80 points per square meter and the density standard deviation exceeds 15, a preliminary fuzzy anomaly is determined. Further analysis of reflection intensity is conducted. If the reflection intensity within the fuzzy area is generally lower than the surrounding area by more than 20% and exhibits non-uniform attenuation, and if, when compared to the device's built-in self-test parameters, this characteristic is consistent with typical manifestations of an obstructed LiDAR window or surface contamination, the LiDAR surface is ultimately confirmed to be fuzzy.

[0055] 204. If the sensor surface is detected to be blurry, a corresponding cleaning method is selected according to the blur level and type of the sensor;

[0056] If blur is detected on the sensor surface, the image clarity index is calculated based on the image data, or the point cloud density attenuation rate is calculated based on the point cloud data. The blur level of the corresponding sensor is determined based on the image clarity index or the point cloud density attenuation rate, and the blur level includes light blur, moderate blur, or heavy blur. The corresponding cleaning method is selected according to the blur level and type of the sensor.

[0057] If camera surface blur is detected, an image clarity index is calculated based on the image data, and the degree of camera blur is determined based on the image clarity index. Specifically, the image clarity index is calculated based on the image data by using the Laplace operator to extract the second-order derivative of each pixel in the image, counting the gradient amplitudes of all pixels and calculating their average and standard deviation. The average gradient amplitude is compared with a preset average threshold, and whether the standard deviation is greater than a preset standard deviation threshold is used as an auxiliary judgment condition. When the average gradient amplitude is less than the preset average threshold and the standard deviation is greater than the preset standard deviation threshold, the area ratio of the blurred area is further analyzed. For example, if the blurred area ratio is less than 20%, it is determined to be lightly blurred, if it is between 20% and 40%, it is determined to be moderately blurred, and if it is greater than 40%, it is determined to be heavily blurred.

[0058] If the LiDAR surface is detected as blurry, the point cloud density decay rate is calculated based on the point cloud data, and the degree of blur of the corresponding sensor is determined based on the point cloud density decay rate. Specifically, the point cloud data is divided into grids, the number of valid reflection points in each grid is counted, the difference in reflection point density between the blurred area and the adjacent normal area is calculated, and the ratio of this difference to the initial density of the blurred area is defined as the point cloud density decay rate. The degree of blur is determined based on the point cloud density decay rate. For example, when the point cloud density decay rate is less than 20%, it is judged as mild blur, when the point cloud density decay rate is between 20% and 50%, it is judged as moderate blur, and when the point cloud density decay rate is greater than 50%, it is judged as severe blur.

[0059] After determining the blur level of the camera or lidar, the corresponding type of camera or lidar must also be determined. Sensor types are categorized as windshield or non-windshield. The steps for selecting the appropriate cleaning method based on blur level and type are similar to those in step 104 above and will not be further detailed here.

[0060] 205. Sending a cleaning instruction to the selected fan or wiper according to the selected cleaning method to start the cleaning operation;

[0061] The execution process of step 205 is similar to that of the above step 105 and will not be repeated here.

[0062] 206. Perform fault diagnosis on the cleaning process and cleaning results.

[0063] Monitor the working status of the fan or wiper. If an abnormal working status of the fan or wiper is detected during the cleaning process, immediately report the corresponding fan or wiper fault and stop the cleaning operation. After cleaning is completed, check again whether the sensor surface is blurred. If the sensor surface is still blurred, the corresponding sensor cleaning is invalid.

[0064] like Figure 3 As shown, Figure 3 This is a system fault diagnosis flowchart. According to the selected cleaning method, a cleaning instruction is sent to the fan or wiper to start the cleaning operation. At the same time, the working status of the fan or wiper is monitored in real time during the cleaning process. Once an abnormality is found, the fault is reported immediately and cleaning is stopped. After cleaning is completed, the sensor surface will be checked again to see if it is blurred. If it is still blurred, it is determined that the cleaning of the corresponding sensor is invalid.

[0065] In an embodiment of the present invention, by detecting whether it is raining in the external environment of the autonomous driving vehicle, if it is detected that it is raining in the external environment, a corresponding cleaning method is selected according to the amount of rainfall and the type of sensor; if it is detected that it is not raining in the external environment, whether the sensor surface is blurred is detected; if it is detected that the sensor surface is blurred, a corresponding cleaning method is selected according to the degree of blur and type of the sensor; according to the selected cleaning method, a cleaning instruction is sent to the selected fan or wiper, the cleaning operation is started, and fault diagnosis is performed on the cleaning process and cleaning results, thereby effectively reducing the vibration and noise of the vehicle, improving ride comfort, and reducing equipment procurement and maintenance costs.

[0066] The above describes the cleaning method of the autonomous driving sensor in the embodiment of the present invention. The following describes the cleaning system of the autonomous driving sensor in the embodiment of the present invention. Figure 4 In one embodiment of the present invention, a cleaning system for an autonomous driving sensor includes:

[0067] The sensor module is used to obtain external environmental data and blur detection data. The autonomous driving controller is used to receive the detection data from the sensor module, select the corresponding cleaning method based on the amount of rain or the blur level of the corresponding sensor, and generate the corresponding cleaning control instructions. The control IO module is used to transmit the cleaning control instructions from the autonomous driving controller to the fan or wiper. The fan is used to activate the fan according to the cleaning control instructions to blow and wash the sensor surface. The wiper is used to activate the wiper according to the control instructions to clean the sensor surface. The sensor module includes a camera and a lidar.

[0068] Furthermore, the system also includes a diagnostic module for monitoring the working status of sensors, fans and wipers in real time, and feeding back status information to the automatic driving controller; when an abnormal state is detected, a fault report is generated and sent to the automatic driving controller for fault handling; a power supply module for providing a stable power supply for the automatic driving controller, sensor module, control IO module, fan and wiper to ensure the normal operation of the system; and connecting components such as signal lines, power lines and pipes to ensure smooth signal transmission, power supply and fluid transmission between modules, thereby ensuring the coordinated operation of the entire cleaning system.

[0069] above Figure 4The cleaning of the autonomous driving sensor in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The electronic device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0070] See also Figure 5 As shown, the electronic device includes a processor 500 and a memory 501, wherein the memory 501 stores machine executable instructions that can be executed by the processor 500, and the processor 500 executes the machine executable instructions to implement the above-mentioned cleaning method of the autonomous driving sensor.

[0071] Furthermore, Figure 5 The electronic device shown further includes a bus 502 and a communication interface 503 , and the processor 500 , the communication interface 503 and the memory 501 are connected via the bus 502 .

[0072] Among them, the memory 501 may include a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), for example, at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 503 (which can be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. can be used. The bus 502 can be an ISA bus, a PCI bus, or an EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0073] The processor 500 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits or software instructions in the processor 500. The processor 500 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may 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. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present disclosure may be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory 501 , and the processor 500 reads the information in the memory 501 and completes the method steps of the aforementioned embodiment in combination with its hardware.

[0074] The present invention also provides an electronic device, wherein the computer device includes a memory and a processor, wherein the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the cleaning method of the autonomous driving sensor in the above-mentioned embodiments.

[0075] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to execute the steps of the cleaning method for the autonomous driving sensor.

[0076] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0077] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program code.

[0078] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for cleaning an autonomous driving sensor, characterized in that: The cleaning method of the autonomous driving sensor includes: Detect whether it is raining outside the autonomous vehicle; If it is detected that the external environment is raining, the corresponding cleaning method is selected according to the amount of rain and the type of sensor; If it is detected that the external environment is not raining, then the sensor surface is checked to see if it is blurred; If the sensor surface is detected to be blurry, a corresponding cleaning method is selected according to the blur degree and type of the sensor; The sensor is cleaned according to a selected cleaning method, which is fan cleaning or wiper cleaning.

2. The cleaning method of the autonomous driving sensor according to claim 1, characterized in that: The detecting whether the external environment of the autonomous driving vehicle is raining includes: Collect image data or point cloud data of the vehicle's external environment through cameras or lidar; detecting whether there are raindrops, water mist, or road surface reflective features based on the image data; if the detected raindrops, water mist, or road surface reflective features are greater than a preset threshold, determining that the external environment is raining; Based on the point cloud data, it is detected whether there is raindrop reflection noise or point cloud density anomaly caused by environmental humidity; if the number of detected noise points exceeds a dynamic threshold or the point cloud scattering characteristics conform to the rainy day model, it is determined that the external environment is raining.

3. The cleaning method of the autonomous driving sensor according to claim 1, characterized in that: The method for determining the amount of rainfall includes: When determining rain using image data collected by a camera, the number of raindrops per unit area of ​​the image is counted. If the number of raindrops is not greater than a first preset number threshold, it is determined to be light rain; if the number of raindrops is greater than the first preset number threshold and not greater than a second preset number threshold, it is determined to be moderate rain; if the number of raindrops is greater than the second preset number threshold, it is determined to be heavy rain; When using point cloud data collected by lidar to determine rain, the number of raindrop reflection noise points in the unit volume point cloud is counted. If the number of noise points is not greater than the first preset noise threshold, it is determined to be light rain; if the number of noise points is greater than the first preset noise threshold and not greater than the second preset noise threshold, it is determined to be moderate rain; if the number of noise points is greater than the second preset noise threshold, it is determined to be heavy rain.

4. The cleaning method of the autonomous driving sensor according to claim 1, characterized in that: The detecting whether the sensor surface is blurred includes: Collect image data or point cloud data through a sensor, wherein the sensor is a camera or a lidar; analyzing a first fuzzy feature in the image data or a second fuzzy feature in the point cloud data; Whether the corresponding sensor surface is blurred is determined according to the first blur feature or the second blur feature.

5. The cleaning method of the autonomous driving sensor according to claim 4, characterized in that: If the sensor surface is detected to be blurry, a corresponding cleaning method is selected according to the blur level and type of the sensor, including: If blur is detected on the sensor surface, an image clarity index is calculated based on the image data, or a point cloud density attenuation rate is calculated based on the point cloud data; Determining a blur level of the sensor based on the image clarity index or the point cloud density decay rate, where the blur level includes light blur, medium blur, or heavy blur; Select a corresponding cleaning method according to the blur level and type of the sensor.

6. The method for cleaning an autonomous driving sensor according to any one of claims 1 to 5, characterized in that: The cleaning process of the sensor according to the selected cleaning method includes: Sending a cleaning instruction to the selected fan or wiper according to the selected cleaning method to start the cleaning operation; Troubleshoot the cleaning process and results.

7. The cleaning method of the autonomous driving sensor according to claim 6, characterized in that: The fault diagnosis of the cleaning process and cleaning results includes: Monitor the working status of the fan or wiper. If an abnormality is detected during the cleaning process, a fault report will be generated immediately and the cleaning operation will be stopped. After cleaning, check the sensor surface again to see if it is blurry; If the sensor surface is still blurry, it is determined that the corresponding sensor cleaning is invalid.

8. A cleaning system for an autonomous driving sensor, characterized in that: The cleaning system for the autonomous driving sensor should be the cleaning method for the autonomous driving sensor according to any one of claims 1 to 7, comprising: Sensor module, used to obtain external environment data and fuzzy detection data; An automatic driving controller is configured to receive detection data from the sensor module, select a corresponding cleaning method based on the amount of rain or the blur level of the corresponding sensor, and the type of sensor, and generate a corresponding cleaning control instruction; A control IO module, configured to transmit cleaning control instructions from the automatic driving controller to the fan or wiper; A fan, configured to start the fan according to the cleaning control instruction to blow clean the sensor surface; The wiper is used to start the wiper according to the control instruction to clean the sensor surface.

9. An electronic device, characterized in that: The electronic device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory so that the electronic device executes the cleaning method for the autonomous driving sensor as described in any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the cleaning method of the autonomous driving sensor according to any one of claims 1 to 7 is implemented.

Images