A car washing machine control method based on laser radar modeling

Abstract

The application provides a car washing machine control method based on laser radar modeling, and the control method comprises the following steps: S1, system initialization; S2, data acquisition; S3, coordinate conversion; S4, profile generation; S5, three-dimensional model construction; and S6, adaptive cleaning. Through multi-radar systemized coordinate conversion, timing accurate compensation and high-density sampling, the application realizes high-precision whole vehicle three-dimensional modeling, can construct a high-precision whole vehicle three-dimensional point cloud model, and the sampling precision and density are significantly better than those of the existing single-point ranging and segmented detection scheme, thereby providing accurate data basis for adaptive control; the lifting mechanism and the opening and closing mechanism driven based on the three-dimensional model can realize continuous stepless following-shape movement, can adaptively match vehicle models, do not need manual preset parameters, and realize 'one machine washing ten thousand vehicles'.

Description

Technical Field

[0001] This invention relates to the field of car wash machines, and more particularly to a car wash machine control method based on lidar modeling. Background Technology

[0002] A car wash machine is a machine that uses computer-controlled brushes and high-pressure water to automatically clean cars. It can be divided into automatic car wash machines and coin-operated car wash machines. It mainly consists of a control system, electrical circuits, air circuits, water circuits, and mechanical structures. Car wash machines are characterized by simple operation, an attractive appearance, and minimal damage to car paint. In recent years, they have been widely used in the automotive service industry. Car wash machines can be classified into tunnel type and reciprocating type according to their working method. In tunnel type operation, the car wash machine remains stationary, and the car is slowly moved through the working area by the machine. The car wash machine cleans the car according to a corresponding instruction program. Examples include continuous tunnel car wash machines and large tunnel type (trolleybus, bus, subway, passenger train) washing machines. In reciprocating type operation, the car remains stationary, and the car wash machine moves back and forth on a guide rail according to a certain program, simultaneously executing washing instructions. Examples include gantry reciprocating car wash machines and large (medium and small) mobile car wash machines.

[0003] Reciprocating car wash machines are widely used in various car wash locations due to their small footprint, high cleaning efficiency, and high degree of automation. Existing vehicle recognition and cleaning control solutions for reciprocating car wash machines mainly fall into the following technical categories: Fixed limit control schemes. This scheme detects the vehicle's extreme boundary positions by installing mechanical limit switches or photoelectric sensors at specific locations on the machine body. The lifting and opening / closing mechanisms can only switch between a few preset fixed positions. However, this scheme cannot perceive the vehicle's true curved contours and cannot provide differentiated responses to features such as roof curvature, hood height, and trunk height differences for different car models. The cleaning effect heavily relies on manually preset parameters. Single-point ranging control schemes sometimes incorporate infrared ranging sensors or single-line laser ranging modules to measure specific points on the vehicle body (such as the highest point of the roof or the widest points on both sides) as the frame moves, adjusting the brush height or width accordingly. However, single-point ranging only acquires discrete local distance information, failing to reconstruct continuous vehicle body surfaces or build a complete three-dimensional contour of the vehicle. The brushing mechanism can only rely on open-loop motion between two points, still unable to achieve true conformal surface fitting. Segmented detection control schemes, in a few cases, divide the vehicle body longitudinally into several segments, setting independent detection sensors in each segment to achieve segmented coarse identification. However, this scheme suffers from large blind spots within segments, significant abrupt changes between segments, and complex multi-sensor calibration. Furthermore, each actuator relies on its own detection signal, lacking a globally unified data benchmark, resulting in poor coordination among multiple mechanisms.

[0004] In summary, the existing car wash machine recognition data cannot be effectively aggregated into a complete model. An incomplete model cannot support precise continuous conformal control. The lack of a unified model leads to the inability of multiple mechanisms to coordinate and work together, ultimately resulting in poor vehicle model compatibility, inconsistent cleaning effects, and serious waste of resources. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the defects of poor vehicle model compatibility, poor consistency of cleaning effect and serious waste of resources in the prior art, and to provide a car wash machine control method based on lidar modeling.

[0006] The present invention solves the above-mentioned technical problems through the following technical solution:

[0007] This invention provides a car wash machine control method based on lidar modeling, the control method comprising the following steps:

[0008] S1. System initialization: Power-on calibration and lidar data initialization; establish global coordinate system reference and lidar calibration parameters; start lidar data acquisition; reset the actuator to the reference position.

[0009] S2. Data acquisition: Control the relative displacement between the lidar and the vehicle along the longitudinal direction of the vehicle body. The lidar acquires polar coordinate point cloud data of the vehicle at a set frequency and records the position and timing information corresponding to each frame of point cloud.

[0010] S3. Coordinate transformation: Convert the polar coordinates of each LiDAR point one by one, and transform them into the global coordinate system according to the installation and calibration parameters of each LiDAR and the position and timing information.

[0011] S4. Profile generation: The point clouds belonging to different lidars after conversion are partitioned, filtered and merged according to preset rules to generate the lateral profile of the vehicle at the current longitudinal position.

[0012] S5. Three-dimensional model construction: All transverse sections are stacked along the longitudinal direction of the vehicle body, and the whole vehicle three-dimensional point cloud model is generated and stored by interpolation fitting.

[0013] S6. Adaptive cleaning: Reuse the three-dimensional point cloud model of the whole vehicle generated in step S5. During the cleaning process, retrieve the corresponding profile contour data according to the current longitudinal position, drive the cleaning actuator to adaptively follow the body contour movement, and complete the cleaning process.

[0014] Preferably, in the global coordinate system, the X-axis corresponds to the vehicle width direction, the Y-axis corresponds to the vehicle length direction and is consistent with the relative displacement direction, and the Z-axis corresponds to the vehicle height direction.

[0015] Preferably, the lidar is installed at different positions on the cleaning actuator, and the lidar has independent X-axis installation offset parameters, Y-axis timing compensation parameters, and unified Z-axis installation reference parameters.

[0016] Preferably, in step S3, local coordinate transformation is performed on any valid point cloud data (L, θ);

[0017] Local X-coordinate = L × Cos(θ);

[0018] Local Z-coordinate = L×Sin(θ);

[0019] Then, based on the X-axis offset, Y-axis timing compensation, and Z-axis installation reference height of each lidar, global coordinate correction is performed, and the legality of the converted coordinate points is filtered to remove invalid point clouds.

[0020] Preferably, in step S4, different lidars collect point cloud data for different lateral side areas of the vehicle, and each lidar retains only the valid point cloud within its corresponding side area.

[0021] The point cloud within the overlapping region is deduplicated, and the profile contour is smoothed using a smoothing filter.

[0022] Preferably, in step S5, the three-dimensional point cloud model of the whole vehicle is established by a single relative displacement scan, and the modeling process and the cleaning process are separated in time.

[0023] The interpolation fitting includes data completion for the gaps between adjacent profiles.

[0024] Preferably, the cleaning actuator includes a reciprocating car wash machine and a tunnel car wash machine;

[0025] The reciprocating car wash machine includes a walking frame, a lifting cleaning module, and an opening and closing cleaning module, which are mounted on the walking frame.

[0026] The tunnel-type car wash machine includes a conveyor belt, a radar module, a cleaning mechanism, and a blower mechanism. The radar module, the cleaning mechanism, and the blower mechanism are arranged sequentially from left to right above the conveyor belt.

[0027] In this technical solution, the cleaning operation of the vehicle is performed by a cleaning actuator.

[0028] Preferably, the walking frame includes a support frame, a drive structure, and a walking track. The support frame and the drive structure are connected by the walking track, and the drive structure drives the support frame to move along the walking track.

[0029] The lifting cleaning module and the opening and closing cleaning module are mounted on a support frame, and a radar acquisition mechanism is installed on the front side of the support frame.

[0030] In this technical solution, the walking frame continuously reuses the whole vehicle 3D point cloud model in all processes such as pre-washing, brushing, and air drying, and there is no need to re-execute scanning and modeling between each process.

[0031] Preferably, the lifting cleaning module includes a lifting assembly, which is mounted on a support frame, and the output execution end of the lifting assembly is equipped with one or more of a top brush assembly, a top spray assembly, and a top drying assembly;

[0032] The opening and closing cleaning module includes an opening and closing component, which is installed on a support frame. The output execution end of the opening and closing component is equipped with one or more of a side brush component, a side spray component, and a side air drying component.

[0033] In this technical solution, the top brush assembly, top spray assembly, and top air drying assembly are driven by the lifting assembly to move adaptively with the top contour of the vehicle body; the side brush assembly, side spray assembly, and side air drying assembly are driven by the opening and closing assembly to move adaptively with the top contour of the vehicle body.

[0034] Preferably, the cleaning mechanism includes a high-pressure water and foam assembly, a small vertical brush assembly, left and right vertical brush assemblies, a top brush assembly, and a wax spraying assembly. The small vertical brush assembly is connected to the output end of the rotary drive assembly, the left and right vertical brush assemblies are connected to the output end of the distance adjustment drive assembly, and the top brush assembly is connected to the output end of the height adjustment drive assembly.

[0035] The blower mechanism includes a top blower assembly and a side blower assembly. The top blower assembly is connected to the output end of the blower height adjustment assembly, and the side blower assembly is connected to the output end of the angle adjustment drive assembly.

[0036] Based on common knowledge in the field, the above-mentioned preferred conditions can be combined arbitrarily to obtain various preferred embodiments of the present invention.

[0037] The positive and progressive effects of this invention are as follows:

[0038] This invention achieves high-precision 3D modeling of the entire vehicle through systematic coordinate transformation of multiple radar systems, precise time-series compensation, and high-density sampling. It can construct a high-precision 3D point cloud model of the entire vehicle, and the sampling accuracy and density are significantly better than existing single-point ranging and segmented detection schemes, providing a precise data foundation for adaptive control.

[0039] The lifting and opening / closing mechanisms driven by 3D models can achieve continuous stepless conformal motion, adaptively matching the roof curvature and body width changes of different car models such as sedans, SUVs, MPVs, and station wagons. No manual parameter preset is required, and it is adaptable to all car models, truly achieving "one machine to wash ten thousand cars".

[0040] The lifting and cleaning module and the opening and closing cleaning module share the same model data source. Based on the same Y-axis position, they synchronously retrieve the corresponding profile data, realizing true linkage and collaborative control. The multi-mechanism linkage eliminates the defects of multiple mechanisms acting independently and asynchronously in traditional solutions.

[0041] The entire process, from pre-washing, foaming, scrubbing to air drying, reuses the same model data, allowing for seamless switching between processes, continuous and consistent control logic, and full process coverage, thus ensuring the stability and reliability of the entire cleaning process.

[0042] The spraying, foaming, and waxing processes can be precisely controlled in terms of spray range, flow rate, and pressure according to the actual contours of the vehicle body. The brushing pressure is adaptively adjusted according to the curved surface, effectively reducing the waste of water, electricity, and chemicals, making efficient use of resources, and reducing operating costs.

[0043] The brush plate can dynamically conform to the concave and convex surfaces of the vehicle body, eliminating blind spots in cleaning. The position of the air knife is adjusted synchronously with the contour, eliminating drying dead angles, improving cleaning quality, and significantly enhancing the overall cleaning effect and efficiency. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the overall process of the car wash machine control method based on lidar modeling according to an embodiment of the present invention.

[0045] Figure 2 for Figure 1 The diagram shows a three-dimensional structure of a reciprocating car wash machine based on lidar modeling control method. Figure 1 .

[0046] Figure 3 for Figure 2 The diagram shows a three-dimensional structure of a reciprocating car wash machine based on lidar modeling control method. Figure 2 .

[0047] Figure 4 for Figure 2 The diagram shows a three-dimensional structure of the walking frame of the car wash machine control method based on lidar modeling.

[0048] Figure 5 for Figure 2 The diagram shows the three-dimensional structure of the support frame, opening and closing assembly, side brush assembly, and side spray assembly of the car wash machine control method based on LiDAR modeling. Figure 1 .

[0049] Figure 6 for Figure 5 The diagram shows the three-dimensional structure of the support frame, opening and closing assembly, side brush assembly, and side spray assembly of the car wash machine control method based on LiDAR modeling. Figure 2 .

[0050] Figure 7 for Figure 2 The diagram shows a three-dimensional structure of the side brush assembly, side spray assembly, and side drying assembly of the car wash machine control method based on lidar modeling.

[0051] Figure 8 for Figure 2 The diagram shows a three-dimensional structure of the lifting and cleaning module in the car wash machine control method based on LiDAR modeling. Figure 1 .

[0052] Figure 9 for Figure 8 The diagram shows a three-dimensional structure of the lifting and cleaning module in the car wash machine control method based on LiDAR modeling. Figure 2 .

[0053] Figure 10 for Figure 1 The diagram shown illustrates the principle of converting LiDAR polar coordinate data into an XYZ Cartesian coordinate system in the left-hand side of the LiDAR-based car wash machine control method.

[0054] Figure 11 for Figure 1 The diagram shown on the right side illustrates the principle of converting LiDAR polar coordinate data into an XYZ rectangular coordinate system for the car wash machine control method based on LiDAR modeling.

[0055] Figure 12 for Figure 1 The diagram shows a three-dimensional structure of a tunnel-type car wash machine based on a LiDAR-modeled control method.

[0056] Figure 13 for Figure 12 The diagram shows a three-dimensional structure of the small vertical brush machine component and the rotary drive component of the car wash machine control method based on LiDAR modeling.

[0057] Figure 14 for Figure 12 The diagram shows a three-dimensional structure of the left and right vertical brush components and the distance adjustment drive component of the car wash machine control method based on LiDAR modeling.

[0058] Figure 15 for Figure 12 The diagram shows a three-dimensional structure of the top brush assembly, height adjustment drive assembly, and wax spraying assembly in the car wash machine control method based on LiDAR modeling.

[0059] Figure 16 for Figure 12 The diagram shows a three-dimensional structure of the top blower component and the blower height adjustment component in the car wash machine control method based on lidar modeling.

[0060] Figure 17 for Figure 12The diagram shows a three-dimensional structure of the side blowing component and the angle adjustment drive component of the car wash machine control method based on lidar modeling.

[0061] Explanation of reference numerals in the attached figures

[0062] 1. Walking frame; 11. Support frame; 12. Drive structure; 13. Walking track;

[0063] 2. Lifting and cleaning module; 21. Lifting assembly; 22. Top brush assembly; 23. Top spray assembly; 24. Top drying assembly;

[0064] 3. Opening and closing cleaning module; 31. Opening and closing assembly; 32. Side brush assembly; 33. Side spray assembly; 34. Side drying assembly;

[0065] 4. Radar data acquisition mechanism;

[0066] 5. Conveyor belt;

[0067] 6. Radar module;

[0068] 7. Cleaning mechanism; 71. High-pressure water and foam assembly; 72. Small vertical brush assembly; 73. Rotary adjustment drive assembly; 74. Left and right vertical brush assemblies; 75. Spacing adjustment drive assembly; 76. Top brush assembly; 77. Height adjustment drive assembly; 78. Wax spraying assembly;

[0069] 8. Hair dryer mechanism; 81. Top blower assembly; 82. Hair dryer height adjustment assembly; 83. Side blower assembly; 84. Angle adjustment drive assembly. Detailed Implementation

[0070] The present invention will be further illustrated by way of embodiments below, but the present invention is not limited to the scope of the embodiments described herein.

[0071] Figures 1 to 17 The diagram shown is a structural schematic of an embodiment of the car wash machine control method based on lidar modeling of the present invention.

[0072] A control method for a car wash machine based on lidar modeling, the control method includes the following steps:

[0073] S1. System initialization: Power-on calibration and lidar data initialization are performed. A global coordinate system reference and lidar calibration parameters are established. LiDAR data acquisition is started, and the actuator is reset to the reference position.

[0074] The control unit loads the global XYZ coordinate system reference, dual lidar installation and calibration parameters, effective lidar scanning range, and distance filtering threshold.

[0075] Start two independent acquisition threads to connect to the left and right lidars respectively, and establish a point cloud cache area;

[0076] The control system synchronizes the walking frame 1, the lifting and cleaning module 2, and the opening and closing cleaning module 3 back to the reference origin O.

[0077] The lidar performs a self-check, filtering out original invalid points with a distance of 0 and a reflection intensity of 0, and waits for the modeling trigger command.

[0078] The effective scanning range of the lidar is 95°~170°, and the distance filtering threshold is 50~5000mm.

[0079] After connecting to the left and right lidars, enable triple data verification of frame header, version, and CRC8, and establish a point cloud cache area.

[0080] S2. Data acquisition: Control the relative displacement between the lidar and the vehicle along the longitudinal direction of the vehicle body. The lidar acquires polar coordinate point cloud data of the vehicle at a set frequency and records the position and timing information corresponding to each frame of point cloud.

[0081] The modeling command is issued, and the drive mechanism causes the lidar and the vehicle to generate a uniform relative displacement along the Y-axis of the vehicle body.

[0082] The left and right misaligned lidars continuously output raw polar coordinate data (detection distance L, scanning angle θ, reflection intensity) at a fixed frequency (e.g., 10Hz).

[0083] Real-time filtering of point cloud data in ineffective angle ranges;

[0084] Record the relative motion timestamp corresponding to each frame of point cloud for subsequent Y-axis timing compensation calculation;

[0085] After the lidar and the vehicle have completed the relative displacement along the entire vehicle length, the lidar data acquisition stops, and the original point cloud data of the entire vehicle is collected.

[0086] The effective angle range and effective distance threshold for lidar data acquisition are preset based on environmental interference conditions.

[0087] S3. Coordinate transformation: The polar coordinates of each LiDAR point are converted point by point, and then uniformly transformed to the global coordinate system based on the installation calibration parameters and position timing information of each LiDAR.

[0088] The raw output data of a lidar is in polar coordinates (detection distance L, scanning angle θ), lacking direct spatial coordinate information. This invention solves and maps each frame of polar coordinate point cloud to the device's global XYZ coordinate system by fixing the installation and calibration parameters.

[0089] S4. Profile generation: The point clouds belonging to different LiDARs after conversion are partitioned, filtered and merged according to preset rules to generate the lateral profile of the vehicle at the current longitudinal position.

[0090] The deduplication mesh size and smoothing filter parameters are preset according to the modeling accuracy requirements.

[0091] A proprietary merging algorithm is adopted, which consists of "independent left and right calculation - misaligned timing compensation - partitioned acquisition - single frame profile generation".

[0092] After completing the global coordinate transformation, each lidar is partitioned and filtered according to its installation location and acquisition area. The lidar on the left retains the point cloud of the left half of the vehicle body, and the lidar on the right retains the point cloud of the right half of the vehicle body. The advantage of this partitioning strategy is that it makes full use of the coverage advantage of multiple lidars while avoiding data redundancy and noise interference caused by overlapping point clouds in the same area.

[0093] For the few overlapping areas that may exist near the partition boundaries, a mesh deduplication algorithm is used. This algorithm uses X-axis mesh cells of a set size as the basic processing unit, and takes the average value of the Z-axis values ​​for multiple points within the same mesh to eliminate redundant noise. Based on this, a sliding window algorithm is used to smooth the profile contour, resulting in a continuous, uninterrupted, and ghost-free XZ transverse profile of the entire vehicle.

[0094] S5. 3D model construction: Stack all transverse sections along the longitudinal direction of the vehicle body, and generate and store the whole vehicle 3D point cloud model through interpolation and fitting.

[0095] The vehicle remains stationary throughout the process, while the equipment moves at a constant speed along the Y-axis of the vehicle using a walking frame 1. The lidar rotates and scans at a fixed frequency. By precisely matching the timing of the walking speed and the scanning frequency, high-density sampling with equal intervals is achieved along the longitudinal direction of the vehicle.

[0096] Each frame of the cross-section corresponds to a complete XZ contour data frame collected by the device at the current Y-axis position; as the device moves at a constant speed along the Y-axis of the vehicle body, the LiDAR continuously collects data and generates a new set of cross-sectional contours at fixed intervals along the Y-axis; by stacking all equally spaced cross-sections sequentially along the Y-axis and interpolating the gaps between adjacent cross-sections, a complete XYZ 3D point cloud model of the whole vehicle can be generated.

[0097] The model has the following characteristics: it is established in one go by a single uniform-speed scanning, and the modeling process and the cleaning process are completely separated in time; the model data is fixed and stored after modeling is completed, serving as the sole data source for all subsequent processes; the sampling density can be flexibly controlled by adjusting the travel speed and scanning frequency to adapt to different accuracy requirements.

[0098] S6. Adaptive cleaning: Reuse the three-dimensional point cloud model of the whole vehicle generated in step S5. During the cleaning process, retrieve the corresponding profile contour data according to the current longitudinal position, drive the cleaning actuator to adaptively follow the body contour movement, and complete the cleaning process.

[0099] This invention completely solves the industry pain points of traditional reciprocating car wash machines, such as poor vehicle compatibility, incomplete cleaning, high chemical consumption, low surface fit, and asynchronous operation of multiple actuators, by using a unified global XYZ coordinate transformation of the laser radar on the top of the frame, physical misalignment timing compensation, high-density whole-vehicle sampling during the reciprocating movement of the frame, whole-vehicle 3D point cloud pre-modeling, and full-process adaptive control of a single set of lifting integrated mechanism and a single set of opening and closing integrated mechanism based on the same source model.

[0100] The entire vehicle relies solely on a unified 3D model generated through a single pre-modeling process to drive the lifting and cleaning module 2 and the opening and closing cleaning module 3 to synchronously and adaptively adjust. No secondary detection or segmented recognition is required, and the entire process features data from the same source and coordinated action of multiple mechanisms.

[0101] The lifting and cleaning module 2 moves up and down along the Z-axis, dynamically adjusting its height and position based on the Z-axis contour data of the three-dimensional model, so as to achieve uniform coverage of the roof surface, dynamic following of the continuous surface, and matching of the air knife's height from the ground.

[0102] The opening and closing cleaning module 3 moves along the X-axis and dynamically adjusts the opening and closing distance according to the X-axis width contour data of the three-dimensional model to adapt to the concave and convex surfaces such as car doors, wheel arches, and rearview mirrors, so as to achieve flexible and seamless cleaning and drying.

[0103] Each process continuously calls the same model data during the reciprocating motion of the equipment. Each cleaning actuator synchronously obtains the corresponding profile based on the current Y-axis position, and works together to ensure the continuity and consistency of the entire process control.

[0104] In the global coordinate system, the X-axis corresponds to the width of the vehicle body, the Y-axis corresponds to the length of the vehicle body and is consistent with the direction of relative displacement, and the Z-axis corresponds to the height of the vehicle body.

[0105] The vehicle and the lidar generate relative displacement.

[0106] The lidar is installed at different locations on the cleaning actuator, preferably at the entrance or the front end of the cleaning actuator. The lidar has independent X-axis installation offset parameters, Y-axis timing compensation parameters, and unified Z-axis installation reference parameters.

[0107] The Y-axis timing compensation parameter is used to eliminate profile timing deviations caused by physical misalignment of multiple lidar sensors.

[0108] In step S3, local coordinate transformation is performed on any valid point cloud data (L, θ);

[0109] Local X-coordinate = L × Cos(θ);

[0110] Local Z-coordinate = L × Sin(θ);

[0111] Then, based on the X-axis offset, Y-axis timing compensation, and Z-axis installation reference height of each lidar, global coordinate correction is performed, and the legality of the converted coordinate points is filtered to remove invalid point clouds.

[0112] Where θ is the scanning angle of the lidar, with the vertical downward direction as the 0-degree reference.

[0113] Global X coordinate = LiDAR X-axis installation offset ± Local X-axis coordinate (select the plus or minus sign according to whether the LiDAR is installed on the left or right side);

[0114] Global Y-coordinate = Relative displacement velocity × Scan time + Y-axis timing compensation;

[0115] Global Z-coordinate = Z-axis mounting reference height - local Z-axis coordinate.

[0116] One of the key innovations of this invention is the Y-axis timing compensation. Because the dual lidars are staggered at the top of the frame to avoid interference from oncoming beams, there is a physical distance between the scanning planes of the two lidars in the Y-axis direction. This compensation corrects the Y-axis alignment of the point clouds from the left and right lidars during the coordinate transformation stage, ensuring that the profile data from different lidars correspond to the same longitudinal position of the vehicle, fundamentally eliminating model ghosting and distortion caused by misalignment.

[0117] In step S4, different lidars collect point cloud data of different lateral side areas of the vehicle, and each lidar retains only the valid point cloud within its corresponding side area.

[0118] The point cloud within the overlapping region is deduplicated, and the profile contour is smoothed using a smoothing filter.

[0119] In step S5, the three-dimensional point cloud model of the whole vehicle is established by a single relative displacement scan, and the modeling process and the cleaning process are separated in time.

[0120] Interpolation fitting includes data completion for the gaps between adjacent profiles.

[0121] The cleaning actuator includes a reciprocating car wash machine, which includes a walking frame 1, a lifting cleaning module 2 and an opening and closing cleaning module 3, with the lifting cleaning module 2 and the opening and closing cleaning module 3 mounted on the walking frame 1.

[0122] The lifting and cleaning module 2 moves along the Z-axis and adaptively adjusts the lifting height based on the model's Z-axis height data.

[0123] The opening and closing cleaning module 3 moves along the X-axis and adaptively adjusts the opening and closing spacing based on the model's X-axis width data.

[0124] In this technical solution, the cleaning operation of the vehicle is performed by a cleaning actuator.

[0125] The walking frame 1 includes a support frame 11, a drive structure 12 and a walking track 13. The support frame 11 and the drive structure 12 are connected by the walking track 13. The drive structure 12 drives the support frame 11 to move along the walking track 13.

[0126] The lifting cleaning module 2 and the opening and closing cleaning module 3 are mounted on the support frame 11.

[0127] A radar acquisition mechanism 4 is installed on the front side of the support frame 11.

[0128] The radar acquisition mechanism 4 can use a single lidar unit in conjunction with moving scanning to achieve 3D modeling, or it can use two or more lidar units to achieve multi-angle synchronous acquisition. The lidar unit can be installed on the top of the rack, or on the side or other locations depending on the equipment structure. It is only necessary to accurately measure its installation offset parameters during the calibration phase and correct them in the coordinate transformation algorithm.

[0129] The walking frame 1 can adopt any walking drive scheme such as servo motor chain drive, synchronous belt drive, gear rack drive or linear motor drive to achieve relative uniform speed displacement between the frame and the vehicle.

[0130] In this technical solution, the walking frame 1 continuously reuses the three-dimensional point cloud model of the whole vehicle in the entire process of pre-washing, brushing, and air drying, and there is no need to re-execute scanning and modeling between each process.

[0131] The lifting cleaning module 2 includes a lifting component 21, which is installed on the support frame 11. The output execution end of the lifting component 21 is equipped with one or more of the top brush component 22, the top spray component 23, and the top air drying component 24.

[0132] In this technical solution, the top brush assembly 22, the top spray assembly 23, and the top air drying assembly 24 are driven by the lifting assembly 21 to move adaptively with the top contour of the vehicle body.

[0133] The opening and closing cleaning module 3 includes an opening and closing component 31, which is installed on the support frame 11. The output execution end of the opening and closing component 31 is equipped with one or more of the following: a side brush component 32, a side spray component 33, and a side air drying component 34.

[0134] In this technical solution, the side brush assembly 32, the side spray assembly 33, and the side air drying assembly 34 are driven by the opening and closing assembly 31 to move adaptively with the side profile of the vehicle body.

[0135] The lifting cleaning module 2 and the opening and closing cleaning module 3 can add or remove cleaning components according to the cleaning process requirements, such as adding a drying device, a coating device, a pre-washing device, etc. As long as the whole vehicle three-dimensional model is used to achieve adaptive following control, they all fall within the protection scope of this invention.

[0136] The frame travel speed can be adjusted from 10mm / s to 50mm / s; the LiDAR scanning frequency can be adjusted from 5Hz to 20Hz; the LiDAR X-axis offset can be adjusted from ±500mm according to the actual size of the equipment; the top mounting reference height can be adjusted from 2000mm to 3000mm according to the equipment structure; and the overlapping grid size can be adjusted from 1mm to 10mm.

[0137] The technical solution of the present invention will be described in detail below using two lidars as an example.

[0138] The radar acquisition mechanism 4 includes a left top lidar and a right top lidar. The left top lidar and the right top lidar are installed on the top of the support frame 11 and adopt a staggered arrangement structure. The scanning planes of the two are physically misaligned by a set distance in the Y-axis direction. This misalignment is eliminated in the coordinate transformation stage by the timing compensation amount to ensure that the left and right profile data correspond to the same longitudinal position of the vehicle.

[0139] The control system is an industrial-grade controller that loads and executes the steps of the control method of this invention.

[0140] The global coordinate system is defined as follows: the origin O is set at the projection point of the top left lidar on the ground at the initial position of the rack, offset 250mm to the right along the X-axis; the X-axis is the horizontal direction of the equipment; the Y-axis is the longitudinal direction of the rack's back-and-forth movement; and the Z-axis is the vertical direction of the ground.

[0141] Calibration parameters for the top left lidar: X-axis offset is -250mm, top mounting reference height is 2500mm, and Y-axis timing compensation is -104.4mm (this value is the actual physical distance between the scanning planes of the two lidars in the Y-axis direction; the positive or negative sign indicates that the lidar is located in front of or behind the other lidar).

[0142] Calibration parameters for the top-right lidar: X-axis offset is 2750mm, and the top mounting reference height is 2500mm.

[0143] The two lidar units are at the same height, with a top mounting reference height of 2500mm.

[0144] It is worth noting that the specific values ​​of the above parameters can be adjusted according to the equipment model and installation conditions.

[0145] The polar coordinates were converted to global coordinates, and the LiDAR returned polar coordinate parameters of 1400mm detection distance and 30° scanning angle.

[0146] Step 1: Local coordinate conversion.

[0147] Calculation of horizontal lateral height difference (local Z-axis component): LocalZ(AB) = 1400 × sin(30°) = 700 mm.

[0148] Calculation of horizontal projection distance (local X component): LocalX (AC) = 1400 × cos(30°) ≈ 1212.44 mm.

[0149] Step 2: Global coordinate correction.

[0150] like Figure 10 As shown, take a certain scanning moment of the left-side lidar O_L as an example.

[0151] The equipment has fixed calibration dimensions. The horizontal projection distance from the origin O to O_L is 250mm, and the installation height of O_L from the ground is 2500mm.

[0152] X-axis global coordinate correction: X = 1212.44 − 250 = 962.44 mm;

[0153] Z-axis global coordinate correction: Z = 2500 − 700 = 1800 mm;

[0154] The final standard coordinates of the scanning point in the global coordinate system of the whole machine are: A (962.44mm, 1800mm).

[0155] like Figure 11 As shown, the example is taken at a certain scanning moment of the right-side lidar O_R.

[0156] The lidar on the right is symmetrically arranged with the lidar on the left, and the coordinate system conversion principle is the same, only the X-axis reference offset direction is opposite.

[0157] Equipment calibration parameters: The total horizontal spacing between the two lidars is 3000mm, the horizontal projection distance from the origin O to O_R is 3000-250=2750mm, and the installation height of O_R from the ground is 2500mm.

[0158] X-axis global coordinate correction: X = 2750 − 1212.44 = 1537.56 mm;

[0159] Z-axis global coordinate correction: Z = 2500 − 700 = 1800 mm;

[0160] The final standard coordinates of the scanning point of the right-side lidar in the global coordinate system of the whole machine are: B (1537.56mm, 1800mm).

[0161] To ensure modeling accuracy, the system performs multi-level filtering on the raw data:

[0162] Level 1, zero-value filtering: points with a detection distance of 0 or a reflection intensity of 0 are directly eliminated;

[0163] The second level, angle range filtering, retains only point cloud data within the valid scan range. In this embodiment, the valid scan range is set to 95°~170°, and data outside this range (such as environmental interference data like ground reflection, rack reflection, and wall reflection) is directly discarded.

[0164] The third level is distance threshold filtering, with the effective detection distance set to 50mm~5000mm. Points outside this range are directly rejected.

[0165] Level 4: Coordinate validity filtering. Points whose global coordinates exceed the reasonable range of the vehicle body, fall into the ground area, or are in the interference area of ​​the frame after conversion are directly removed.

[0166] After the coordinate transformation is completed, each LiDAR needs to perform partition filtering: the left LiDAR retains only valid points with X≤0 (corresponding to the left and middle areas of the vehicle body), and the right LiDAR retains only valid points with X≥0 (corresponding to the right and middle areas of the vehicle body), to ensure that the left and right data are complementary in space and do not overlap.

[0167] The matching relationship between the traveling speed of the walking frame 1 and the scanning frequency of the lidar, and the calculation method of the sampling interval.

[0168] The scanning speed of the walking frame 1 is 100mm / s, and the LiDAR scanning frequency is 10Hz (i.e., 10 frames of data are output per second).

[0169] The longitudinal travel distance of the rack corresponding to each frame profile is: sampling interval = travel speed / scanning frequency = 100 / 10 = 10mm.

[0170] This means that a complete XZ profile is captured every 10mm along the longitudinal direction of the vehicle body. Based on the standard sedan body length of approximately 4500mm, the whole vehicle scan can obtain approximately 2064 profile frames. Each profile frame contains hundreds of effective point clouds, and the total number of point clouds can reach millions, achieving high-density 3D reconstruction of the whole vehicle.

[0171] The above parameters can be flexibly adjusted according to actual accuracy requirements: increasing the travel speed or decreasing the scanning frequency can obtain a coarser model (suitable for rapid modeling scenarios), while decreasing the travel speed or increasing the scanning frequency can obtain a more accurate model (suitable for fine cleaning scenarios of high-end vehicles). Adjustments to each parameter do not depart from the protection scope of this invention.

[0172] The complete process of merging, deduplication, smoothing, and generating single-frame profiles for the left and right LiDAR point clouds is as follows: After coordinate transformation and partitioning, the left LiDAR contributes the point cloud data for the region X≤0, and the right LiDAR contributes the point cloud data for the region X≥0. (In the area near the longitudinal centerline of the vehicle body at X=0), there may be a small amount of overlapping point cloud data between the left and right LiDARs.

[0173] Overlapping mesh deduplication is performed by dividing the entire X-axis range into several meshes with 5mm as the X-axis mesh unit. For multiple point clouds existing within the same mesh (which may originate from the left LiDAR, right LiDAR, or both), the average of their Z-axis coordinates is taken as the final height value for that mesh location. This process eliminates redundant data in overlapping areas and effectively suppresses random noise.

[0174] Contour smoothing filtering employs a sliding window algorithm with a window size of 3 to smooth the Z-axis height sequence. Specifically, the height value of each point is obtained by a weighted average of its own height and the height values ​​of its left and right adjacent points, making the profile curve more continuous and smoother, and more closely resembling the actual surface morphology of the vehicle.

[0175] Single-frame profile stitching arranges all processed valid point clouds in ascending order of X-axis coordinates to generate a complete XZ lateral contour curve from the left to the right side of the vehicle body, and binds the Y-axis position coordinates corresponding to the current profile.

[0176] The process from single-frame profile to temporal stacking, interpolation fitting to complete 3D model generation is as follows: as the equipment moves at a constant speed along the Y-axis of the vehicle body (from the front to the rear or in the opposite direction), a frame of XZ profile is generated at fixed intervals, and each frame of profile is accompanied by its corresponding Y-axis position coordinates.

[0177] All cross-sections are arranged in ascending order of their Y-axis coordinates and stacked longitudinally along the Y-axis to form a point cloud set in three-dimensional space. Since the sampling interval is 10mm, the gap between adjacent cross-sections is extremely small. However, for areas with drastic curvature changes (such as the junction of the roof and the windshield), interpolation completion is still required to ensure the continuity of the model.

[0178] The interpolation fitting method employs linear interpolation for the gap between two adjacent cross-sections. At the corresponding X-axis position, the Z-axis height at the intermediate position is calculated based on the Z-axis height values ​​of the two cross-sections in proportion to the Y-axis distance, generating a transition cross-section. After interpolation processing, a complete, continuous, and uninterrupted XYZ 3D point cloud model of the entire vehicle is obtained.

[0179] The model is permanently stored, with the 3D model data stored in a structured format on the controller's local storage medium, serving as the sole data source for all subsequent cleaning processes. Simultaneously, the model data can be exported via a communication interface to standard point cloud file formats (such as .xyz, .pcd, etc.) for offline simulation, debugging analysis, and system optimization.

[0180] The process by which the lifting and cleaning module 2 drives the top cleaning component to adaptively operate based on the Z-axis height data of the 3D model is as follows: After the cleaning process is started, the equipment reciprocates along the Y-axis of the vehicle body. During the movement, the control unit obtains the current Y-axis position in real time, retrieves the XZ profile data corresponding to that position from the 3D model, and extracts the Z-axis height curve.

[0181] In the top spraying process, based on the current Z-axis height and its changing trend at the Y position, the control unit calculates the desired height position of the lifting assembly 21 and sends a position command to the servo drive system to maintain the optimal spraying distance between the top spraying assembly 23 and the roof surface. Simultaneously, the spray flow rate and pressure are adjusted according to changes in the roof height; the flow rate is appropriately increased at higher areas of the roof to ensure uniform coverage, while the flow rate is appropriately reduced at lower areas to avoid pesticide splashing and waste.

[0182] In the top brush washing process, the lifting component 21 drives the top brush component 22 to dynamically rise and fall in real time, following the continuous height curves of the roof, hood, and trunk. When the equipment moves to the roof area, the top brush component 22 maintains a constant contact pressure with the roof surface; when moving to the hood area, the lifting component 21 automatically lowers the height of the top brush component 22 to follow the lower body surface; when moving to the trunk area, it automatically raises or lowers to match the height changes of the trunk lid. This achieves continuous, close-fitting washing from the front to the rear of the vehicle.

[0183] In the top drying process, the top drying component 24 is raised and lowered synchronously to match the roof contour and adjust the height of the air blades of the top drying component 24 from the ground to eliminate water accumulation on the roof surface and ensure the drying effect.

[0184] The process by which the opening and closing mechanism drives the side cleaning component to operate adaptively based on the X-axis width data of the three-dimensional model is as follows: the control unit retrieves the XZ profile data at the current Y position from the three-dimensional model, extracts the X-axis boundary positions on the left and right sides of the vehicle body, and calculates the actual width of the vehicle body at the current Y position.

[0185] In the side high-pressure spraying process, the control unit calculates the expected opening and closing distance of the opening and closing component 31 based on the actual width of the vehicle body and the position of the two side boundaries, and sends position commands to the servo drive system so that the side spraying component 33 maintains the optimal spraying distance with the two side surfaces of the vehicle body, accurately covers both sides of the vehicle body, and avoids water overflow.

[0186] In the side brush washing process, the opening and closing distance of the side brush assembly 32 is dynamically adjusted according to the changes in the X-axis width on both sides of the vehicle body. When the equipment moves to the door area, the side brush assembly 32 remains in contact with the door surface; when it moves to the wheel arch area, the side brush assembly 32 automatically expands outward to adapt to the concave curvature of the wheel arch; when it moves to the rearview mirror area, the side brush assembly 32 automatically adjusts the spacing to avoid the protruding parts of the rearview mirror, achieving flexible contact cleaning and avoiding rigid collisions.

[0187] In the side drying process, the side drying component 34 adjusts its opening and closing width synchronously with the horizontal contour of the vehicle body to achieve thorough drying of both sides of the vehicle body and the gaps in the wheel arches.

[0188] The complete cleaning process of a reciprocating car wash machine based on the same 3D model is as follows:

[0189] Step 1: The vehicle drives into the cleaning station and is positioned stationary with the wheels in the positioning device.

[0190] Step 2: The traveling frame 1 moves at a constant speed from the front to the rear of the vehicle, while the radar acquisition mechanism 4 acquires data synchronously, completing the three-dimensional scanning and modeling of the entire vehicle, and then storing the model.

[0191] Step 3: The walking frame 1 returns from the rear of the vehicle to the front position, and at the same time performs the pre-washing spraying process. The lifting and lowering cleaning module 2 adjusts the top spraying height according to the model, and the opening and closing cleaning module 3 adjusts the side spraying width according to the model.

[0192] Step 4: The walking frame 1 moves from the front to the rear of the vehicle and performs the foam spraying process. The lifting cleaning module 2 and the opening and closing cleaning module 3 are adjusted synchronously and adaptively according to the model to make the foam evenly cover the entire surface of the vehicle.

[0193] Step 5: The walking frame 1 moves from the rear of the vehicle to the front, performing a combined brushing process of the top brush assembly 22 and the side brush assembly 32. The lifting assembly 21 rises and falls adaptively according to the roof outline, and the opening and closing assembly 31 opens and closes adaptively according to the width of the vehicle body. The two mechanisms move synchronously and in coordination.

[0194] Step six: The walking frame 1 moves from the front to the rear of the vehicle to perform the wax spraying process, and the mechanism adaptively adjusts according to the model.

[0195] Step 7: The traveling frame 1 moves from the rear of the vehicle to the front, and performs the air drying process in conjunction with the top air drying component 24 and the opening and closing component 31. The lifting component 21 and the opening and closing component 31 adjust the position of the air knife synchronously according to the model to complete the air drying of the entire vehicle.

[0196] The entire process requires no secondary LiDAR scanning, and all procedures continuously reuse the pre-modeled 3D data; seamless switching between procedures and continuous and consistent control logic.

[0197] The cleaning actuator also includes a tunnel car wash machine, which includes a conveyor belt 5, a radar module 6, a cleaning mechanism 7, and a blower mechanism 8. The radar module 6, the cleaning mechanism 7, and the blower mechanism 8 are arranged sequentially from left to right above the conveyor belt 5.

[0198] In use, the vehicle stays on the conveyor belt 5 and moves with the operation of the conveyor belt 5. During the movement, it passes through the radar module 6, the cleaning mechanism 7, and the blower mechanism 8 in sequence. When passing through the radar module 6, the vehicle's three-dimensional data is scanned using lidar. When passing through the cleaning mechanism 7, the vehicle is cleaned in sequence using small vertical brushes, left and right vertical brushes, and top brushes. During this process, high-pressure water, foam, and wax are sprayed to assist in the vehicle cleaning. When passing through the blower mechanism 8, air is blown from the top and sides of the vehicle to dry the vehicle's surface.

[0199] The cleaning mechanism 7 includes a high-pressure water and foam assembly 71, a small vertical brush assembly 72, left and right vertical brush assemblies 74, a top brush assembly 76, and a wax spraying assembly 78. The small vertical brush assembly 72 is connected to the output end of the rotary drive assembly 73, the left and right vertical brush assemblies 74 are connected to the output end of the distance adjustment drive assembly 75, and the top brush assembly 76 is connected to the output end of the height adjustment drive assembly 77.

[0200] Two small vertical brush machine components 72 are installed on both sides of the conveyor belt 5.

[0201] The angle of the small brushing component 72 is adjusted by the rotary drive component 73, which also provides rotational power to the small brushing component 72.

[0202] The rotary drive assembly 73 includes a first column, a servo bracket connected to the surface of the first column, a vertical brush suspension rotatably mounted on the surface of the first column, a servo motor mounted on the servo bracket, the output end of the servo motor connected to the vertical brush suspension, a small vertical brush rotary motor mounted on one end of the vertical brush suspension, and a small vertical brush machine assembly 72 connected to the output end of the small vertical brush rotary motor.

[0203] In use, the servo motor drives the vertical brush suspension to rotate, which in turn drives the small vertical brush assembly 72 to rotate. The angle of the small vertical brush assembly 72 is adjusted so that it is close to the vehicle. At the same time, the small vertical brush rotation motor drives the high-pressure water and foam assembly 71 to rotate, and the vehicle is cleaned.

[0204] The adjustable drive assembly 75 includes a load-bearing frame and a belt drive assembly. The belt drive assembly is mounted on the load-bearing frame. The two output ends of the belt drive assembly are respectively connected to movable square tube trolleys. Both movable square tube trolleys are equipped with large vertical brush rotary motors. The output ends of the large vertical brush rotary motors are connected to left and right vertical brush assemblies 74.

[0205] In use, the distance between the two movable square tube trolleys is adjusted by the belt drive assembly, so that the rotary drive assemblies 73 on both sides are brought closer to the vehicle. Then, the large vertical brush rotary motor drives the left and right vertical brush assemblies 74 to rotate, thereby using the left and right vertical brush assemblies 74 to perform the cleaning operation on the vehicle.

[0206] The height adjustment drive assembly 77 includes a support frame, a wax spraying assembly 78 is installed on the top of the support frame, and an adjustment frame is rotatably connected to the side of the support frame. The adjustment frame is connected to the output end of the top brush angle adjustment motor, which is installed on the side.

[0207] One end of the adjustment frame is rotatably connected to a top brush assembly 76, which is connected to the output end of a top brush rotary motor. The top brush rotary motor is installed on the adjustment frame.

[0208] A top brush counterweight is installed at the other end of the adjustment frame.

[0209] In use, the top brush angle adjustment motor drives the adjustment frame and top brush assembly 76 to rotate, adjusting the height of the top brush assembly 76 so that the top brush assembly 76 is close to the vehicle. Then, the top brush rotation motor drives the top brush assembly 76 to rotate, cleaning the vehicle.

[0210] The blower mechanism 8 includes a top blower assembly 81 and a side blower assembly 83. The top blower assembly 81 is connected to the output end of the blower height adjustment assembly 82, and the side blower assembly 83 is connected to the output end of the angle adjustment drive assembly 84.

[0211] The blower height adjustment assembly 82 includes a gantry frame and a belt drive lifting control assembly. The belt drive lifting control assembly is installed on the gantry frame. The output end of the belt drive lifting control assembly is connected to a lifting square tube trolley. The lifting square tube trolley is installed on the side of the lifting platform. Multiple top blower assemblies 81 are installed on the top of the lifting platform.

[0212] The drive wheel of the belt-driven lifting control assembly uses double synchronous pulleys, and synchronous belts are used on both the left and right sides for transmission.

[0213] In use, the height of the lifting square tube trolley, the lifting platform and the top blowing assembly 81 are adjusted by the belt drive lifting control component, so that the top blowing assembly 81 is close to the vehicle. Then, the top blowing assembly 81 blows air onto the vehicle from above to dry the vehicle.

[0214] Side blowing components 83 are installed on both sides of the conveyor belt 5.

[0215] The angle adjustment drive assembly 84 includes a second column, a motor bracket connected to the surface of the second column, a mounting bracket rotatably mounted on the surface of the second column, an angle adjustment motor mounted on the motor bracket, the output end of the angle adjustment motor connected to the mounting bracket, and a side blowing assembly 83 mounted on one end of the mounting bracket.

[0216] In use, the angle adjustment motor drives the mounting suspension and side air blowing assembly 83 to rotate, so that the side air blowing assembly 83 is close to the vehicle, and then the side air blowing assembly 83 blows air onto the vehicle from the side to dry the vehicle.

[0217] The complete cleaning process of a tunnel-type car wash machine based on the same 3D model is as follows:

[0218] Step 1: The vehicle enters the tunnel-type car wash machine and stops on the conveyor belt 5. The conveyor belt 5 moves the vehicle, and the radar module 6 begins to scan and collect point cloud data, and performs coordinate transformation according to the algorithm.

[0219] Step 2: Activate the high-pressure water and foam assembly 71 to perform high-pressure water rinsing and foam spraying;

[0220] Step 3: Enter the small vertical brushing stage. Based on the modeling data on both sides, control the small vertical brushing machine component 72 to always maintain a constant distance from the side of the vehicle to brush the lower part of the vehicle body, tires, and wheel rims.

[0221] Step 4: Enter the left and right vertical brush stage. Based on the modeling data on both sides, control the rotation of the large vertical brush motors on both sides, and drive the moving square tube trolley and the left and right vertical brush assemblies 74 to move through the belt drive assembly. This ensures that the left and right vertical brush assemblies 74 can always brush the sides of the vehicle body with a certain force. Because of the front-mounted 3D modeling data, the position of the vehicle's rearview mirror can be accurately identified. Before the left and right vertical brush assemblies 74 come into contact with the rearview mirror, the left and right vertical brush assemblies 74 can be controlled to avoid it in advance.

[0222] Step 5: Enter the top brush stage. Based on the roof modeling data, control the rotation of the top brush angle adjustment motor to control the height of the top brush. At the same time, use the top brush rotation motor to drive the top brush assembly 76 to rotate and wash the vehicle.

[0223] Step 6: After the top brush is finished, use the wax spraying component 78 to automatically spray wax.

[0224] Step 7: Enter the top blowing stage. Based on the roof modeling data, control the height of the top blowing component 81 through the belt drive lifting control component to make the air outlet close to the roof to blow away the residual moisture.

[0225] Step 8: Perform the side blowing stage. Based on the modeling data on both sides, control the rotation of the angle adjustment motor to ensure that the outlet of the side blowing component 83 always maintains a constant distance from the side of the vehicle, thus blowing away the moisture on the side.

[0226] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these are merely illustrative examples, and the scope of protection of the present invention is defined by the appended claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, but all such changes and modifications fall within the scope of protection of the present invention.

Claims

1. A car wash machine control method based on lidar modeling, characterized in that: The control method includes the following steps: S1. System initialization: Power-on calibration and lidar data initialization; establish global coordinate system reference and lidar calibration parameters; start lidar data acquisition; reset the actuator to the reference position. S2. Data acquisition: Control the relative displacement between the lidar and the vehicle along the longitudinal direction of the vehicle body. The lidar acquires polar coordinate point cloud data of the vehicle at a set frequency and records the position and timing information corresponding to each frame of point cloud. S3. Coordinate transformation: Convert the polar coordinates of each LiDAR point one by one, and transform them into the global coordinate system according to the installation calibration parameters of each LiDAR and the position timing information. S4. Profile generation: The point clouds belonging to different lidars after conversion are partitioned, filtered and merged according to preset rules to generate the lateral profile of the vehicle at the current longitudinal position. S5. Three-dimensional model construction: All transverse sections are stacked along the longitudinal direction of the vehicle body, and the whole vehicle three-dimensional point cloud model is generated and stored by interpolation fitting. S6. Adaptive cleaning: Reuse the three-dimensional point cloud model of the whole vehicle generated in step S5. During the cleaning process, retrieve the corresponding profile contour data according to the current longitudinal position, drive the cleaning actuator to adaptively follow the body contour movement, and complete the cleaning process.

2. The car wash machine control method based on lidar modeling as described in claim 1, characterized in that: In the global coordinate system, the X-axis corresponds to the width of the vehicle body, the Y-axis corresponds to the length of the vehicle body and is consistent with the relative displacement direction, and the Z-axis corresponds to the height of the vehicle body.

3. The car wash machine control method based on lidar modeling as described in claim 1, characterized in that: The lidar is installed at different positions on the cleaning actuator. The lidar has independent X-axis installation offset parameters, Y-axis timing compensation parameters, and unified Z-axis installation reference parameters.

4. The car wash machine control method based on lidar modeling as described in claim 1, characterized in that: In step S3, local coordinate transformation is performed on any valid point cloud data (L, θ); Local X-coordinate = L × Cos(θ); Local Z-coordinate = L×Sin(θ); Then, based on the X-axis offset, Y-axis timing compensation, and Z-axis installation reference height of each lidar, global coordinate correction is performed, and the validity of the converted coordinate points is filtered to remove invalid point clouds.

5. The car wash machine control method based on lidar modeling as described in claim 1, characterized in that: In step S4, different lidars collect point cloud data of different lateral side areas of the vehicle, and each lidar retains only the effective point cloud within its corresponding side area. The point cloud within the overlapping region is deduplicated, and the profile contour is smoothed using a smoothing filter.

6. The car wash machine control method based on lidar modeling as described in claim 1, characterized in that: In step S5, the three-dimensional point cloud model of the whole vehicle is established by a single relative displacement scan, and the modeling process and the cleaning process are separated in time. The interpolation fitting includes data completion for the gaps between adjacent profiles.

7. The car wash machine control method based on lidar modeling as described in claim 1, characterized in that: The cleaning actuators include reciprocating car wash machines and tunnel car wash machines; The reciprocating car wash machine includes a walking frame (1), a lifting cleaning module (2) and an opening and closing cleaning module (3), which are mounted on the walking frame (1); The tunnel-type car wash machine includes a conveyor belt (5), a radar module (6), a cleaning mechanism (7) and a blower mechanism (8). The radar module (6), the cleaning mechanism (7) and the blower mechanism (8) are arranged sequentially from left to right above the conveyor belt (5).

8. The car wash machine control method based on lidar modeling as described in claim 7, characterized in that: The walking frame (1) includes a support frame (11), a drive structure (12) and a walking track (13). The support frame (11) and the drive structure (12) are connected by the walking track (13). The drive structure (12) drives the support frame (11) to move along the walking track (13). The lifting cleaning module (2) and the opening and closing cleaning module (3) are mounted on the support frame (11), and a radar acquisition mechanism 4 is installed on the front side of the support frame 11.

9. The car wash machine control method based on lidar modeling as described in claim 8, characterized in that: The lifting cleaning module (2) includes a lifting assembly (21), which is installed on the support frame (11). The output execution end of the lifting assembly (21) is equipped with one or more of the top brush assembly (22), the top spray assembly (23), and the top air drying assembly (24). The opening and closing cleaning module (3) includes an opening and closing component (31), which is installed on the support frame (11). The output execution end of the opening and closing component (31) is equipped with one or more of the following: a side brush component (32), a side spray component (33), and a side air drying component (34).

10. The car wash machine control method based on lidar modeling as described in claim 7, characterized in that: The cleaning mechanism (7) includes a high-pressure water and foam assembly (71), a small vertical brush assembly (72), left and right vertical brush assemblies (74), a top brush assembly (76), and a wax spraying assembly (78). The small vertical brush assembly (72) is connected to the output end of the rotary drive assembly (73), the left and right vertical brush assemblies (74) are connected to the output end of the distance adjustment drive assembly (75), and the top brush assembly (76) is connected to the output end of the height adjustment drive assembly (77). The blower mechanism (8) includes a top blower assembly (81) and a side blower assembly (83). The top blower assembly (81) is connected to the output end of the blower height adjustment assembly (82), and the side blower assembly (83) is connected to the output end of the angle adjustment drive assembly (84).

Images