Vehicle cleaning method and device, electronic equipment, storage medium and program product

Abstract

The embodiment of the invention provides a vehicle cleaning method and device, electronic equipment, a storage medium and a program product. The method comprises the following steps: carrying out stain identification on a vehicle body picture by adopting a pre-trained identification model to obtain a vehicle body picture marked with a stain identifier; based on the vehicle body picture marked with the stain identifier, constructing a vehicle body point cloud marked with the stain identifier; determining a cleaning track of the cleaning robot based on the vehicle body point cloud and the category information of the stains, and controlling the cleaning robot to clean the target vehicle according to the cleaning track; and in response to an evaluation result of the user for the cleaned target vehicle, optimizing the recognition model based on the evaluation result, and performing stain recognition on the vehicle body pictures of the other vehicles by adopting the optimized recognition model. The method is used for dynamically adjusting the cleaning mode according to different types of stains, and the automatic cleaning effect of the vehicle is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle cleaning control technology, and in particular to a vehicle cleaning method, apparatus, electronic device, storage medium and program product. Background Technology

[0002] With the rapid development of automated vehicle cleaning technology, society has increasingly higher requirements for the accuracy of stain recognition during the automated vehicle cleaning process.

[0003] Currently, common automated vehicle cleaning methods generally use pre-trained recognition models to identify stains on the vehicle body. However, the diversity of training samples for these recognition models is limited, which severely restricts the accuracy of stain recognition. In real-world scenarios, the types of stains on vehicle bodies are very diverse. Using fixed recognition models for stain recognition can easily lead to missed or incorrect identifications, resulting in inadequate automated vehicle cleaning, severely impacting the cleaning effect and damaging the user experience. Summary of the Invention

[0004] This application provides a vehicle cleaning method, apparatus, electronic device, storage medium, and program product to improve the accuracy of vehicle body stain recognition, dynamically adjust cleaning strategies, and enhance vehicle cleaning effectiveness.

[0005] In a first aspect, embodiments of this application provide a vehicle cleaning method applied to a vehicle cleaning system, the vehicle cleaning system including a cleaning robot and a data acquisition device; the cleaning robot is used to clean a target vehicle parked in the target area, and the data acquisition device is used to acquire images of the target vehicle's body;

[0006] The method includes:

[0007] A pre-trained recognition model is used to identify stains in the vehicle body image, resulting in a vehicle body image marked with stain identifiers. The stain identifiers in the vehicle body image are used to mark the image location corresponding to the stains on the target vehicle, and the stain identifiers contain the category information of the stains.

[0008] Based on the vehicle body image marked with the stain identifier, a vehicle body point cloud marked with the stain identifier is constructed; the stain identifier in the vehicle body point cloud is used to mark the spatial location of the stain on the target vehicle.

[0009] Based on the vehicle body point cloud and the category information of the stains, the cleaning trajectory of the cleaning robot is determined, and the cleaning robot is controlled to clean the target vehicle according to the cleaning trajectory;

[0010] In response to the user's evaluation of the target vehicle after cleaning, the recognition model is optimized based on the evaluation results, and the optimized recognition model is used to identify stains on the body images of the remaining vehicles.

[0011] In one possible implementation, the evaluation results include unwashed markings made by the user on the vehicle image, the unwashed markings being used to mark the image locations corresponding to stains on the target vehicle after cleaning;

[0012] The optimization of the recognition model based on the evaluation results includes:

[0013] The union of the vehicle body image marked with the stain label and the vehicle body image marked with the unclean label is obtained to obtain the vehicle body image marked with both the stain label and the unclean label.

[0014] The vehicle images marked with both the stain label and the unclean label are added to the training set of the recognition model, and the recognition model is retrained using the training set to optimize the recognition model.

[0015] In one possible implementation, determining the cleaning trajectory of the cleaning robot based on the vehicle body point cloud and the category information of the stains includes:

[0016] The three-dimensional points in the vehicle body point cloud are clustered and divided into point cloud regions corresponding to multiple vehicle body parts.

[0017] Based on the spatial location and category information of the stains within the corresponding point cloud area of ​​each of the aforementioned vehicle body components, the sub-cleaning trajectory corresponding to each of the aforementioned point cloud areas is determined.

[0018] Based on the preset cleaning sequence of each of the vehicle body parts, the sorted sub-cleaning trajectories are smoothed to obtain the cleaning trajectory.

[0019] In one possible implementation, determining the sub-cleaning trajectory corresponding to each point cloud region based on the spatial location and category information of the stains within the corresponding point cloud region of each of the vehicle body components includes:

[0020] A preset vehicle body component cleaning strategy is adopted to determine the first sub-cleaning trajectory of the point cloud region corresponding to each vehicle body component; the vehicle body component cleaning strategy includes a first cleaning method corresponding to multiple vehicle body components;

[0021] Using a preset stain type cleaning strategy, the second cleaning trajectory corresponding to stains of different categories is determined in the point cloud area corresponding to each of the vehicle body parts; the stain type cleaning strategy includes a second cleaning method corresponding to stains of multiple categories.

[0022] By fusing the first sub-cleaning trajectory and the second cleaning trajectory, sub-cleaning trajectories corresponding to each point cloud region are obtained.

[0023] In one possible implementation, the method further includes:

[0024] Based on the assessment results, the cleaning strategy for the stain type is adjusted.

[0025] In one possible implementation, the second cleaning method is provided with multiple cleaning levels, with different cleaning levels indicating different stain cleaning intensities;

[0026] The step of adjusting the cleaning strategy based on the evaluation results includes:

[0027] Find the intersection of the vehicle body image marked with the stain label and the vehicle body image marked with the unclean label to obtain the adjustment area marked with both the stain label and the unclean label.

[0028] Based on the stain identifier marked in the adjustment area, the category information of the stain to be adjusted is determined;

[0029] In the stain type cleaning strategy, the cleaning level corresponding to the category information of the stain to be adjusted is increased.

[0030] Secondly, this application provides a vehicle washing device for use in a vehicle washing system, the vehicle washing system including a cleaning robot and a data acquisition device; the cleaning robot is used to clean target vehicles parked in the target area, and the data acquisition device is used to acquire images of the target vehicle body;

[0031] The device includes:

[0032] The recognition module is used to identify stains on the vehicle body image using a pre-trained recognition model, resulting in a vehicle body image marked with stain identifiers; the stain identifiers in the vehicle body image are used to mark the image location corresponding to the stains on the target vehicle, and the stain identifiers contain the category information of the stains;

[0033] A construction module is used to construct a vehicle point cloud marked with the stain identifier based on a vehicle image marked with the stain identifier; the stain identifier in the vehicle point cloud is used to mark the spatial location of the stain on the target vehicle;

[0034] The control module is used to determine the cleaning trajectory of the cleaning robot based on the vehicle body point cloud and the category information of the stains, and control the cleaning robot to clean the target vehicle according to the cleaning trajectory;

[0035] An optimization module is used to respond to the user's evaluation results for the cleaned target vehicle, optimize the recognition model based on the evaluation results, and use the optimized recognition model to identify stains on the body images of other vehicles.

[0036] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0037] The memory stores computer-executed instructions;

[0038] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0039] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0040] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0041] The vehicle cleaning method, apparatus, electronic device, storage medium, and program products provided in this application achieve accurate identification and customized cleaning of stains on the vehicle body by relying on a pre-trained recognition model and three-dimensional vehicle body point cloud mapping. By incorporating user evaluation results, the recognition model is continuously and dynamically optimized, avoiding the situation where the recognition model becomes rigid and unable to adapt to changing scenarios. This allows the continuously updated recognition model to identify as many stains as possible on the vehicle body and accurately identify the type of each stain, thereby improving the stain recognition accuracy. This enables subsequent cleaning robots to perform more targeted cleaning of the vehicle based on more accurate stain recognition results, thereby improving the cleaning effect of automated cleaning. Attached Figure Description

[0042] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0043] Figure 1 This application provides a structural schematic diagram of a vehicle washing system;

[0044] Figure 2 A flowchart illustrating the vehicle cleaning method provided in this application;

[0045] Figure 3A schematic diagram of the vehicle washing device provided in this application;

[0046] Figure 4 A schematic diagram of the structure of the electronic device provided in this application.

[0047] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0048] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0049] like Figure 1 As shown in the figure, this application provides a vehicle cleaning system 100, wherein the vehicle cleaning system 100 includes a data collection device 110, a controller 120 and a cleaning robot 130, and the controller 120 is connected to the data collection device 110 and the cleaning robot 130 respectively.

[0050] The cleaning robot 130 is connected to at least one receiving tank via a pipe;

[0051] When the controller 120 receives a vehicle cleaning instruction from the server of the car wash service provider, it generates a collection instruction and sends it to the collection device 110 to control the collection device 110 to collect images of the target vehicle parked in the target area.

[0052] In one embodiment, the vehicle washing system 100 is equipped with acquisition devices 110 at multiple locations to acquire images of the target vehicle parked within the target area from multiple angles. Alternatively, the acquisition devices 110 can be mounted on a movable base or a sliding track. After receiving an acquisition command, the acquisition devices 110 can move along a preset path or along the sliding track to acquire images of the target vehicle parked within the target area.

[0053] In one embodiment, the cleaning robot 130 can also be set on another movable base or sliding track. The cleaning robot 130 can move along another preset path or along the sliding track to clean the target vehicle parked in the target area according to the cleaning trajectory.

[0054] It should be noted that the aforementioned sliding track can be set on the ground surrounding the target area, on the ceiling of the corresponding target area, or on the side wall of the corresponding target area.

[0055] In one embodiment, the data acquisition device 110 can be directly installed at the end of the cleaning robot 130. When the controller 120 receives the vehicle cleaning instruction from the server of the car wash service provider, it can generate a data acquisition instruction and send it to the data acquisition device 110, and at the same time generate a movement instruction and send it to the cleaning robot 130 to control the cleaning robot 130 to move along a preset path or along a sliding track, thereby driving the data acquisition device 110 to collect images of the vehicle body around the target vehicle.

[0056] As an example, the acquisition device 110 can be a camera, a spectrometer, or a multispectral camera—devices capable of identifying spectral features. When the acquisition device 110 is a spectrometer or a multispectral camera, the controller 120 can obtain the spectral information of each pixel from the vehicle image acquired by the acquisition device 110.

[0057] In one embodiment, a vehicle cleaning method is provided, which is applied to a controller in the aforementioned vehicle cleaning system. For example... Figure 2 As shown, the vehicle cleaning method includes:

[0058] Step 202: Using a pre-trained recognition model, perform stain recognition on the vehicle body image to obtain a vehicle body image marked with stain labels; the stain labels in the vehicle body image are used to mark the image location corresponding to the stains on the target vehicle, and the stain labels contain stain category information.

[0059] The vehicle images can be obtained through a data acquisition device in the vehicle washing system.

[0060] As an example, the acquisition device can acquire video of the vehicle body of a target vehicle parked in the target area. The controller can perform blur filtering on each frame of the vehicle body image contained in the video to remove the blurrier vehicle body images and retain the remaining clear vehicle body images.

[0061] In this embodiment, the controller can use a trained recognition model to mark the pixels on the vehicle image that indicate stains with stains, thereby indicating that the current pixel is a stain and achieving pixel-level accurate segmentation.

[0062] In one embodiment, the controller can further identify the pixels corresponding to the stain using a trained recognition model to determine the stain type and other category information corresponding to different pixels. The stain type is used to indicate whether the stain corresponding to the current pixel is dirt, oil, water, bird droppings, tree sap, etc.

[0063] Category information is used to indicate the specific type of stains on the vehicle body. The stain types included in the category information can indicate mud, oil stains, water stains, bird droppings, tree sap, etc.

[0064] Stain indicators are used to indicate graphics or symbols that are visually displayed on the vehicle body image. The controller can generate stain indicators for each pixel based on the category information of each pixel, and overlay or superimpose the generated stain indicators onto the corresponding pixels in the vehicle body image.

[0065] Step 204: Based on the vehicle body image marked with stain labels, construct a vehicle body point cloud marked with stain labels; the stain labels in the vehicle body point cloud are used to mark the spatial location of the stains on the target vehicle.

[0066] As an example, vehicle body point clouds can be obtained by processing vehicle body images acquired by a data acquisition device.

[0067] The controller, upon receiving a vehicle cleaning instruction from the server of the car wash service provider, issues a collection instruction and sends it to the collection device to control the collection device to collect multiple images of the target vehicle parked in the target area.

[0068] Furthermore, the controller can construct a point cloud of the target vehicle body based on multiple vehicle body images. This process may include, for example, acquiring the pose information of each vehicle body image acquired by the acquisition device; and performing 3D mapping on the image feature points contained in each vehicle body image based on the pose information to obtain the spatial feature points corresponding to the image feature points in each vehicle body image, thus forming a vehicle body point cloud.

[0069] Image feature points typically appear at corners and areas of dramatic texture change in vehicle images, containing rich local information. An image feature point is not merely a single point in a vehicle image; it includes a series of local information, sometimes even a small region of area. For example, image feature points correspond to corner points, edges, and textures in a vehicle image, and can also correspond to edge points of different body parts on the target vehicle.

[0070] As an example, methods for extracting image feature points from vehicle images may include: using one of the following algorithms to extract image feature points contained in the vehicle image: Oriented FAST and Rotated BRIEF (ORB), Scale-Invariant Feature Transform (SIFT), or Speeded Up Robust Features (SURF).

[0071] Alternatively, methods for extracting image feature points from vehicle images may include: using a deep learning-based SuperPoint feature extraction algorithm or a Local Feature Matching Transformer (LoFTR) to extract image feature points contained in the vehicle image.

[0072] It should be noted that this application does not limit the method for extracting image feature points contained in a vehicle body image, as long as it can extract the local invariant features of the vehicle body image. Different methods for extracting image feature points contained in a vehicle body image should all be included within the protection scope of this application.

[0073] Furthermore, the process of the terminal matching the spatial feature points corresponding to each image feature point can usually be achieved by calculating the distance between descriptors (such as Euclidean distance, Hamming distance, etc.) to find the correspondence between image feature points and spatial feature points. Commonly used matching methods include brute-force matching, K-nearest neighbor matching (KNN), and the FastLibrary for Approximate Nearest Neighbors (FLANN). Moreover, since erroneous matches may occur during the matching process, various filtering methods (such as ratio testing, RANSAC, etc.) are needed to remove erroneous matches and ensure the quality of the matching.

[0074] It should be noted that in the process of matching the spatial feature points corresponding to each image feature point, it is necessary to obtain the spatial coordinate information of the spatial feature points corresponding to the image feature points. This process can be achieved by the Perspective-n-Point (pnp) algorithm.

[0075] This application does not limit the method for obtaining the spatial coordinate information of spatial feature points corresponding to image feature points in vehicle images, as long as it can realize the mapping between two-dimensional points in the image and three-dimensional points in the scene. Different methods for mapping between two-dimensional points in the image and three-dimensional points in the scene should be included within the protection scope of this application.

[0076] It should be noted that after matching the spatial feature points corresponding to each image feature point, the controller can also remove sparse spatial feature points that are significantly off-target.

[0077] It should be noted that the vehicle washing command refers to the instruction that controls the cleaning robot in the vehicle washing system to wash the vehicle body parked in the target area.

[0078] The vehicle cleaning command can be issued by the user through the terminal's human-machine interface. Specifically, the terminal's human-machine interface can display a platform interface pre-designated by the service provider for vehicle cleaning. The user can issue the vehicle cleaning command by clicking on specific components within the platform interface. It should be noted that the user can also adjust the cleaning robot's cleaning time and the pressure of the cleaning liquid sprayed onto the vehicle by clicking on pre-integrated virtual components on the platform interface that indicate different cleaning durations and intensities. The aforementioned terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc., while portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Additionally, the terminal can also be a human-machine interface device located near the target area and wiredly connected to the controller, such as a touchscreen mounted on a wall or a touchscreen mounted on a separate pillar. This embodiment does not limit the form of the terminal for issuing vehicle cleaning commands; any terminal capable of wired or wireless communication connection to the controller in the vehicle cleaning system should be included within the scope of protection of this application.

[0079] Step 206: Based on the point cloud of the vehicle body and the category information of the stains, determine the cleaning trajectory of the cleaning robot, and control the cleaning robot to clean the target vehicle according to the cleaning trajectory.

[0080] In this embodiment, the controller may, for example, pre-store a first basic cleaning trajectory, which is used to identify a cleaning path that covers the entire body of the vehicle from top to bottom and from front to back.

[0081] It should be noted that the first basic cleaning trajectory corresponds to a specific vehicle model, and the controller needs to correct the first basic cleaning trajectory according to the actual shape of the target vehicle parked in the target area.

[0082] The controller can make preliminary adjustments to the actual spatial position of the first basic cleaning trajectory based on the actual point cloud of the target vehicle body, so that the end effector of the cleaning robot can clean along the body of the target vehicle without colliding with it. Secondly, the controller can dynamically adjust the cleaning method of the end effector of the cleaning robot at the corresponding pixel of the stain based on the category information of the stain on the target vehicle body, and finally obtain the cleaning trajectory of the cleaning robot, so that when the end effector of the cleaning robot cleans the pixel corresponding to the stain along the first basic cleaning trajectory after preliminary adjustment, it can perform targeted cleaning of the stain.

[0083] For example, when the category information indicates that the stain is bird droppings, the controller can control the end of the cleaning robot to apply greater pressure and use a spiral progressive movement to wipe the corresponding pixel. When the category information indicates that the stain is oil, the controller can control the end of the cleaning robot to apply medium pressure and use a high-frequency vibration reciprocating movement to wipe the corresponding pixel.

[0084] Alternatively, the controller can generate a specific second basic motion trajectory for the target vehicle's body point cloud. Specifically, the target vehicle's body point cloud defines the precise geometry and spatial range of the target vehicle's body parts or structural surfaces. For example, the controller can determine multiple stopping points along the body point cloud at preset intervals, and further smooth and filter the sequence of determined stopping points to finally obtain the second basic motion trajectory of the cleaning robot's end effector for the target vehicle's body point cloud.

[0085] Furthermore, the controller can dynamically adjust the cleaning method of the cleaning robot's end effector at the corresponding pixel of the stain based on the category information of the stain on the target vehicle body, and finally obtain the cleaning trajectory of the cleaning robot, so that when the end effector of the cleaning robot cleans the pixel corresponding to the stain along the second basic cleaning trajectory, it can perform targeted cleaning of the stain.

[0086] It should be noted that the first and second basic movement trajectories mentioned above not only indicate the movement path of the cleaning robot's end effector, but also include the rotation angles of each joint of the cleaning robot.

[0087] Step 208: In response to the user's evaluation of the cleaned target vehicle, optimize the recognition model based on the evaluation results, and use the optimized recognition model to identify stains on the body images of the remaining vehicles.

[0088] The evaluation result refers to the feedback information from the user on the cleaning robot's cleaning effect on the target vehicle after the vehicle cleaning is completed. This includes specific evaluations such as whether the cleaning is satisfactory or unsatisfactory. Specifically, the user can mark the unsatisfactory cleaning location on any vehicle image and send the marked image to the controller along with the evaluation result.

[0089] As an example, the controller can use a data acquisition device to capture images of the target vehicle body after cleaning, and display the cleaned images and images marked with stain labels to the user. The user can combine the images marked with stain labels with the actual cleaning effect of the target vehicle body to mark the unsatisfactory cleaning locations on the cleaned images. The markings can also include the reasons for incomplete cleaning, such as unidentified stains (and the type of stain), misjudgment of stain type, insufficient scrubbing force, etc.

[0090] In this embodiment, the controller can adjust the parameters and supplement the data of the pre-trained recognition model based on the user's evaluation results. For example, if there are unrecognized stains, the vehicle images with the user-labeled unrecognized stains and their category information are added to the training dataset; if there are misclassified stain types, the vehicle images with the user-labeled correct stain category information are added to the training dataset to adjust the feature extraction parameters and classifier parameters of the recognition model, thereby improving the subsequent recognition accuracy of the recognition model.

[0091] The aforementioned vehicle cleaning method, relying on a pre-trained recognition model and 3D vehicle body point cloud mapping, achieves accurate identification and customized cleaning of stains on the vehicle body. By incorporating user evaluation results, it enables continuous dynamic optimization of the recognition model, avoiding the situation where the recognition model becomes rigid and unable to adapt to changing scenarios. This allows the continuously updated recognition model to identify as many stains as possible on the vehicle body and accurately identify the type of each stain, thereby improving the stain recognition accuracy. This enables subsequent cleaning robots to perform more targeted cleaning based on more accurate stain recognition results, thus improving the cleaning effect of automated cleaning.

[0092] In some optional embodiments, the evaluation results include unwashed markings made by the user on images of the vehicle body, which are used to mark the image locations corresponding to stains on the target vehicle after cleaning;

[0093] Step 208 includes:

[0094] Find the union of the images of the car body marked with the stain label and the images of the car body marked with the unclean label, and you will get the images of the car body marked with both the stain label and the unclean label.

[0095] Images of vehicles marked with both dirt and unclean markings are added to the training set of the recognition model, and the recognition model is retrained using the training set to optimize it.

[0096] In this embodiment, after the target vehicle is cleaned, the controller can send another collection command to the collection device to control the collection device to collect images of the cleaned vehicle body and display them on the user terminal's feedback interface, while also providing a marking tool.

[0097] Users can observe the actual vehicle or compare images of the cleaned vehicle with images marked with stain labels. If they find any remaining stains on the vehicle, they can use a marker to select the specific location of the stain on the cleaned image and add information about the stain's type. After marking, the user submits feedback, and the controller automatically associates and stores the images of the vehicle marked with the "not clean" label with the corresponding images marked with the "stain" label. The cleaned images have the same viewing angle and resolution as the images taken before cleaning, allowing for easy comparison.

[0098] Furthermore, the controller can perform a union operation on two associated stored vehicle images (the image of the vehicle marked as "not clean" and the corresponding image of the vehicle marked as "stain"). Specifically, the controller can use image alignment technology to ensure that the pixel coordinates of the two vehicle images correspond exactly, and then overlay the coordinate information and category information of the "stain" and "not clean" labels onto the same vehicle image.

[0099] If a pixel has both a stain marker and an unclean marker, it means that the recognition model has identified the stain, but the cleaning robot has not cleaned it properly. The controller can mark the pixel with a first-type marker to indicate that the recognition model has correctly identified the stain, but the cleaning robot has not cleaned it thoroughly.

[0100] If a pixel only has an "unwashed" mark, it means that the recognition model did not recognize the stain and the cleaning robot did not clean it properly. The controller can mark the pixel with a second type of mark to indicate that the recognition model missed the stain.

[0101] If a pixel only has a stain marker, it means that the recognition model has identified the stain and it has been cleaned by the cleaning robot. The controller does not need to mark that pixel.

[0102] Furthermore, the controller can supplement the training dataset of the recognition model with vehicle images labeled with the first and second categories of identifiers, and use the supplemented training dataset to optimize the parameters of the recognition model to obtain an optimized recognition model.

[0103] The aforementioned vehicle cleaning method, through user-marked uncleaned labels, accurately locates stains that the identification model fails to recognize or misidentifies, as well as areas where the cleaning robot has not cleaned thoroughly. This allows for targeted optimization of the identification model, significantly improving its accuracy. Through user interaction, the most authentic, accurate, and user-focused optimized samples can be obtained for parameter tuning of the identification model, ensuring that the continuously optimized model best meets user needs and thereby improving user satisfaction with the vehicle cleaning system.

[0104] In some alternative embodiments, step 206 includes:

[0105] Clustering is performed on the 3D points in the vehicle body point cloud to divide the vehicle body point cloud into point cloud regions corresponding to multiple vehicle body parts;

[0106] Based on the spatial location and category information of stains within the corresponding point cloud area of ​​each vehicle body component, the sub-cleaning trajectory corresponding to each point cloud area is determined.

[0107] Based on the preset cleaning sequence of each vehicle body component, the sorted sub-cleaning trajectories are smoothed to obtain the cleaning trajectory.

[0108] A point cloud region refers to a collection of three-dimensional points that are physically and / or geometrically independent. Each point cloud region typically corresponds to a body part or structural surface of the target vehicle. For example, a point cloud region may correspond to a door panel, a wheel hub surface, or a section of the roof profile.

[0109] In this step, the initial vehicle point cloud usually has problems such as noise and uneven density. The controller can use methods such as statistical filtering or radius filtering to remove noise points that are significantly deviated from the main vehicle point cloud range in order to perform noise reduction processing on the vehicle point cloud.

[0110] Furthermore, the controller can employ Euclidean clustering to cluster the 3D points in the vehicle body point cloud. The controller can primarily rely on the spatial location information of the 3D points to calculate the spatial distance between them, and then classify the 3D points in the vehicle body point cloud based on these spatial distances. Euclidean clustering can group 3D points that are spatially close together into one class.

[0111] Alternatively, the controller can use region growing clustering to cluster the 3D points in the vehicle body point cloud. The region growing clustering method not only considers the spatial distance between 3D points, but also pays attention to the similarity in normal vectors and curvatures of 3D points. The controller can start from any 3D point and traverse its neighboring points. If the angle between the normal vector of a neighboring point and the normal vector of this 3D point is less than a certain threshold and the curvatures are similar, it is regarded as the same type of 3D point, and so on, until there are no neighboring points whose normal vector angles with the normal vector of this 3D point are less than a certain threshold and the curvatures are similar.

[0112] Furthermore, it should be noted that the controller can determine the sub-cleaning trajectories corresponding to each point cloud region based on the spatial position, category information, and adhesion strength information of the stains within the point cloud regions corresponding to each vehicle body component.

[0113] Specifically, the controller can match the trajectory type of each point cloud region from a mapping table of multiple vehicle body components and various cleaning trajectories. Subsequently, based on the spatial positions of the 3D points in the point cloud region and a preset incremental distance, multiple waypoints that conform to the trajectory type are determined. Among them, there is a sequential relationship between the multiple waypoints. Further, the controller can adjust the waypoints corresponding to the 3D points with stain marks in the point cloud region according to the category information and adhesion strength information of the stains corresponding to these 3D points, obtaining multiple adjusted waypoints within the current point cloud region. Finally, interpolation and smoothing processing are performed on the waypoints to obtain multiple spatially continuous waypoints as the sub-cleaning trajectory; the waypoints are in the direction of the normal vector of neighboring 3D points.

[0114] Specifically, when the trajectory type corresponding to the point cloud region is an equidistant grid scanning trajectory, the controller can, for example, automatically calculate the row spacing and column spacing of the grid according to the boundary of the point cloud region and generate a "field" - shaped waypoint covering the entire region. When the trajectory type corresponding to the point cloud region is a progressive spiral coverage trajectory, the controller can, for example, fit the central axis and the maximum radius of the point cloud region and then generate spiral waypoints with a fixed pitch from the inside out. Subsequently, interpolation and filtering are performed on the generated multiple waypoints to ensure the smooth movement of the end of the cleaning robot.

[0115] The controller can, for example, adjust the trajectory type and the incremental distance corresponding to the 3D points with stain marks according to the category information and adhesion strength information of the stains corresponding to the 3D points within the point cloud region.

[0116] Specifically, when the trajectory type corresponding to the point cloud region is an equidistant grid scanning trajectory, the controller can, for example, automatically calculate the row spacing and column spacing of the grid according to the boundary of the point cloud region, generate a "field" - shaped waypoint covering the entire region. Subsequently, for the category information and adhesion strength information of the stains corresponding to several three - dimensional points with stain marks in the point cloud region, when the category information indicates that the stains on the vehicle body corresponding to these three - dimensional points are bird droppings and the adhesion strength information indicates that the adhesion strength value of the stain is relatively high, the controller can adopt a spiral progressive movement method to adjust the reference points corresponding to these three - dimensional points with stain marks, which have a sequential relationship, and reduce the distances between these reference points and adjacent three - dimensional points, so as to increase the pressure on the end of the cleaning robot to scrub the stains.

[0117] After the controller adjusts the waypoints corresponding to the three - dimensional points with stain marks and obtains multiple adjusted waypoints in the current point cloud region, it can sort the cleaning order of the multiple adjusted waypoints in the current point cloud region. In one embodiment, the controller can determine the first order of the multiple adjusted waypoints according to the trajectory type corresponding to the point cloud region, and determine the second order of the waypoints corresponding to the multiple three - dimensional points with stain marks in the point cloud region according to the trajectory type corresponding to the multiple three - dimensional points with stain marks.

[0118] When the controller controls the end of the cleaning robot to move according to the first order until it moves to the waypoint corresponding to the first three - dimensional point with a stain mark, the controller can continue to move according to the second order, and after moving to the waypoint corresponding to the last three - dimensional point with a stain mark, continue to move according to the first order to the remaining waypoints to perform the automatic cleaning of the target vehicle.

[0119] The above vehicle cleaning method, by establishing the correspondence between vehicle body parts and trajectory types and presetting the cleaning order of vehicle body parts, organizes the separate cleaning tasks on different vehicle body surfaces into continuous and coherent cleaning tasks. A coherent cleaning trajectory is obtained after the smooth processing of multiple sub - cleaning trajectories. Moreover, during the generation process of each sub - cleaning trajectory, detailed adjustments of the cleaning trajectory can be made for different stains. This can not only ensure the coherence of the cleaning process of the cleaning robot and avoid secondary contamination of the already cleaned areas on the vehicle body surface, but also ensure targeted cleaning of the stain areas, thus significantly improving the cleaning quality of automatic cleaning.

[0120] In some optional embodiments, determining the sub - cleaning trajectory corresponding to each point cloud region based on the spatial position and category information of the stains in the point cloud region corresponding to each vehicle body part includes:

[0121] A preset cleaning strategy for vehicle body parts is adopted to determine the first sub-cleaning trajectory of the point cloud region corresponding to each vehicle body part; the cleaning strategy for vehicle body parts includes the first cleaning method corresponding to multiple vehicle body parts;

[0122] Using a preset stain type cleaning strategy, the second sub-cleaning trajectory corresponding to stains of different categories is determined in the point cloud area corresponding to each body part; the stain type cleaning strategy includes the second cleaning method corresponding to stains of multiple categories.

[0123] By fusing the first and second sub-cleaning trajectories, the sub-cleaning trajectories corresponding to each point cloud region are obtained.

[0124] The body component cleaning strategy refers to a set of preset cleaning rules specifically designed for different body components. The controller can match the corresponding first cleaning method for each body component based on the body component cleaning strategy.

[0125] As an example, the first cleaning method includes information such as the cleaning tools (e.g., soft brushes, high-pressure water guns, sponges) corresponding to each body part, the cleaning motion mode (e.g., horizontal sweeping, circular wiping, linear reciprocating), and basic cleaning parameters (e.g., the distance between the cleaning tool and the surface of the part, and the movement speed).

[0126] Stain type cleaning strategy refers to a set of preset cleaning rules specifically designed for different types of stains. The controller can match a corresponding second cleaning method for each type of stain based on the stain type cleaning strategy.

[0127] As an example, the second cleaning method includes cleaning intensity (such as water spray pressure and scrubbing force), cleaning sequence (such as wetting first, then scrubbing, and finally rinsing), cleaning time, and whether to use a special cleaning agent. The basis for the determination is the adhesion of the stain (such as tree sap has strong adhesion, while dust has weak adhesion), chemical properties (such as oil stains require degreasing agents, while bird droppings contain acidic substances and require neutral cleaning agents), and removal difficulty (such as mud and sand stains with weak adhesion, which can be treated with "low-pressure water spray + rapid sweeping"; oil stains with strong adhesion, which require "high-pressure water spray + special degreasing agent + slow reciprocating scrubbing").

[0128] The first sub-cleaning trajectory is a three-dimensional motion path planned within the corresponding point cloud area based on the vehicle body component cleaning strategy, which adapts to the basic cleaning needs of the vehicle body component. The first sub-cleaning trajectory only considers the characteristics of the vehicle body component itself, such as material and shape, and does not involve specific stain types. It is the basic cleaning trajectory framework for the corresponding point cloud area of ​​the vehicle body component.

[0129] The second sub-cleaning trajectory refers to a three-dimensional motion path planned within the corresponding point cloud area for specific types of stains, based on a stain type cleaning strategy. This path is adapted to the stain removal needs and can focus on the characteristics of the stain (such as type and spatial location). The second sub-cleaning trajectory is a targeted optimization of the first sub-cleaning trajectory. For example, the second sub-cleaning trajectory for oil stains in the point cloud area of ​​a car door might be "a circular reciprocating path within a 10cm radius around the coordinate point of the oil stain, with a water spray pressure of 0.5MPa, a brushing time of 8 seconds, and the application of a degreaser."

[0130] For stains of different types in any body part, the second sub-cleaning trajectory has a higher priority than the first sub-cleaning trajectory. This can be understood as follows: for any point cloud area of ​​any body part, the pixels corresponding to the stains are cleaned according to the second sub-cleaning trajectory, while the remaining pixels are cleaned according to the first sub-cleaning trajectory.

[0131] The controller ultimately merges the first and second sub-cleaning trajectories to form a sub-cleaning trajectory. This maintains the basic framework for cleaning the vehicle body parts, ensuring the sub-cleaning trajectory conforms to the shape of the parts and protects their materials. It also precisely adapts to the removal needs of all stains within the corresponding point cloud area of ​​the vehicle body part, representing the optimal operating path for the cleaning robot within that area. For example, the final sub-cleaning trajectory for the point cloud area corresponding to the car door might be: "A basic longitudinal reciprocating path from the left to the right side of the door. The area containing oil stains (X-axis 800-950mm, Y-axis 1500-1650mm) is converted to a circular reciprocating path at a speed of 10cm / s and a pressure of 0.5MPa, with simultaneous spraying of degreaser; the remaining area maintains a longitudinal reciprocating path of 15cm / s and 0.2MPa."

[0132] The aforementioned vehicle cleaning method can fully integrate the characteristics of each vehicle body component, such as material, curvature, and fragility, to plan a basic cleaning trajectory framework, thereby minimizing damage to the vehicle body during the cleaning process and ensuring cleaning safety. Furthermore, it can develop customized cleaning solutions for different types of stains based on their adhesion, chemical properties, and removal difficulty, significantly improving the accuracy of stain removal and avoiding incomplete or over-cleaning. By integrating the first and second sub-cleaning trajectories, it combines the basic needs of the components with the needs of stain removal. This approach can significantly improve cleaning efficiency while ensuring cleaning quality.

[0133] In some optional embodiments, the vehicle washing method further includes:

[0134] Based on the assessment results, adjust the cleaning strategy according to the type of stain.

[0135] Specifically, the second cleaning method has multiple cleaning levels, with each level indicating a different level of stain removal intensity;

[0136] Based on the assessment results, the steps to adjust the cleaning strategy according to the type of stain include:

[0137] Find the intersection of the vehicle body image marked with a stain label and the vehicle body image marked with an unclean label to obtain the adjustment area marked with both stain and unclean labels.

[0138] Based on the stain identification of the adjustment area marker, determine the category information of the stain to be adjusted;

[0139] In the stain type cleaning strategy, the cleaning level corresponding to the category information of the stain to be adjusted will be increased.

[0140] As shown in the above embodiments, the controller can perform a union operation on the vehicle body image marked with an "unwashed" label and the corresponding vehicle body image marked with a "stain" label:

[0141] If a pixel has both a stain marker and an unclean marker, it means that the recognition model has identified the stain, but the cleaning robot has not cleaned it properly. The controller can mark the pixel with a first-type marker to indicate that the recognition model has correctly identified the stain, but the cleaning robot has not cleaned it thoroughly.

[0142] If a pixel only has an "unwashed" mark, it means that the recognition model did not recognize the stain and the cleaning robot did not clean it properly. The controller can mark the pixel with a second type of mark to indicate that the recognition model missed the stain.

[0143] If a pixel only has a stain marker, it means that the recognition model has identified the stain and it has been cleaned by the cleaning robot. The controller does not need to mark that pixel.

[0144] For vehicle images with the first type of label, it can be understood that the reason for the incomplete cleaning of the vehicle is that the cleaning intensity of the cleaning robot is insufficient. However, the location and category information of the stains identified by the recognition model are accurate. If the vehicle images reported by the user only carry the first type of label, the controller can adjust the cleaning strategy based on the type of stain without retraining the recognition model.

[0145] In this embodiment, the controller can, for example, increase the cleaning level of the stain category corresponding to pixels that simultaneously have stain markings and not clean markings in the stain type cleaning strategy by one or more levels. For example, if the user marks the pixels corresponding to oil stains as not clean, the controller can, for example, increase the cleaning level of oil stains in the stain type cleaning strategy, such as increasing the water spray pressure of oil stains from 0.5MPa to 0.8MPa, and the cleaning time from 8 seconds to 12 seconds, to enhance the cleaning intensity of this type of stain and ensure that subsequent cleaning can completely remove this type of stain.

[0146] As an example, the increase in cleaning level can be determined based on the degree of incomplete cleaning. Slight residue can be increased by 1 level, and severe residue can be increased by 2 levels. The degree of incomplete cleaning can be selected by the user when marking the uncleaned area, thus avoiding the controller from excessively increasing the cleaning intensity, which could lead to a waste of cleaning resources or damage to vehicle body parts.

[0147] The aforementioned vehicle cleaning method can fundamentally solve the problem of cleaning residue by increasing the cleaning level of the corresponding stain type, continuously improving the overall cleaning effect. During the adjustment process, the adjustment area determined by finding the intersection can directly lock the location of the stain that is not thoroughly cleaned, clearly identify the stain type to be adjusted, avoid blind adjustment, and ensure that the adjustment only targets the problematic stain type without affecting the cleaning strategy for other types of stains. By dynamically adjusting the cleaning level, it can avoid insufficient cleaning caused by the cleaning level being too low, and also prevent the waste of resources caused by blindly increasing the cleaning level, achieving efficient use of resources such as water and cleaning agents.

[0148] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0149] Based on the same inventive concept, this application also provides a vehicle washing apparatus for implementing the vehicle washing method described above. The solution provided by this vehicle washing apparatus is similar to the solution described in the vehicle washing method above; therefore, the specific limitations in one or more apparatus embodiments provided below can be found in the limitations of the vehicle washing method described above, and will not be repeated here.

[0150] In one embodiment, such as Figure 3 As shown, a vehicle washing device 300 is provided, applied to the aforementioned vehicle washing system 100. The vehicle washing device 300 includes:

[0151] The recognition module 302 is used to perform stain recognition on the vehicle body image using a pre-trained recognition model to obtain a vehicle body image marked with stain labels; the stain labels in the vehicle body image are used to mark the image location corresponding to the stains on the target vehicle, and the stain labels contain stain category information;

[0152] Module 304 is used to construct a vehicle point cloud marked with stains based on a vehicle image marked with stains; the stains in the vehicle point cloud are used to mark the spatial location of the stains on the target vehicle.

[0153] The control module 306 is used to determine the cleaning trajectory of the cleaning robot based on the vehicle body point cloud and the category information of the stains, and to control the cleaning robot to clean the target vehicle according to the cleaning trajectory.

[0154] The optimization module 308 is used to respond to the user's evaluation results for the cleaned target vehicle, optimize the recognition model based on the evaluation results, and use the optimized recognition model to identify stains on the body images of other vehicles.

[0155] In some optional embodiments, the evaluation results include unwashed markings made by the user on images of the vehicle body, which are used to mark the image locations corresponding to stains on the target vehicle after cleaning;

[0156] Optimization module 308 is also configured as follows:

[0157] Find the union of the images of the car body marked with the stain label and the images of the car body marked with the unclean label, and you will get the images of the car body marked with both the stain label and the unclean label.

[0158] Images of vehicles marked with both dirt and unclean markings are added to the training set of the recognition model, and the recognition model is retrained using the training set to optimize it.

[0159] In some alternative embodiments, the control module 306 is further configured to:

[0160] Clustering is performed on the 3D points in the vehicle body point cloud to divide the vehicle body point cloud into point cloud regions corresponding to multiple vehicle body parts;

[0161] Based on the spatial location and category information of stains within the corresponding point cloud area of ​​each vehicle body component, the sub-cleaning trajectory corresponding to each point cloud area is determined.

[0162] Based on the preset cleaning sequence of each vehicle body component, the sorted sub-cleaning trajectories are smoothed to obtain the cleaning trajectory.

[0163] In some alternative embodiments, the control module 306 is further configured to:

[0164] A preset cleaning strategy for vehicle body parts is adopted to determine the first sub-cleaning trajectory of the point cloud region corresponding to each vehicle body part; the cleaning strategy for vehicle body parts includes the first cleaning method corresponding to multiple vehicle body parts;

[0165] Using a preset stain type cleaning strategy, the second cleaning trajectory corresponding to stains of different categories is determined in the point cloud area corresponding to each body part; the stain type cleaning strategy includes the second cleaning method corresponding to stains of multiple categories.

[0166] By fusing the first and second sub-cleaning trajectories, the sub-cleaning trajectories corresponding to each point cloud region are obtained.

[0167] In some alternative embodiments, the control module 306 is further configured to:

[0168] Based on the assessment results, adjust the cleaning strategy according to the type of stain.

[0169] In some optional embodiments, the second cleaning method is provided with multiple cleaning levels, with different cleaning levels indicating different stain cleaning intensities;

[0170] Control module 306 is also configured as follows:

[0171] Find the intersection of the vehicle body image marked with a stain label and the vehicle body image marked with an unclean label to obtain the adjustment area marked with both stain and unclean labels.

[0172] Based on the stain identification of the adjustment area marker, determine the category information of the stain to be adjusted;

[0173] In the stain type cleaning strategy, the cleaning level corresponding to the category information of the stain to be adjusted will be increased.

[0174] Each module in the above-mentioned device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0175] Figure 4 A schematic diagram of the structure of the electronic device provided in this application. Figure 4 As shown, the electronic device 400 provided in this embodiment includes at least one processor 401 and a memory 402. Optionally, the device 400 further includes a communication component 403. The processor 401, memory 402, and communication component 403 are connected via a bus 404.

[0176] In a specific implementation, at least one processor 401 executes computer execution instructions stored in memory 402, causing at least one processor 401 to perform the above-described method.

[0177] The specific implementation process of processor 401 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0178] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0179] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0180] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0181] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0182] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0183] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0184] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0185] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0186] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0187] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0188] If a function is implemented as 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 this invention, or the part that contributes to the prior art, or a part 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 of the various embodiments of this invention. 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.

[0189] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0190] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A vehicle washing method characterized by, The application is applied to a vehicle cleaning system, the vehicle cleaning system comprises a cleaning robot and a collection device; the cleaning robot is used for cleaning a target vehicle parked in a target area, and the collection device is used for collecting a vehicle body picture of the target vehicle; The method comprises: adopting a pre-trained identification model to perform stain identification on the vehicle body picture, to obtain a vehicle body picture marked with a stain identifier; the stain identifier in the vehicle body picture is used for marking a picture position corresponding to a stain on the target vehicle, and the stain identifier contains category information of the stain; based on the vehicle body picture marked with the stain identifier, a vehicle body point cloud marked with the stain identifier is constructed; the stain identifier in the vehicle body point cloud is used for marking a spatial position corresponding to a stain on the target vehicle; based on the vehicle body point cloud and the category information of the stain, a cleaning trajectory of the cleaning robot is determined, and the cleaning robot is controlled to clean the target vehicle according to the cleaning trajectory; in response to an evaluation result of a user for the cleaned target vehicle, the identification model is optimized based on the evaluation result, and the optimized identification model is adopted to perform stain identification on vehicle body pictures of remaining vehicles.

2. The method of claim 1, wherein, The evaluation result comprises an unwashed identifier marked by the user on the vehicle body picture, and the unwashed identifier is used for marking a picture position corresponding to a stain on the cleaned target vehicle; The optimization of the identification model based on the evaluation result comprises: taking a union of the vehicle body picture marked with the stain identifier and the vehicle body picture marked with the unwashed identifier, to obtain a vehicle body picture marked with both the stain identifier and the unwashed identifier; adding the vehicle body picture marked with both the stain identifier and the unwashed identifier into a training set of the identification model, and retraining the identification model by using the training set, to optimize the identification model.

3. The method of claim 2, wherein, The determination of the cleaning trajectory of the cleaning robot based on the vehicle body point cloud and the category information of the stain comprises: performing cluster division on three-dimensional points in the vehicle body point cloud, to divide the vehicle body point cloud into point cloud regions corresponding to vehicle body parts; determining a sub cleaning trajectory corresponding to each point cloud region based on spatial positions and category information of stains in each point cloud region corresponding to a vehicle body part; performing smoothing processing on the sorted multiple sub cleaning trajectories based on a preset cleaning sequence of each vehicle body part, to obtain the cleaning trajectory.

4. The method of claim 3, wherein, The determination of the sub cleaning trajectory corresponding to each point cloud region based on spatial positions and category information of stains in each point cloud region corresponding to a vehicle body part comprises: adopting a preset vehicle body part cleaning strategy to determine a first sub cleaning trajectory of each point cloud region corresponding to a vehicle body part; the vehicle body part cleaning strategy comprises a first cleaning mode corresponding to multiple vehicle body parts; adopting a preset stain type cleaning strategy to determine a second sub cleaning trajectory corresponding to stains of different category information in each point cloud region corresponding to a vehicle body part; the stain type cleaning strategy comprises a second cleaning mode corresponding to stains of multiple category information. Fuse the first sub-cleaning track and the second sub-cleaning track to obtain a sub-cleaning track corresponding to each point cloud region.

5. The method of claim 4, wherein, The method further comprises: Adjusting the stain type cleaning strategy based on the evaluation result.

6. The method of claim 5, wherein, The second cleaning mode is provided with multiple cleaning levels, and different cleaning levels indicate different stain cleaning intensities. The adjusting the stain type cleaning strategy based on the evaluation result comprises: Intersecting the vehicle body picture marked with the stain mark and the vehicle body picture marked with the unwashed mark to obtain an adjustment region marked with both the stain mark and the unwashed mark; Determining the category information of the stain to be adjusted based on the stain mark marked on the adjustment region; Upgrading the cleaning level corresponding to the category information of the stain to be adjusted in the stain type cleaning strategy.

7. A vehicle washing apparatus characterized by comprising: The vehicle washing system comprises a cleaning robot and a collection device; the cleaning robot is used to clean a target vehicle parked in a target area, and the collection device is used to collect a vehicle body picture of the target vehicle. The device comprises: An identification module configured to perform stain identification on the vehicle body picture by using a pre-trained identification model to obtain a vehicle body picture marked with a stain mark; the stain mark in the vehicle body picture is used to mark a picture position of a stain on the target vehicle, and the stain mark contains category information of the stain; A construction module configured to construct a vehicle body point cloud marked with the stain mark based on the vehicle body picture marked with the stain mark; the stain mark in the vehicle body point cloud is used to mark a spatial position of the stain on the target vehicle; A control module configured to determine a cleaning track of the cleaning robot based on the vehicle body point cloud and the category information of the stain, and control the cleaning robot to clean the target vehicle according to the cleaning track; An optimization module configured to optimize the identification model based on an evaluation result of a user for the cleaned target vehicle in response to the evaluation result, and perform stain identification on vehicle body pictures of remaining vehicles by using the optimized identification model. 8.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-7. The processor executes the computer program to realize the steps of the method in claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method in claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method in claims 1 to 6.

Images