A modular autonomous cleaning machine control method and system suitable for high tilt angle photovoltaic modules

Abstract

The application relates to the technical field of autonomous mobile body control. A modular autonomous cleaning machine control method and system suitable for high-inclination photovoltaic modules are provided, wherein the method comprises the following steps: performing pollution feature analysis based on image features obtained by an image acquisition unit to obtain a pollution level division result; marking a pollution area according to the pollution level division result to generate a pollution distribution map; obtaining a current cleaning execution unit type through a preset cleaning unit recognition interface; performing cleaning parameter linkage adjustment according to the pollution distribution map, the cleaning execution unit type and a cleaning mode to obtain an adaptive cleaning strategy; controlling a preset differential motion unit to perform a cleaning action based on a cleaning path and the adaptive cleaning strategy to generate a task execution state; and uploading the task execution state to an external system through a preset state feedback unit, so that the moving stability, cleaning adaptability and safety decision capability of the robot are improved, the cleaning effect is optimized and energy consumption is reduced.

Description

Technical Field

[0001] This invention relates to the field of autonomous mobile body control technology, and in particular to a modular autonomous cleaning machine control method and system suitable for high-tilt photovoltaic modules. Background Technology

[0002] With the widespread application of photovoltaic power generation technology, the problem of decreased power generation efficiency caused by surface contamination of photovoltaic modules has become increasingly prominent. Especially in large-scale photovoltaic power plants, regular cleaning has become a key aspect of ensuring power generation efficiency. Due to their installation characteristics, high-tilt photovoltaic modules place higher demands on the mobility stability and path planning capabilities of cleaning equipment.

[0003] However, traditional mobile cleaning equipment lacks sophisticated posture perception and dynamic motion control technologies, making it difficult to operate stably on steeply tilted photovoltaic panels and prone to instability. Existing cleaning equipment does not employ a modular, switchable operating structure design, making it impossible to match cleaning methods to different types of pollutants, resulting in an imbalance between cleaning effectiveness and energy consumption. Furthermore, traditional cleaning equipment lacks intelligent edge risk identification, pollution status assessment, and autonomous path optimization technologies, making it impossible to coordinate cleaning coverage, efficiency, and energy consumption control while ensuring safety. Summary of the Invention

[0004] Therefore, it is necessary to provide a modular autonomous cleaning machine control method and system suitable for high-tilt photovoltaic modules to address the above-mentioned technical problems, so as to improve the robot's mobility stability, cleaning adaptability and safety decision-making ability, thereby optimizing the cleaning effect and reducing energy consumption.

[0005] In a first aspect, this application provides a modular autonomous cleaning machine control method suitable for high-tilt photovoltaic modules, the method comprising:

[0006] The robot receives external cleaning task instructions through a preset communication unit and parses the boundary of the area to be cleaned and the cleaning mode; it obtains the robot body size and edge safety distance through a preset structural parameter acquisition unit; it obtains photovoltaic panel image data through a preset image acquisition unit; it obtains depth data through a preset depth vision unit; and based on the boundary of the area to be cleaned, photovoltaic panel image data, and depth data, it identifies passable areas and generates a local environmental map containing safe passable areas.

[0007] Based on the local environment map, robot body size, and edge safety distance, path boundary calculation is performed to generate a safe passage area; based on the safe passage area and cleaning mode, cleaning path generation processing is performed to obtain a cleaning path with applied tilt angle safety constraints;

[0008] Based on the image features of the photovoltaic panel obtained by the image acquisition unit, pollution feature analysis is performed to obtain pollution level classification results; according to the pollution level classification results, pollution areas are marked to generate a pollution distribution map;

[0009] The current cleaning execution unit type is obtained through a preset cleaning unit identification interface; based on the pollution distribution map, the cleaning execution unit type, and the cleaning mode, the cleaning parameters are adjusted in a coordinated manner to obtain an adaptive cleaning strategy;

[0010] Based on the cleaning path and adaptive cleaning strategy, the system controls the preset differential motion unit to perform cleaning actions and generates a task execution status; the task execution status is then uploaded to an external system through a preset status feedback unit.

[0011] In one embodiment, a cleaning path generation process is performed based on the safe passage area and the cleaning mode to obtain a cleaning path with applied tilt angle safety constraints, including:

[0012] Based on the cleaning mode, determine the path coverage strategy, which includes a reciprocating coverage strategy, a zoned coverage strategy, or a localized enhanced cleaning strategy.

[0013] Based on the path coverage strategy and safe passage area, the initial path generation process is performed to obtain the initial clean path;

[0014] Based on the preset tilt angle safety constraint rules and the robot body size, the turning and edge movement actions in the initial cleaning path are optimized to generate a cleaning path with tilt angle safety constraints.

[0015] In one embodiment, based on preset tilt angle safety constraint rules and robot body dimensions, the turning and edge movement actions in the initial cleaning path are optimized to generate a cleaning path with applied tilt angle safety constraints, including:

[0016] Based on the robot's body dimensions, the turning safety margin and edge movement safety margin are calculated. The turning safety margin is used to limit actions with a turning radius smaller than a safety threshold; the edge movement safety margin is used to limit lateral movement actions within the edge risk area.

[0017] Based on the tilt angle safety constraint rules and the turning safety margin, the turning radius of the turning action in the initial cleaning path is optimized to obtain the optimized turning path segment;

[0018] Based on the tilt angle safety constraint rules and the edge movement safety margin, the edge movement action in the initial cleaning path is optimized by edge distance to obtain the optimized edge path segment;

[0019] The optimized turning path segment, the optimized edge path segment, and the regular path segment in the initial cleaning path are merged to generate a cleaning path with tilt angle safety constraints.

[0020] In one embodiment, an adaptive cleaning strategy is obtained by adjusting cleaning parameters in conjunction with the pollution distribution map, the type of cleaning execution unit, and the cleaning mode, including:

[0021] Based on the pollution distribution map, pollution areas are classified into first-level pollution areas and second-level pollution areas. The first-level pollution areas correspond to areas with pollution scores exceeding a set threshold, while the second-level pollution areas correspond to areas with pollution scores below the set threshold.

[0022] Based on the type of cleaning execution unit, parameter rules are matched to obtain cleaning parameter adjustment rules;

[0023] Based on the area marking information of the first pollution level, the cleaning intensity is calculated and a first-class cleaning intensity instruction is generated.

[0024] Based on the area marking information of the second pollution level, the cleaning intensity is calculated to generate a second type of cleaning intensity instruction;

[0025] Based on the cleaning mode, the cleaning instructions are modified to obtain the modified first type of cleaning intensity instructions and the modified second type of cleaning intensity instructions;

[0026] The modified first type of cleaning intensity command, the modified second type of cleaning intensity command, and the cleaning parameter adjustment rules are integrated and processed to generate an adaptive cleaning strategy.

[0027] In one embodiment, based on a cleaning path and an adaptive cleaning strategy, a preset differential motion unit is controlled to perform cleaning actions, generating a task execution state, including:

[0028] Based on the cleaning path, the motion command is decomposed to generate the target linear velocity, target angular velocity, and target velocities of the left and right tracks;

[0029] Based on the adaptive cleaning strategy, the cleaning action is analyzed to obtain the roller brush rotation speed command, water spray volume command, and movement speed correction command.

[0030] Based on the target linear velocity, target angular velocity, target speed of left and right tracks, roller brush rotation speed command, water spray volume command and movement speed correction command, coordinated control is performed to drive the differential motion unit to complete the cleaning action;

[0031] Real-time collection of cleaning area coverage data, abnormal triggering events, and robot status parameters;

[0032] Based on the cleaning area coverage data, abnormal triggering events, and robot state parameters, the task state is synthesized to generate the task execution state.

[0033] In one embodiment, based on a local environment map, robot body dimensions, and edge safety distance, path boundary calculation is performed to generate a safe passage area, including:

[0034] Based on the local environment map, extract the boundaries of photovoltaic modules and the boundaries of passable areas;

[0035] Calculate the robot's safety envelope area based on the robot's body size and edge safety distance;

[0036] Within the boundaries of the photovoltaic modules, path boundary calculations are performed based on the traversable area boundary and the robot's safety envelope area to obtain the safe traversable area.

[0037] Secondly, this application also provides a modular autonomous cleaning machine control system suitable for high-tilt photovoltaic modules, the system comprising:

[0038] The task parsing module is used to receive external cleaning task instructions through a preset communication unit and parse the boundary of the area to be cleaned and the cleaning mode; obtain the robot body size and edge safety distance through a preset structural parameter acquisition unit; obtain photovoltaic panel image data through a preset image acquisition unit; obtain depth data through a preset depth vision unit; and identify passable areas based on the boundary of the area to be cleaned, photovoltaic panel image data and depth data to generate a local environment map containing safe passable areas.

[0039] The path planning module is used to perform path boundary calculations based on the local environment map, robot body size, and edge safety distance to generate a safe passage area; based on the safe passage area and cleaning mode, it performs cleaning path generation processing to obtain a cleaning path with applied tilt angle safety constraints;

[0040] The pollution assessment module is used to perform pollution feature analysis based on the photovoltaic panel image features acquired by the image acquisition unit, and obtain pollution level classification results; based on the pollution level classification results, the pollution area is marked and a pollution distribution map is generated.

[0041] The strategy generation module is used to obtain the current cleaning execution unit type through a preset cleaning unit identification interface; and to adjust the cleaning parameters in conjunction with the pollution distribution map, the cleaning execution unit type, and the cleaning mode to obtain an adaptive cleaning strategy.

[0042] The execution control module is used to control the preset differential motion unit to perform cleaning actions based on the cleaning path and adaptive cleaning strategy, and generate the task execution status; the task execution status is uploaded to the external system through the preset status feedback unit.

[0043] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods in the first aspect of this application.

[0044] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the methods in the first aspect of this application.

[0045] This application provides a modular autonomous cleaning robot control method and system applicable to high-tilt photovoltaic modules. The method receives and parses external cleaning task instructions, simultaneously acquires robot body dimensions, edge safety distances, photovoltaic panel images, and depth data, identifies passable areas, and constructs a local environmental map containing safe passable areas. Then, based on the local environmental map, robot body dimensions, and safety distances, the safe passable areas are delineated, and a cleaning path with tilt angle safety constraints is generated. This helps improve the robot's movement stability when operating on high-tilt photovoltaic panels and provides a reliable path basis for safety decisions.

[0046] This method analyzes the image features of photovoltaic panels to classify pollution levels and mark polluted areas. It also identifies the current cleaning execution unit type and adjusts cleaning parameters in conjunction with the pollution distribution map, unit type, and cleaning mode. This improves the robot's adaptability to cleaning different types of pollution. By relying on the planned cleaning path and adaptive cleaning strategy to control the differential motion unit to complete the cleaning action and upload the task execution status, it further enhances the safety decision-making capability, thereby optimizing the cleaning effect of photovoltaic modules and reducing the energy consumption of the operation. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating a modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to one embodiment of the present invention;

[0049] Figure 2 This is a flowchart of a process in one embodiment of the present invention for performing path boundary calculation and generating a safe passage area based on a local environment map, robot body size and edge safety distance;

[0050] Figure 3This is a structural diagram of a modular autonomous cleaning machine control system for high-tilt photovoltaic modules, according to one embodiment of the present invention. Detailed Implementation

[0051] To make the above-mentioned objects, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0052] First, the application scenarios of the embodiments of this application are described. In the embodiments of this application, a modular autonomous cleaning machine control method is provided that is applicable to, but not limited to, scenarios such as cleaning high-tilt photovoltaic modules, maintenance of tilted solar panels, precision cleaning of outdoor panels on slopes, and maintenance of tilted energy harvesting equipment.

[0053] In illustrative purposes, the modular autonomous cleaning machine control method provided in this application embodiment can also be applied to other application scenarios such as automated slope cleaning, intelligent operation and maintenance of inclined surfaces, maintenance of large outdoor inclined facilities, and autonomous operation of special inclined slabs. This is only an example and does not limit the specific application scenarios.

[0054] like Figure 1 As shown, this application provides a modular autonomous cleaning machine control method suitable for high-tilt photovoltaic modules, the method comprising:

[0055] S101: Receive external cleaning task instructions through a preset communication unit and parse the boundary of the area to be cleaned and the cleaning mode; obtain the robot body size and edge safety distance through a preset structural parameter acquisition unit; obtain photovoltaic panel image data through a preset image acquisition unit; obtain depth data through a preset depth vision unit; and identify passable areas based on the boundary of the area to be cleaned, photovoltaic panel image data, and depth data to generate a local environment map containing safe passable areas.

[0056] For example, based on a preset communication unit, the cleaning robot control terminal receives external cleaning task instructions, performs integrity verification and protocol parsing on the instructions, and extracts the boundary of the area to be cleaned and the cleaning mode from the instruction data after the verification and parsing are completed.

[0057] Based on the preset structural parameter acquisition unit, the cleaning robot control terminal collects the original data of the robot body size and the original data of the edge safety distance. It performs unit uniform calibration and measurement error correction on the size data, and performs benchmark alignment calibration and threshold compliance verification on the safety distance data. The calibrated parameters are stored in the parameter cache unit and a benchmark parameter index is established.

[0058] Using a pre-set image acquisition unit, the cleaning robot control terminal acquires raw image data of the photovoltaic panel surface, and sequentially performs noise filtering, distortion correction, feature extraction and format standardization to obtain standardized photovoltaic panel surface image feature data; simultaneously, it acquires raw spatial depth data through a pre-set depth vision unit, performs outlier removal, coordinate registration and scale normalization to obtain calibrated spatial depth feature data, and unifies the spatial reference of the two types of data.

[0059] Using multi-source processed data as input, the cleaning robot control terminal invokes a passable area recognition algorithm to complete feature fusion, boundary fitting, and safety threshold determination calculations. Based on the recognition results of the passable area set, the cleaning robot control terminal performs spatial topology association construction and safety constraint fusion processing to generate a local environmental map containing safe passable areas.

[0060] S102: Based on the local environment map, robot body size and edge safety distance, perform path boundary calculation to generate a safe passage area; based on the safe passage area and cleaning mode, perform cleaning path generation processing to obtain a cleaning path with applied tilt angle safety constraints.

[0061] For example, relying on a local environment map, the cleaning robot control terminal analyzes the spatial topology information and regional boundary features contained therein, matches the space occupancy range corresponding to the robot body size, defines the boundary buffer range in combination with the edge safety distance, eliminates spatial areas that exceed the buffer range, selects valid areas that meet the body space occupancy conditions, performs path boundary calculation, and generates a safe passage area.

[0062] For the generated safe passage area, the cleaning robot control terminal identifies its spatial distribution characteristics and regional connectivity attributes, analyzes the cleaning coverage requirements and operation methods corresponding to the cleaning mode, matches the motion restriction rules corresponding to the tilt angle safety constraints, plans the path direction within the area according to the constraint rules, adjusts the path nodes to avoid high-risk motion actions, performs cleaning path generation processing, and obtains a cleaning path with tilt angle safety constraints applied.

[0063] S103: Based on the image features of the photovoltaic panel obtained by the image acquisition unit, perform pollution feature analysis to obtain pollution level classification results; based on the pollution level classification results, mark the pollution areas and generate a pollution distribution map.

[0064] For example, based on the image features of the photovoltaic panel obtained by the image acquisition unit, the cleaning robot control terminal extracts texture features, grayscale distribution features, edge gradient features, and color entropy features pixel by pixel. Spatial alignment, redundant feature removal, and feature dimension normalization are performed on the extracted multi-dimensional features to construct a standardized high-dimensional pollution feature matrix. This feature matrix is ​​then input into a nonlinear pollution level analysis model. Pollution level quantification is completed through multi-feature weighted fusion and entropy constraint correction. The core determination process uses a nonlinear weighted fusion formula:

[0065]

[0066] In the formula, This represents the quantified value of the pollution level. Represents a non-linear activation function; This represents the total number of pollution feature dimensions; Indicates the first Dynamic weighting coefficients for pollution characteristics; Indicates the first Nonlinear transformation function for dimensional pollution characteristics; Represents a standardized high-dimensional pollution feature matrix; This represents the entropy constraint penalty coefficient; The information entropy function represents the pollution feature matrix.

[0067] The cleaning robot control terminal classifies pollution levels based on the quantified pollution level values, thus obtaining the pollution level classification results.

[0068] Based on the established pollution level classification, the cleaning robot control terminal spatially maps the level quantification value to the pixel coordinates of the photovoltaic panel image, uses an adaptive threshold segmentation algorithm to delineate the spatial boundaries corresponding to each pollution level, performs connected component clustering and edge smoothing optimization on the closed boundary regions, assigns exclusive level labels to regions with different pollution levels, completes the marking of the pollution areas on the entire panel surface, and generates a pollution distribution map.

[0069] S104: Obtain the current cleaning execution unit type through the preset cleaning unit identification interface; adjust the cleaning parameters in conjunction with the pollution distribution map, cleaning execution unit type and cleaning mode to obtain an adaptive cleaning strategy.

[0070] For example, through a preset cleaning unit identification interface, the cleaning robot control terminal sends a type query command to the cleaning execution unit, receives the attribute identification information fed back by the cleaning execution unit, performs format parsing and feature matching processing on the attribute identification information, extracts the core functional parameters and operation adaptation characteristics of the cleaning execution unit, and determines the current cleaning execution unit type.

[0071] The cleaning robot control terminal analyzes the pollution distribution map, including the distribution range of polluted areas, differences in pollution levels, and the degree of regional concentration. It clarifies the operational capability boundaries, adaptation parameter ranges, and operational constraints corresponding to the cleaning execution unit type. Simultaneously, it breaks down the operational process requirements, coverage standards, and efficiency targets specified by the cleaning mode, establishes a parameter linkage mapping relationship among the three, matches the corresponding cleaning intensity, operation speed, and coverage frequency according to the characteristics of the polluted area, adjusts the operation mode adaptation parameters in combination with the cleaning execution unit type, optimizes the operation sequence and process connection according to the cleaning mode, and completes the linkage adjustment of cleaning parameters through multi-dimensional parameter collaborative adaptation to obtain an adaptive cleaning strategy.

[0072] S105: Based on the cleaning path and adaptive cleaning strategy, control the preset differential motion unit to perform cleaning actions and generate task execution status; upload the task execution status to the external system through the preset status feedback unit.

[0073] For example, by combining the cleaning path and the adaptive cleaning strategy, the cleaning robot control terminal analyzes the spatial nodes and direction planning of the cleaning path, decomposes the operating parameters such as cleaning intensity and operation speed included in the adaptive cleaning strategy, integrates the two types of information into motion control commands that can be recognized by the preset differential motion unit, sends motion control commands to the preset differential motion unit, drives the preset differential motion unit to execute the corresponding cleaning action according to the command sequence, collects the operating status data in real time during the execution of the cleaning action, and integrates the operating status data to form the task execution status.

[0074] Based on the preset status feedback unit, the cleaning robot control terminal performs format standardization processing on the generated task execution status to ensure that the task execution status meets the information reception specifications of the external system. A communication link with the external system is established through the preset status feedback unit, and the standardized task execution status is uploaded to the external system through the communication link to complete the feedback transmission of the task execution status.

[0075] One embodiment of this application provides a modular autonomous cleaning robot control method applicable to high-tilt photovoltaic modules. By receiving and parsing external cleaning task instructions, it simultaneously acquires robot body dimensions, edge safety distances, photovoltaic panel images, and depth data, identifies passable areas, and constructs a local environmental map containing safe passable areas. Then, based on the local environmental map, robot body dimensions, and safety distances, it delineates safe passable areas and generates a cleaning path with tilt angle safety constraints. This helps improve the robot's movement stability when operating on high-tilt photovoltaic panels and provides a reliable path basis for safety decisions.

[0076] This method analyzes the image features of photovoltaic panels to classify pollution levels and mark polluted areas. It also identifies the current cleaning execution unit type and adjusts cleaning parameters in conjunction with the pollution distribution map, unit type, and cleaning mode. This improves the robot's adaptability to cleaning different types of pollution. By relying on the planned cleaning path and adaptive cleaning strategy to control the differential motion unit to complete the cleaning action and upload the task execution status, it further enhances the safety decision-making capability, thereby optimizing the cleaning effect of photovoltaic modules and reducing the energy consumption of the operation.

[0077] In one embodiment, a cleaning path generation process is performed based on the safe passage area and the cleaning mode to obtain a cleaning path with applied tilt angle safety constraints, including:

[0078] (1) Determine the path coverage strategy based on the cleaning mode, where the path coverage strategy includes reciprocating coverage strategy, partitioned coverage strategy or local enhanced cleaning strategy.

[0079] For example, for the core operational attributes of the cleaning mode, the cleaning robot control terminal analyzes the operational coverage requirements, area cleaning priorities and operational efficiency targets contained therein, extracts the core operational features corresponding to the cleaning mode one by one, and clarifies the attributes and quantification direction of each core operational feature.

[0080] Based on the extracted core operational features, the cleaning robot control terminal compares the features with a pre-defined path coverage strategy feature library dimension by dimension. The feature fit of each dimension is quantitatively evaluated, and the appropriate path coverage strategy is selected based on the comprehensive result of the fit of each dimension. The feature fit evaluation formula is as follows:

[0081]

[0082] In the formula, Indicates the core operation characteristics and the first The overall suitability of the class path coverage strategy; This represents the total number of dimensions for feature comparison; Indicates the first The comparison weight coefficients of dimensional features; The first part represents the core operational characteristics of the cleaning mode. Dimensional quantization value; Indicates the first The first in the class path coverage strategy feature library Standard quantization value of dimensional features.

[0083] The core operational characteristics of the cleaning mode include the requirement for comprehensive coverage, the criteria for prioritizing area cleaning, and the quantitative standards for operational efficiency targets. The path coverage strategy feature library contains the coverage logic features, area adaptation features, and efficiency matching features corresponding to various path coverage strategies.

[0084] (2) Based on the path coverage strategy and safe passage area, perform initial path generation process to obtain the initial clean path.

[0085] For example, by breaking down the determined path coverage strategy, the cleaning robot control terminal clarifies the path planning logic, area traversal order, coverage spacing standards, and coverage integrity requirements contained therein, and clarifies the core rules and parameter constraints for strategy execution.

[0086] Based on the spatial characteristics of the safe passage area, the cleaning robot control terminal analyzes its outer boundary range, internal segmentation boundary, regional connectivity attributes, and effective sub-region division results. It then integrates and adapts the execution rules of the path coverage strategy with the spatial characteristics of the safe passage area in a layered manner. The coordinates of the starting node, intermediate nodes, and ending node of the path are planned according to the priority order of regional traversal. The path trajectories between nodes are then smoothly connected according to the coverage spacing standard. Finally, path completion planning is performed for the coverage blind spots in the safe passage area to obtain the initial cleaning path.

[0087] The core execution rules of the path coverage strategy include the priority setting of path planning logic, the arrangement criteria of region traversal order, the adaptation range of coverage spacing, and the judgment conditions for coverage integrity. The spatial characteristics of the safe passage area include the geometric shape of the outer boundary, the distribution of the internal dividing boundary, the judgment result of the region connectivity attribute, and the division dimensions of the effective operation sub-region.

[0088] (3) Based on the preset tilt angle safety constraint rules and robot body size, optimize the turning and edge movement in the initial cleaning path to generate a cleaning path with tilt angle safety constraints.

[0089] For example, the cleaning robot control terminal retrieves a preset tilt angle safety constraint rule file, analyzes the maximum allowable turning angle, tilt surface motion stability constraint threshold, minimum safety distance for edge operations, and upper limit of path slope change rate contained therein, and clarifies the specific execution requirements of each constraint.

[0090] Key parameters of the robot's dimensions are extracted. The robot control terminal acquires information on the robot's length, width, and the extension range of the operating mechanism. The actual turning angle of each turning motion node in the initial cleaning path is calculated, and the calculation results are compared and verified with the maximum allowable turning angle. The shortest distance to the boundary of the safe passage area is measured for each edge movement segment in the initial cleaning path, and the measurement results are analyzed for compatibility with the minimum safe distance for edge operations and the sum of the robot's half-width. The slope change rate of the overall trajectory of the initial cleaning path is calculated, and the calculation results are verified for compliance with the inclined surface motion stability constraint threshold.

[0091] For non-constrained motion nodes and path segments in the initial cleaning path, the cleaning robot control terminal corrects the turning angle and smooths the trajectory for non-compliant turning motion nodes; adjusts the path offset and compensates for the safety distance for edge movement motion segments that do not meet the safety distance standard; and replans the trajectory and smooths the slope for path segments with excessive slope change rate. The core optimization process adopts a multi-constraint multi-objective optimization model.

[0092]

[0093]

[0094]

[0095]

[0096] In the formula, This represents the final cleaning path after optimizing the tilt angle safety constraints. This represents the set of paths to be optimized that participate in the iterative computation. This represents the initial cleaning path that has not been adjusted for constraints. The weighting coefficients represent the target weights for smoothing path turns. This represents an evaluation index for the smoothness of the turning trajectory of a path. The weighting coefficients represent the target weights for optimizing safety at the edge of the path. Indicators representing the safety level of operations at the edge of the path; The weight coefficients representing the target weights for path slope adaptation optimization; Evaluation indicators for the compatibility between the representative path and the inclined working surface; This represents the actual turning angle of a single turning node within the path. Represents an independent path node within a path; Represents the set of all path nodes that involve turning. Represents the straight-line distance from the edge path node to the boundary of the safe passage area; Represents edge-like path nodes within a path; The outline of the safe passage area; This represents the overall width of the cleaning robot. Represents the entire set of edge-type path segments; This represents the actual slope change rate of a single path segment. This represents a path segment with varying slope within the path. This represents the set of all path segments with varying slopes.

[0097] Based on the aforementioned multi-constraint, multi-objective optimization model, the cleaning robot control terminal completes the full-dimensional optimization and adjustment of the initial cleaning path, generating a cleaning path with applied tilt angle safety constraints.

[0098] The specific constraints of the tilt angle safety constraint rules include the numerical definition of the maximum allowable turning angle, the judgment criteria for the threshold of tilt surface motion stability constraint, the basis for setting the minimum safety distance for edge operations, and the quantitative requirements for the upper limit of the path slope change rate. Key parameters of the robot's body dimensions include the measurement dimension of the body length, the reference range of the body width, and the maximum limit of the extension range of the operating mechanism. Specific methods for path optimization include the adjustment range standard for turning angle correction, the algorithm logic for trajectory smoothing optimization, the distance range for path offset adjustment, the implementation method for safety distance compensation, and the path reconstruction rules for slope smoothing processing.

[0099] In one embodiment, based on preset tilt angle safety constraint rules and robot body dimensions, the turning and edge movement actions in the initial cleaning path are optimized to generate a cleaning path with applied tilt angle safety constraints, including:

[0100] (1) Calculate the turning safety margin and edge movement safety margin based on the robot body size. The turning safety margin is used to limit the turning radius to less than the safety threshold. The edge movement safety margin is used to limit the lateral movement within the edge risk area.

[0101] For example, the cleaning robot control terminal extracts parameters from the robot's dimensions, including body length, body width, wheelbase, and turning radius of the work execution mechanism. Combining these with the basic safety requirements of tilt angle safety constraints, it calculates the minimum safe space required for turning movements using a geometric kinematics model, deriving the turning safety margin. Simultaneously, based on the robot's body width and the buffering requirements for edge operations, it quantifies the safe distance threshold between the edge region and the robot body, obtaining the edge movement safety margin. The core formula for calculating the safety margin is as follows:

[0102]

[0103]

[0104] In the formula, Indicates the safety margin for turning; This indicates the minimum permissible turning radius specified in the tilt angle safety constraint rules; Indicates the length of the robot's body; Indicates the width of the robot's body; This indicates the basic safety buffer amount corresponding to the tilt angle; Indicates the safety margin for edge movement; This indicates the basic edge safety distance specified in the tilt angle safety constraint rules; This represents the dynamic buffer size during robot operation.

[0105] Among them, the turning safety margin is the amount of space buffer that ensures the robot does not tip over or collide when turning on an inclined surface; the edge movement safety margin is the distance threshold that limits the robot from exceeding the safe range when moving at the edge of a safe passage area.

[0106] (2) Based on the tilt angle safety constraint rules and the turning safety margin, the turning radius of the turning action in the initial cleaning path is optimized to obtain the optimized turning path segment.

[0107] For example, the cleaning robot control terminal analyzes the maximum allowable tilt angle, turning speed limit, and turning radius threshold requirements of the tilt angle safety constraint rules related to turning actions. It extracts the turning radius, turning angle, and turning speed parameters corresponding to all turning action nodes in the initial cleaning path. The extracted parameters are compared and analyzed with the turning safety margin. For action nodes with a turning radius less than the safety threshold, the curvature of the turning path is adjusted through a trajectory interpolation algorithm to expand the turning radius to a safe range, thereby generating an optimized turning path segment.

[0108] Among them, the optimized turning path segment is the part of the path corresponding to the turning action after radius adjustment, which meets the requirements of tilt angle safety constraints and turning safety margin.

[0109] (3) Based on the tilt angle safety constraint rules and the edge movement safety margin, the edge movement action in the initial cleaning path is optimized by edge distance to obtain the optimized edge path segment.

[0110] For example, the cleaning robot control terminal parses the edge operation tilt angle limit and minimum safety distance requirement related to edge movement in the tilt angle safety constraint rules, extracts the actual distance parameters between the path and the boundary of the safe passage area corresponding to all edge movement segments in the initial cleaning path, performs an adaptation check between the actual distance and the edge movement safety margin, and corrects the path position through a translation adjustment algorithm for path segments whose actual distance is less than the edge movement safety margin, increasing the distance between the path and the area boundary to a safe range, and generating optimized edge path segments.

[0111] Among them, the optimized edge path segment is the path part corresponding to the edge movement action after distance adjustment, which meets the requirements of tilt angle safety constraints and edge movement safety margin.

[0112] (4) The optimized turning path segment, the optimized edge path segment and the regular path segment in the initial cleaning path are merged to generate a cleaning path with tilt angle safety constraints.

[0113] For example, the cleaning robot control terminal identifies regular path segments in the initial cleaning path that do not involve turning or edge movement, extracts the spatial coordinates, path direction, and length information of the regular path segments, aligns the coordinates of the connection nodes of the optimized turning path segments and the regular path segments, and processes the trajectory curvature changes at the connection points through a smooth transition algorithm; at the same time, it performs endpoint matching between the optimized edge path segments and the regular path segments and the optimized turning path segments, supplements the transition path to ensure the continuity of the overall path, and generates a cleaning path with tilt angle safety constraints after merging all path segments.

[0114] Among them, the cleaning path with tilt angle safety constraints is a complete operation path that integrates and optimizes turning path segments, optimized edge path segments, and regular path segments, and meets the tilt angle safety constraint rules and robot body size adaptation requirements throughout.

[0115] In one embodiment, an adaptive cleaning strategy is obtained by adjusting cleaning parameters in conjunction with the pollution distribution map, the type of cleaning execution unit, and the cleaning mode, including:

[0116] (1) Based on the pollution distribution map, the pollution area is divided into levels to obtain the first pollution level area marking information and the second pollution level area marking information. The first pollution level area corresponds to the area where the pollution degree score exceeds the set threshold; the second pollution level area corresponds to the area where the pollution degree score is lower than the set threshold.

[0117] For example, by parsing the core information of the pollution distribution map, the cleaning robot control terminal extracts the pollution concentration, pollution coverage area, and pollution aggregation density features of each area on the photovoltaic panel, normalizes and quantifies each feature, calculates the pollution level score of each area through multi-feature weighted fusion, compares the pollution level score with the preset pollution level classification threshold one by one, marks the areas with pollution level scores exceeding the set threshold, and generates the first pollution level area label information; at the same time, it marks the areas with pollution level scores below the set threshold and generates the second pollution level area label information.

[0118] Among them, the pollution level score is a quantitative indicator that comprehensively reflects the severity of regional pollution; the first pollution level area labeling information is the identification data of the spatial location, range and pollution characteristics of high pollution areas; the second pollution level area labeling information is the identification data of the spatial location, range and pollution characteristics of low pollution areas.

[0119] (2) Based on the type of cleaning execution unit, perform parameter rule matching to obtain the cleaning parameter adjustment rules.

[0120] For example, key attributes of the cleaning execution unit are extracted. The cleaning robot control terminal obtains the cleaning method, power adjustment range, upper limit of work efficiency, and appropriate pollution type parameters corresponding to the cleaning execution unit type. These parameters are then matched against a preset cleaning parameter rule library in all dimensions to filter out parameter adjustment criteria, parameter value ranges, and parameter linkage logic that are highly compatible with the current cleaning execution unit type, thus obtaining the cleaning parameter adjustment rules. The formula for calculating the parameter rule matching degree is as follows:

[0121]

[0122] In the formula, This indicates the degree of matching between the cleaning execution unit parameters and the rule base; Indicates the total number of dimensions for parameter matching; Indicates the first Matching weight coefficients for dimensional parameters; The first cleaning execution unit Dimensional actual parameters; Indicates the first rule in the rule base Standard parameters; This represents the parameter matching degree function (returns 1 for a complete match, the matching ratio for a partial match, and 0 for no match).

[0123] Among them, the cleaning parameter adjustment rules are a set of criteria that guide the dynamic adaptation of cleaning parameters, including the adjustment standards and linkage relationships of parameters such as cleaning intensity, operation speed, and power distribution; the cleaning parameter rule base is a database that stores the parameter adjustment logic corresponding to various cleaning execution units.

[0124] (3) Based on the first pollution level area marking information, perform cleaning intensity calculation and generate the first type of cleaning intensity instruction.

[0125] For example, focusing on the marking information of the first pollution level area, the cleaning robot control terminal analyzes the spatial range, peak pollution concentration, and pollution accumulation area distribution of the first pollution level area. Combining the deep cleaning requirements of the cleaning operation, and based on the mapping relationship between pollution level and cleaning intensity, it calculates the required basic cleaning intensity for the area. Considering the maximum operating capacity of the cleaning execution unit, it performs an intensity upper limit check. Then, it optimizes the spatial distribution of cleaning intensity using a gradient enhancement algorithm to generate a first-type cleaning intensity instruction.

[0126]

[0127] In the formula, This indicates the core strength value of the first type of cleaning intensity directive; This represents the baseline value for the basic intensity of cleaning operations; This represents the intensity adjustment factor corresponding to the pollution level score; This represents the average pollution level score for areas classified as Level 1 pollution. This represents the intensity adjustment factor corresponding to the peak pollution concentration. This represents the peak pollution concentration quantification value of the first pollution level area.

[0128] The first type of cleaning intensity instruction is an operation control instruction with high cleaning intensity formulated for highly polluted areas, which includes parameters such as cleaning power, operation frequency, and cleaning duration.

[0129] (4) Based on the area marking information of the second pollution level, perform cleaning intensity calculation and generate the second type of cleaning intensity instruction.

[0130] For example, based on the second pollution level area marking information, the cleaning robot control terminal analyzes the spatial range, pollution uniformity, and light pollution distribution characteristics of the second pollution level area, combines the routine cleaning requirements of the cleaning operation, calculates the basic cleaning intensity required for the area according to the adaptation relationship between pollution degree and cleaning intensity, performs intensity lower limit verification with reference to the energy-saving operation parameters of the cleaning execution unit, adjusts the uniformity of cleaning intensity through a smoothing optimization algorithm, and generates a second type of cleaning intensity instruction.

[0131] The second type of cleaning intensity instruction is an operation control instruction formulated for low-pollution areas that takes into account both cleaning effectiveness and energy-saving requirements. It includes parameters such as cleaning power, operation frequency, and cleaning duration.

[0132] (5) Based on the cleaning mode, the cleaning instructions are modified to obtain the modified first type of cleaning intensity instructions and the modified second type of cleaning intensity instructions.

[0133] For example, the cleaning robot control terminal interprets the operation requirements of the cleaning mode, extracts the operation priority, energy-saving target, efficiency standard and cleaning quality requirements included in the cleaning mode, analyzes the constraints of the above requirements on the cleaning intensity, compares and adapts the first type of cleaning intensity command with the constraints of the cleaning mode, corrects and adjusts the cleaning intensity parameters that exceed the constraints, and obtains the corrected first type of cleaning intensity command; at the same time, the second type of cleaning intensity command is adapted and verified with the constraints of the cleaning mode, and parameters that do not meet the requirements are corrected, and the corrected second type of cleaning intensity command is obtained.

[0134] (6) The modified first type of cleaning intensity command, the modified second type of cleaning intensity command and the cleaning parameter adjustment rules are integrated and the strategy generation process is performed to obtain an adaptive cleaning strategy.

[0135] For example, by integrating the two types of cleaning intensity commands and cleaning parameter adjustment rules, the cleaning robot control terminal merges the modified first type of cleaning intensity command and the modified second type of cleaning intensity command with the parameter adaptation logic in the cleaning parameter adjustment rules, matches the power adjustment range and operation speed adaptation interval corresponding to the cleaning execution unit type, optimizes the linkage relationship between cleaning intensity and operation parameters, supplements the operation transition parameters between different areas, constructs a cleaning strategy framework covering the entire photovoltaic panel, performs strategy generation processing, and obtains an adaptive cleaning strategy.

[0136] Among them, the adaptive cleaning strategy is a complete operation plan that takes into account the distribution of pollution, the capabilities of the cleaning execution unit and the requirements of the cleaning mode, and can dynamically adapt to different polluted areas. It includes core information such as cleaning parameters, operation sequence and transition method for each area.

[0137] In one embodiment, based on a cleaning path and an adaptive cleaning strategy, a preset differential motion unit is controlled to perform cleaning actions, generating a task execution state, including:

[0138] (1) Based on the cleaning path, decompose the motion command to generate the target linear velocity, target angular velocity and left and right track target velocities.

[0139] For example, the cleaning robot control terminal reads the path node coordinates, travel trajectory, path curvature characteristics, and turning position information stored inside the cleaning path. It then decomposes the overall travel motion behavior according to the differential transmission motion logic of the preset differential motion unit, distinguishes the different motion control logic corresponding to the straight road segment and the turning road segment, solves the target linear velocity corresponding to the overall travel based on the planar kinematics conversion relationship, calculates the target angular velocity in combination with the path deflection trend, completes the velocity component allocation according to the dual track differential ratio rule, determines the running speed of the left track and the running speed of the right track, and completes the decomposition of motion commands.

[0140] Among them, the target linear velocity represents the baseline translational velocity of the robot's overall movement, the target angular velocity represents the attitude deflection velocity of the robot during movement, and the target velocities of the left and right tracks represent the independent operating speeds of the drive structures on both sides of the differential motion unit.

[0141] (2) Based on the adaptive cleaning strategy, perform cleaning action analysis to obtain the roller brush speed command, water spray volume command and moving speed correction command.

[0142] For example, the cleaning robot control terminal retrieves the pre-compiled adaptive cleaning strategy, analyzes the zoning cleaning operation standards, operation intensity parameters corresponding to different pollution levels, and slope operation travel restrictions contained in the adaptive cleaning strategy, divides the execution boundaries of mechanical roller brush cleaning operations and spray cleaning operations, matches the roller brush rotation operation standards corresponding to high pollution areas and low pollution areas, organizes them into roller brush speed instructions, determines the fluid spray control standards based on the dry and wet cleaning needs of the board surface, organizes them into water spray volume instructions, and sorts out the speed offset that needs to be fine-tuned during the travel process based on the changes in travel resistance caused by the working environment resistance and the inclination angle of the board surface, and organizes them into movement speed correction instructions.

[0143] Among them, the roller brush speed command is used to control the rotation speed of the mechanical cleaning components, the water spray volume command is used to control the total amount of fluid sprayed by the wet cleaning components, and the movement speed correction command is used to fine-tune the basic travel speed according to the environment.

[0144] (3) Based on the target linear velocity, target angular velocity, target speed of left and right tracks, roller brush rotation speed command, water spray volume command and movement speed correction command, perform coordinated control to drive the differential motion unit to complete the cleaning action.

[0145] For example, the cleaning robot control terminal aggregates all movement control parameters and cleaning operation control commands, applies the movement speed correction command to the target linear velocity to complete the basic movement speed correction processing, uniformly calibrates the matching relationship between the target angular velocity and the target speed of the left and right tracks, establishes the timing linkage relationship between the movement motion control logic and the on-site cleaning operation control logic, and synchronously sends all integrated control signals to the preset differential motion unit according to the preset operation execution sequence of the cleaning path, and drives the preset differential motion unit to complete the predetermined cleaning action in sequence according to the action connection specification.

[0146] Among them, collaborative control is a management and control method that enables the synchronous and coordinated operation of the movement and the cleaning operation, which can ensure that the movement trajectory and the cleaning operation range are highly matched.

[0147] (4) Collect cleaning area coverage data, abnormal trigger events and robot status parameters in real time.

[0148] For example, the cleaning robot control terminal activates the omnidirectional perception and acquisition device on the robot body, and continuously collects information during the entire process of performing cleaning actions following the preset differential motion unit. It relies on the spatial positioning perception device to record the range information of the board area that has been cleaned in real time, and summarizes and integrates all range information to form cleaning area coverage data. It relies on the operation fault monitoring device to capture various events that deviate from normal working conditions in real time, such as operation jams, component abnormalities, and boundary crossings, and collects them to form abnormal trigger events. It relies on the body status sensors to collect relevant information such as the overall machine operating power consumption, body tilt posture, and actual operating conditions of each execution unit, and organizes and summarizes them to form robot status parameters.

[0149] Among them, the cleaning area coverage data is used to reflect the completed range of the panel cleaning operation, the abnormal triggering event is used to record various abnormal operating conditions that occur during the operation, and the robot status parameters are used to reflect the real-time operating status of the cleaning robot.

[0150] (5) Based on the cleaning area coverage data, abnormal triggering events and robot state parameters, the task state is synthesized to generate the task execution state.

[0151] For example, the cleaning robot control terminal performs work completion progress statistics and coverage integrity judgment on the collected cleaning area coverage data, classifies the impact level of the summarized abnormal trigger events and analyzes the degree of work obstruction, conducts a comprehensive evaluation of the overall machine operation stability and work adaptability on the collected robot status parameters, integrates the progress judgment results, abnormal judgment results and working condition evaluation results into multi-dimensional information, completes the information regularization and arrangement according to the unified task status arrangement format of the external system, performs task status synthesis processing, and generates task execution status.

[0152] Among them, the task execution status is a standardized status information that integrates three types of information: operation progress, operation abnormality, and overall machine condition, which can fully reflect the overall operation status of the cleaning operation.

[0153] like Figure 2 As shown, based on the local environment map, robot body dimensions, and edge safety distance, path boundary calculations are performed to generate a safe passage area, including:

[0154] S201: Extract the boundaries of photovoltaic modules and passable areas based on the local environment map.

[0155] For example, relying on the loaded local environment map, the cleaning robot control terminal classifies and identifies the spatial contour points and regional attribute annotation information inside the map one by one. It distinguishes the photovoltaic module deployment area and the external vacant area based on the regional attribute identifier. It collects continuous spatial coordinates along the outer contour of the photovoltaic module to complete the complete extraction of the photovoltaic module boundary. At the same time, it selects the spatial range inside the map that has no obstacles and meets the basic travel conditions, collects the outer contour coordinate information of the spatial range, and completes the extraction of the boundary of the passable area.

[0156] Among them, the local environment map is a spatial mapping map that records the spatial layout of the work site, the distribution of obstacles, and the location of the components. The photovoltaic module boundary is the spatial outline limit corresponding to the physical shape of the photovoltaic panel. The passable area boundary is the outer boundary limit corresponding to the unobstructed travel space.

[0157] S202: Calculate the robot's safety envelope area based on the robot's body size and edge safety distance.

[0158] For example, by combining the preset edge safety distance standard and the robot body size parameters stored in the system, the cleaning robot control terminal determines the basic planar space range occupied by the robot's overall shape when it is stationary. Based on this basic shape outline, the buffer space corresponding to the edge safety distance is uniformly expanded outward. At the same time, the range is supplemented by taking into account the space occupied by the robot during dynamic movement such as turning and changing speed. The static shape space and the dynamic reserved space are integrated, and the robot's safety envelope area is calculated by expanding the edge of the planar space.

[0159] Among them, the robot body size is a set of physical parameters that characterize the length, width and working range of the robot as a whole; the edge safety distance is the anti-collision buffer distance reserved when the robot moves close to the edge of the area; and the robot safety envelope area is the complete safety space range that the robot needs to occupy throughout the entire process of being stationary and moving.

[0160] S203: Within the boundary of the photovoltaic module, perform path boundary calculation based on the passable area boundary and the robot safety envelope area to obtain the safe passable area.

[0161] For example, with the photovoltaic module boundary as the core spatial constraint, the cleaning robot control terminal restricts all operational behaviors to the internal space enclosed by the photovoltaic module boundary. The extracted passable area boundary is used as the basic travel path range. Invalid spaces extending beyond the photovoltaic module boundary are trimmed and eliminated. Based on the spatial occupancy of the robot's safety envelope area, the inner and outer contours of the basic travel path range are synchronously contracted and adjusted. All narrow spaces and high-risk spaces near the edge that cannot fully accommodate the smooth passage of the robot's safety envelope area are discarded. The compliance correction and redefinition of the overall path boundary are completed, resulting in a safe passage area.

[0162] The safe passage area is a dedicated workspace that simultaneously meets the requirements for component space constraints, barrier-free passage conditions, and safe movement space for the robot as a whole.

[0163] It should be understood that although the steps in the flowcharts of the embodiments described above 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 embodiments described above 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.

[0164] In one embodiment, such as Figure 3 As shown, this application also provides a modular autonomous cleaning machine control system 300 suitable for high-tilt photovoltaic modules, the system 300 comprising:

[0165] The task parsing module 301 is used to receive external cleaning task instructions through a preset communication unit and parse the boundary of the area to be cleaned and the cleaning mode; obtain the robot body size and edge safety distance through a preset structural parameter acquisition unit; obtain photovoltaic panel image data through a preset image acquisition unit; obtain depth data through a preset depth vision unit; and, based on the boundary of the area to be cleaned, the photovoltaic panel image data, and the depth data, identify passable areas and generate a local environment map containing safe passable areas.

[0166] The path planning module 302 is used to perform path boundary calculations based on the local environment map, robot body size and edge safety distance, and generate a safe passage area; based on the safe passage area and cleaning mode, it performs cleaning path generation processing to obtain a cleaning path with applied tilt angle safety constraints;

[0167] The pollution assessment module 303 is used to perform pollution feature analysis based on the photovoltaic panel image features acquired by the image acquisition unit to obtain pollution level classification results; and to mark polluted areas and generate a pollution distribution map based on the pollution level classification results.

[0168] The strategy generation module 304 is used to obtain the current cleaning execution unit type through a preset cleaning unit identification interface; and to adjust the cleaning parameters in a linked manner according to the pollution distribution map, the cleaning execution unit type and the cleaning mode to obtain an adaptive cleaning strategy.

[0169] The execution control module 305 is used to control the preset differential motion unit to perform cleaning actions based on the cleaning path and adaptive cleaning strategy, and generate the task execution status; the task execution status is uploaded to the external system through the preset status feedback unit.

[0170] Specifically, the cleaning robot control terminal includes a task parsing module 301, a path planning module 302, a pollution assessment module 303, a strategy generation module 304, and an execution control module 305.

[0171] The task parsing module receives external cleaning task instructions via a preset communication unit, deconstructs and analyzes the instructions to extract the boundaries and cleaning patterns of the area to be cleaned, and retrieves the robot's body size and edge safety distance using a preset structural parameter acquisition unit. It also captures visual data of the photovoltaic panel's appearance using a preset image acquisition unit and collects depth data reflecting spatial relationships using a preset depth vision unit. Combining obstacle features in the photovoltaic panel image data with the spatial dimensional relationships reflected by the depth data, it filters out unobstructed areas suitable for robot movement, integrates the information of this area with the safety passage area definition standards, and generates a local environmental map containing the safety passage area.

[0172] Among them, the boundary of the area to be cleaned is used to delineate the spatial range of the cleaning operation, the cleaning mode is used to clarify the standards and requirements for the operation, the robot body size and edge safety distance are the core basis for defining the safe space of the operation, and the local environment map can completely record the spatial layout, passable range and safe travel boundary of the operation area.

[0173] The path planning module retrieves a local environment map, combines the robot's body size and edge safety distance to calculate and determine the safe boundary range of the path to generate a safe passage area. Based on this safe passage area, and combined with the operational requirements of the predetermined cleaning mode, it plans a continuous travel trajectory that can cover the area to be cleaned. At the same time, it incorporates anti-rollover and anti-derailment safety constraints in high-tilt scenarios to complete the optimization and generation of the cleaning path, resulting in a cleaning path with tilt angle safety constraints.

[0174] The safe passage area is a compliant travel space that is adapted to the robot's size and safe distance. The cleaning path with tilt angle safety constraints not only meets the safe operation requirements of high tilt angle environment, but also ensures the integrity of the clean coverage of photovoltaic panels.

[0175] The pollution assessment module extracts pollution-related visual features from photovoltaic panel images, such as the color depth, coverage area, and distribution density of stains. Based on the feature intensity and preset grading standards, it determines the severity of pollution and obtains standardized pollution level classification results. According to the classification results, it marks the spatial location and boundary of different pollution level areas and integrates all the marked information to generate a pollution distribution map that intuitively presents the pollution distribution on the panel surface.

[0176] Among them, pollution-related visual features are the core basis for distinguishing the severity of pollution. The pollution level classification results clarify the severity of pollution in each area, and the pollution distribution map provides a pollution distribution reference for subsequent cleaning parameter adjustments.

[0177] The strategy generation module identifies the type of cleaning execution unit currently in use through a preset cleaning unit identification interface, clarifies its core operational performance such as operating power, cleaning method, and parameter adjustment range, and combines the pollution situation of different areas presented by the pollution distribution map, the upper limit of the operating capacity of the cleaning execution unit, and the standard requirements of the established cleaning mode to adjust core parameters such as cleaning intensity, operating speed, water spray volume, and roller brush speed in a coordinated manner. This ensures that the parameters not only match the cleaning needs of areas with different pollution levels, but also adapt to the operating capabilities of the cleaning execution unit, generating an adaptive cleaning strategy that is suitable for the on-site operating conditions.

[0178] Among them, the type of cleaning execution unit determines the actual cleaning operation capacity boundary on site, and the adaptive cleaning strategy can flexibly and dynamically match the optimal operation parameters according to multiple site conditions.

[0179] The execution control module, based on the determined cleaning path and adaptive cleaning strategy, issues control commands such as travel speed, steering angle, and start / stop of cleaning actions to the preset differential motion unit, ensuring that it completes all cleaning operations in an orderly manner. Simultaneously, it collects various operational information in real time, such as work coverage progress, equipment operating status, and abnormal triggering conditions, and integrates them into a task execution status that can fully reflect the work progress and working conditions. Through the preset status feedback unit, the task execution status is uniformly uploaded to the external system, ensuring that the external system can monitor the work dynamics in real time.

[0180] Among them, the preset differential motion unit is the core execution component that enables the robot to move and clean the board surface. The task execution status can provide comprehensive feedback on the progress, equipment status and work results of the entire cleaning operation.

[0181] The path planning module 302 is also used for:

[0182] Based on the cleaning mode, determine the path coverage strategy, which includes a reciprocating coverage strategy, a zoned coverage strategy, or a localized enhanced cleaning strategy.

[0183] Based on the path coverage strategy and safe passage area, the initial path generation process is performed to obtain the initial clean path;

[0184] Based on the preset tilt angle safety constraint rules and the robot body size, the turning and edge movement actions in the initial cleaning path are optimized to generate a cleaning path with tilt angle safety constraints.

[0185] The path planning module 302 is also used for:

[0186] Based on the robot's body dimensions, the turning safety margin and edge movement safety margin are calculated. The turning safety margin is used to limit actions with a turning radius smaller than a safety threshold; the edge movement safety margin is used to limit lateral movement actions within the edge risk area.

[0187] Based on the tilt angle safety constraint rules and the turning safety margin, the turning radius of the turning action in the initial cleaning path is optimized to obtain the optimized turning path segment;

[0188] Based on the tilt angle safety constraint rules and the edge movement safety margin, the edge movement action in the initial cleaning path is optimized by edge distance to obtain the optimized edge path segment;

[0189] The optimized turning path segment, the optimized edge path segment, and the regular path segment in the initial cleaning path are merged to generate a cleaning path with tilt angle safety constraints.

[0190] The policy generation module 304 is also used for:

[0191] Based on the pollution distribution map, pollution areas are classified into first-level pollution areas and second-level pollution areas. The first-level pollution areas correspond to areas with pollution scores exceeding a set threshold, while the second-level pollution areas correspond to areas with pollution scores below the set threshold.

[0192] Based on the type of cleaning execution unit, parameter rules are matched to obtain cleaning parameter adjustment rules;

[0193] Based on the area marking information of the first pollution level, the cleaning intensity is calculated and a first-class cleaning intensity instruction is generated.

[0194] Based on the area marking information of the second pollution level, the cleaning intensity is calculated to generate a second type of cleaning intensity instruction;

[0195] Based on the cleaning mode, the cleaning instructions are modified to obtain the modified first type of cleaning intensity instructions and the modified second type of cleaning intensity instructions;

[0196] The modified first type of cleaning intensity command, the modified second type of cleaning intensity command, and the cleaning parameter adjustment rules are integrated and processed to generate an adaptive cleaning strategy.

[0197] The execution control module 305 is also used for:

[0198] Based on the cleaning path, the motion command is decomposed to generate the target linear velocity, target angular velocity, and target velocities of the left and right tracks;

[0199] Based on the adaptive cleaning strategy, the cleaning action is analyzed to obtain the roller brush rotation speed command, water spray volume command, and movement speed correction command.

[0200] Based on the target linear velocity, target angular velocity, target speed of left and right tracks, roller brush rotation speed command, water spray volume command and movement speed correction command, coordinated control is performed to drive the differential motion unit to complete the cleaning action;

[0201] Real-time collection of cleaning area coverage data, abnormal triggering events, and robot status parameters;

[0202] Based on the cleaning area coverage data, abnormal triggering events, and robot state parameters, the task state is synthesized to generate the task execution state.

[0203] The path planning module 302 is also used for:

[0204] Based on the local environment map, extract the boundaries of photovoltaic modules and the boundaries of passable areas;

[0205] Calculate the robot's safety envelope area based on the robot's body size and edge safety distance;

[0206] Within the boundaries of the photovoltaic modules, path boundary calculations are performed based on the traversable area boundary and the robot's safety envelope area to obtain the safe traversable area.

[0207] In one embodiment, a multi-layer control module, hereinafter referred to as the control module, is used to uniformly schedule and control the various functional modules of the robot, and is the core module for the robot to perform autonomous cleaning tasks. This module adopts a hierarchical structure of upper-layer computing and planning units and lower-layer motion control units. The upper layer is responsible for task management, perception processing, pollution assessment, path planning, and safety decision-making, while the lower layer is responsible for motion execution, speed feedback, motor control, and real-time safety protection.

[0208] This module mainly includes the following parts:

[0209] 1. Upper-level computational planning unit

[0210] The upper-level computational planning unit receives external cleaning tasks, processes image and depth vision data, and performs contamination assessment, local environment modeling, cleaning path planning, and task status management. Based on the area to be cleaned, contamination distribution, edge safety distance, robot status, power information, and cleaning execution unit type, this unit generates cleaning paths, target motion information, and cleaning control parameters, and sends corresponding instructions to the lower-level motion control unit.

[0211] 2. Low-level motion control unit

[0212] The underlying motion control unit is used to execute left and right track motor control, analog speed regulation signal output, enable control, direction control, braking control, speed feedback acquisition, speed closed-loop control, and safe stopping actions. This unit receives the target linear velocity, target angular velocity, or left and right track target velocity from the upper-level calculation and planning unit, and combines it with ToF edge detection, IMU attitude detection, and speed feedback results to dynamically limit the robot's movement, reverse, stop, or brake it, enabling the robot to safely and stably perform cleaning tasks on the surface of high-tilt photovoltaic modules.

[0213] In this embodiment, the upper-layer computational planning unit uses a Raspberry Pi and runs the ROS2 system to perform sensor data processing, local environment modeling, path planning, external communication, and task scheduling. The lower-layer motion control unit uses an Arduino Due to perform DAC analog speed control signal output, left and right track motor control, PG speed feedback acquisition, speed closed-loop control, and safe shutdown execution. The upper-layer computational planning unit and the lower-layer motion control unit interact with each other through a communication interface to achieve coordination between task planning, path following, and lower-layer real-time control.

[0214] In one embodiment, the differential motion module enables the robot to continuously walk, turn, back, and adjust its posture on the surface of a high-tilt photovoltaic module. Through independent drive and differential control of the left and right tracks, this module allows the robot to perform actions such as straight-line movement, turning, low-speed backtracking, and posture correction on the photovoltaic module surface, providing a motion execution basis for cleaning path following, edge avoidance, and safe stopping.

[0215] This module mainly includes the following parts:

[0216] 1. Track drive structure

[0217] The track drive structure includes left and right tracks, left and right drive motors, drive wheels, driven wheels, a tensioning mechanism, and a reduction gear transmission mechanism. The left and right tracks are each driven by an independent drive motor, and the motor output is transmitted to the drive wheels via the reduction gear transmission mechanism. The contact area between the tracks and the photovoltaic module surface uses flexible, non-slip, and wear-resistant materials to improve the robot's adhesion to high-angle surfaces and reduce damage to the photovoltaic module surface.

[0218] 2. Motor driver

[0219] This robot uses 24V brushless DC motors as the drive source for its left and right tracks. The motor drivers receive speed control signals from the control module and control the motors' enable, direction, braking, and speed. The underlying motion control unit in the control module outputs analog control signals via a DAC, which are then converted to voltage and sent to the motor drivers, enabling continuous and smooth adjustment of the left and right track speeds. This method facilitates dynamic speed limiting, soft start, and soft stop for the robot, and, in conjunction with braking control, improves the robot's controllability and safety when moving on the surface of high-tilt photovoltaic modules. The motor drivers also provide a PG speed feedback interface. The control module obtains the actual rotational speed by reading the PG pulse signals from the left and right motors, providing a data basis for subsequent speed closed-loop control and slippage detection.

[0220] 3. Motion command and safety constraint interface

[0221] The motion command and safety constraint interface is used to receive target motion information and safety constraint information from the control module. Target motion information includes target linear velocity, target angular velocity, or left and right track target speeds; safety constraint information includes speed limit commands, stop commands, reversal commands, and safety control commands triggered by edge risks, abnormal postures, or slippage tendencies. The differential motion module executes corresponding walking, turning, deceleration, stopping, or reversing actions according to the above commands, enabling the robot to safely and stably perform cleaning tasks on the surface of high-tilt photovoltaic modules.

[0222] In one embodiment, a modular cleaning execution module is used to perform dust removal, brushing, squeegeeing, wiping, and combined cleaning operations on the surface of photovoltaic modules. This module is located at the front or bottom of the robot body and connects to the robot body via a unified installation interface, allowing the robot to change the corresponding cleaning execution unit according to different types of contamination and cleaning tasks.

[0223] This module mainly includes the following parts:

[0224] 1. Installation connection structure

[0225] The mounting and connection structure includes a guiding structure, a positioning structure, and a locking structure, used to install the cleaning actuator in a predetermined position on the robot body and to prevent it from swaying, shifting, or falling off during robot movement and cleaning. Different types of cleaning actuators connect to the robot body through this unified interface, enabling quick replacement and stable installation.

[0226] 2. Cleaning execution and media supply components

[0227] The cleaning execution and media supply components include a roller brush, flexible scraper, wiping device, dry dust removal component, water pump, piping, nozzles, and corresponding drive mechanisms. The roller brush is used to remove dust and sand, the nozzles are used to spray clean water or cleaning solution onto the photovoltaic module surface, the flexible scraper is used to scrape off residual water film after cleaning, the wiping device is used to treat localized stains or water spots, and the dry dust removal component is used in lightly polluted or water-scarce scenarios. These components can constitute a cleaning execution unit individually, or two or more can be combined to form a composite cleaning unit to adapt to different types of pollution and cleaning needs.

[0228] 3. Identification and Control Interface

[0229] The identification and control interface includes an electrical connection port, a cleaning unit identification port or identification signal, and an actuator control port. Cleaning unit identification methods include coded plugs, resistor identification, electronic tags, communication identification, or mechanical limit identification. The control module determines the type of the currently installed cleaning actuator through the identification interface and loads corresponding control parameters, such as brush speed, water spray volume, wiping action, wiping frequency, and robot movement speed matching the cleaning action. When the robot enters edge avoidance, abnormal posture, retreat, or stop mode, the control module synchronously reduces or stops the cleaning actuator's action to prevent the cleaning mechanism from continuing to operate in an unsafe state.

[0230] In one embodiment, the perception and safety detection module is used to acquire data such as edge distance, robot posture, panel image, depth information, pollution distribution, and environmental status when the robot operates on the surface of the photovoltaic module. This provides a perception basis for robot edge avoidance, posture stabilization control, cleaning path planning, pollution assessment, and environmental modeling. This module is connected to the control module and converts the perception results into data inputs required for robot control and path planning.

[0231] This module mainly includes the following parts:

[0232] 1. ToF edge detection component

[0233] The ToF edge detection component includes ToF distance sensors located at the four corners of the robot body to detect changes in distance between the robot's four corner areas and the edges of the photovoltaic modules, gaps between modules, or areas with height differences. Based on the ToF distance measurement data from the four corners, the control module determines whether the robot is approaching the edge of the photovoltaic panel, whether there is a risk of one-sided suspension, and whether the robot meets safety distance requirements when turning, reversing, or traveling along the edge.

[0234] 2. IMU Attitude Detection Component

[0235] An IMU (Integrated Mutual Measure) attitude detection component is installed inside the robot body to acquire pitch angle, roll angle, acceleration, and attitude change information when the robot operates on the surface of high-tilt photovoltaic modules. The control module uses the IMU data to determine whether the robot is experiencing sideslip, slippage, abnormal tilting, or attitude instability, and uses the results for dynamic speed limiting, dangerous turn restriction, motion correction, and safe shutdown.

[0236] 3. Image and Depth Vision Components

[0237] The image and depth vision component includes an image acquisition device and a depth camera mounted in front of the robot to acquire image, depth, and spatial structure information of the photovoltaic (PV) modules in front of the robot. This component identifies PV module boundaries, obstacles on the panel surface, traversable areas, and localized contamination, and generates contaminated areas, contamination levels, or contamination scores based on image and depth data. The control module updates the cleaning path, adjusts cleaning intensity, determines traversable areas, and assists in robot localization based on the contamination assessment results. This component is also used for robot local environmental perception and SLAM environmental modeling, providing data for cleaning path planning, local obstacle avoidance, contamination assessment, and task execution status determination.

[0238] 4. Sensing Data Interface

[0239] The perception data interface transmits data acquired by the ToF edge detection component, IMU attitude detection component, and image and depth vision component to the control module. Its output includes four-corner ranging results, edge risk markers, edge distance margins, attitude status, image information, depth information, contamination assessment results, obstacle information, and a local environment model. Based on this perception data, the control module executes actions such as speed limiting, stopping, reversing, path replanning, cleaning intensity adjustment, or stopping cleaning, thereby ensuring the robot safely and stably performs cleaning tasks on the surface of high-tilt photovoltaic modules.

[0240] In one embodiment, the energy and power management module provides power to the robot's various functional modules and detects, manages, and reports the overall power status of the robot, providing an energy foundation for the robot to continuously perform high-tilt photovoltaic module cleaning tasks. This module mainly includes a main battery unit and branch power supply modules.

[0241] 1. Main battery unit

[0242] The main battery unit provides the basic power for the robot. This robot uses a 24V lithium battery as its main power source and is equipped with a charging interface that can connect to a cleaning base station or external charging equipment. It supports recharging during task breaks and can be configured with fast charging functionality. The main battery unit detects battery voltage, current, remaining charge, charging status, and abnormal conditions, and reports the charge information to the control module. This allows the control module to determine whether the robot should continue cleaning, return to the base station, or enter a safe shutdown state.

[0243] 2. Branch power supply module

[0244] The branch power supply module is used to convert the main battery output into the operating voltage required by the differential motion module, modular cleaning execution module, sensing and safety detection module, communication and data interaction module, and control module. This module can detect or estimate the voltage, current, and power of each power supply branch, providing a basis for the control module to perform energy consumption statistics, task path adjustment, cleaning intensity control, and low battery protection.

[0245] In one embodiment, the communication and data interaction module is used to realize information transmission between the robot and the host computer, cleaning base station, drone inspection system, overall scheduling system, and other cleaning robots, enabling the robot to receive cleaning tasks, upload operating status, and support single-robot or multi-robot collaborative cleaning operations. This module mainly includes a task receiving unit, a status uploading unit, and a collaborative data interface.

[0246] 1. Task Receiving Unit

[0247] The task receiving unit receives cleaning tasks and control commands from external systems, including the area to be cleaned, cleaning mode, contaminated area information, path reference information, return-to-base station command, and emergency stop command. The control module determines whether to initiate a cleaning task based on the task information and the robot's current status.

[0248] 2. Status Upload Unit

[0249] The status upload unit is used to upload the robot's working status, cleaning progress, edge risks, posture status, battery status, cleaning execution module status, and fault information, enabling external systems to monitor and schedule the robot's task execution.

[0250] 3. Collaborative Data Interface

[0251] The collaborative data interface supports collaborative operations between robots and drones, cleaning base stations, the overall scheduling system, and other cleaning robots. In multi-robot working scenarios, this interface can be used to exchange information such as task area, current location, cleaning progress, remaining power, and fault status, thereby avoiding duplicate cleaning, path conflicts, or task omissions, and improving the cleaning efficiency of large-area photovoltaic modules.

[0252] In one embodiment, a safe path planning and motion control method for high-tilt photovoltaic modules is used to address the problems of robot slippage, sideslippage, boundary crossing, or dangerous turns when moving on the surface of high-tilt photovoltaic modules, such as those with a maximum tilt angle of approximately 35°. The method mainly includes the following steps:

[0253] 1. Determining the boundaries of the cleaning area and path

[0254] The control module determines the cleaning area where the robot can safely pass through based on the boundary information of the photovoltaic modules to be cleaned, the robot's body dimensions, the effective width of the cleaning execution unit, and the edge safety distance. For the edges of the photovoltaic modules, gaps between modules, or areas with height differences, the control module reserves a safety distance to prevent the robot's path from getting too close to dangerous areas.

[0255] 2. Cleaning path generation

[0256] The control module generates a cleaning path suitable for the current work surface based on the cleaning area. The cleaning path can employ a reciprocating coverage path, a zoned coverage path, or a localized focused cleaning path to ensure comprehensive cleaning coverage of the photovoltaic module surface. During path generation, the control module comprehensively considers the width of the cleaning execution unit, the robot's turning space, the path length, and the cleaning coverage effect to reduce missed areas and unnecessary duplicate paths.

[0257] 3. High tilt angle safety restraint

[0258] In high-angle operation, the control module imposes safety constraints on the robot's movement, restricting dangerous actions such as sharp turns, rapid rotations in place, and lateral movements near edges. For areas requiring turning, the control module prioritizes planning gentler turning paths or reserves sufficient turning space at the edges to reduce the risk of sideslip, slippage, or boundary crossing.

[0259] 4. Path following and speed control

[0260] As the robot moves along the planned path, the control module generates target linear velocity, target angular velocity, or target speeds for the left and right tracks based on the target path, and sends the target motion information to the differential motion module. The underlying motion control unit performs closed-loop speed control based on the speed feedback from the left and right motors (PGs), enabling the left and right tracks to follow the target speed relatively stably.

[0261] 5. Edge, posture, and slippage judgment

[0262] During movement, the control module reads ToF ranging data from the four corners, IMU attitude data, and PG speed feedback data from the left and right motors in real time. The ToF data is used to determine whether the robot is approaching the edge of the photovoltaic module, the gap between the modules, or the area of ​​height difference; the IMU data is used to determine whether the robot is exhibiting side slip, sliding, abnormal tilting, or a tendency to lose posture; and the PG speed feedback is used to determine whether there is an abnormal deviation between the actual speed of the left and right tracks and the target speed.

[0263] 6. Dynamic risk avoidance and path adjustment

[0264] When insufficient edge distance, abnormal posture, or slippage is detected, the control module performs actions such as dynamic speed limiting, steering restriction, stopping forward movement, slow-speed reversal, or replanning the path based on the risk level. Through these methods, the robot can safely and stably complete cleaning path following and movement operations on the surface of high-tilt photovoltaic modules.

[0265] In one embodiment, an adaptive cleaning control method based on contamination assessment is used to address the problems of traditional cleaning robots, such as fixed cleaning intensity, high energy consumption, and difficulty in adapting to different types of contamination. This method mainly includes the following steps:

[0266] 1. Pollution Information Acquisition

[0267] The robot acquires image, depth, and local environmental information from the surface of the photovoltaic module using image and depth vision components. The control module identifies or assesses contamination levels such as dust, mud, water spots, and bird droppings based on the module's image features, depth variations, and local environmental conditions.

[0268] 2. Classification of Polluted Areas and Levels

[0269] The control module generates contaminated areas, contamination levels, or contamination scores based on the contamination identification results. Areas with higher contamination levels are marked as priority cleaning areas, while areas with lower contamination levels are marked as general cleaning areas or quick-pass areas. This result also serves as the basis for cleaning path planning and cleaning intensity adjustment.

[0270] 3. Cleaning parameters are adaptively adjusted.

[0271] The control module dynamically adjusts cleaning parameters based on the contamination level and the type of cleaning execution unit currently installed. For heavily contaminated areas, the control module increases the roller brush speed, increases the water spray volume, decreases the robot's movement speed, or increases the number of repeated cleaning cycles; for lightly contaminated areas, the control module decreases the roller brush speed, reduces the water spray volume, or adopts a fast-pass strategy to reduce unnecessary energy consumption and actuator wear.

[0272] 4. Matching control of different cleaning execution units

[0273] When the robot is equipped with a roller brush cleaning unit, the control module mainly adjusts the roller brush rotation speed and movement speed; when the robot is equipped with a water spray scrubbing unit, the control module adjusts the water spray volume, scrubbing intensity, and passing speed; when the robot is equipped with a squeegee or wiping unit, the control module performs secondary treatment based on the residual water film, water spots, or local stains; when the robot is in a water shortage or light pollution scenario, the control module calls the dry dust removal unit to perform low-energy cleaning.

[0274] 5. Cleaning path and cleaning intensity are linked.

[0275] The control module combines pollution assessment results with cleaning paths, planning focused or repeated cleaning paths for heavily polluted areas and reducing dwell time or cleaning intensity for lightly polluted areas. By linking path planning with cleaning parameters, the robot can reduce ineffective movement and over-cleaning while ensuring cleaning effectiveness.

[0276] 6. Energy Constraints and Task Optimization

[0277] The control module optimizes the cleaning strategy by combining the remaining power, branch power, and power consumption status provided by the energy and power management module. When the remaining power is insufficient or the power load is high, the control module reduces the cleaning intensity, decreases the number of repeated cleaning cycles, shortens the remaining task path, or controls the robot to return to the base station. Through these methods, the robot can improve operational efficiency and reduce energy consumption while ensuring cleaning effectiveness.

[0278] After the cleaning task is completed, the control module records the cleaned area, cleaning parameters, abnormal information and task results, and uploads them to the external system through the communication and data interaction module for task review, subsequent scheduling and cleaning effect analysis.

[0279] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0280] In one embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0281] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and 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 modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0282] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A modular autonomous cleaning machine control method suitable for high-tilt photovoltaic modules, characterized in that, The method includes: The robot receives external cleaning task instructions through a preset communication unit and parses the boundary of the area to be cleaned and the cleaning mode; it obtains the robot body size and edge safety distance through a preset structural parameter acquisition unit; it obtains photovoltaic panel image data through a preset image acquisition unit; it obtains depth data through a preset depth vision unit; and based on the boundary of the area to be cleaned, the photovoltaic panel image data, and the depth data, it identifies passable areas and generates a local environment map containing safe passable areas. Based on the local environment map, the robot body size, and the edge safety distance, path boundary calculation is performed to generate a safe passage area; based on the safe passage area and the cleaning mode, cleaning path generation processing is performed to obtain a cleaning path with applied tilt angle safety constraints; Based on the photovoltaic panel image features acquired by the image acquisition unit, pollution feature analysis is performed to obtain pollution level classification results; according to the pollution level classification results, pollution areas are marked to generate a pollution distribution map; The current cleaning execution unit type is obtained through a preset cleaning unit identification interface; the cleaning parameters are adjusted in conjunction with the pollution distribution map, the cleaning execution unit type, and the cleaning mode to obtain an adaptive cleaning strategy. Based on the cleaning path and the adaptive cleaning strategy, a preset differential motion unit is controlled to perform cleaning actions, generating a task execution status; the task execution status is then uploaded to an external system through a preset status feedback unit.

2. The modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to claim 1, characterized in that, The step of generating a cleaning path based on the safe passage area and the cleaning mode to obtain a cleaning path with applied tilt angle safety constraints includes: Based on the cleaning mode, a path coverage strategy is determined, wherein the path coverage strategy includes a reciprocating coverage strategy, a zoned coverage strategy, or a localized enhanced cleaning strategy. Based on the path coverage strategy and the safe passage area, an initial path generation process is performed to obtain an initial clean path; Based on the preset tilt angle safety constraint rules and the robot body dimensions, the turning and edge movement actions in the initial cleaning path are optimized to generate the cleaning path with the applied tilt angle safety constraint.

3. The modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to claim 2, characterized in that, The step of optimizing the turning and edge movement actions in the initial cleaning path according to the preset tilt angle safety constraint rules and the robot body dimensions to generate the cleaning path with applied tilt angle safety constraints includes: Based on the robot's body dimensions, a turning safety margin and an edge movement safety margin are calculated. The turning safety margin is used to limit actions with a turning radius smaller than a safety threshold. The edge movement safety margin is used to limit lateral movement actions within the edge risk area. Based on the tilt angle safety constraint rules and the turning safety margin, the turning radius of the turning action in the initial cleaning path is optimized to obtain the optimized turning path segment; Based on the tilt angle safety constraint rule and the edge movement safety margin, the edge movement action in the initial cleaning path is optimized by edge distance to obtain the optimized edge path segment; The optimized turning path segment, the optimized edge path segment, and the regular path segment in the initial cleaning path are merged to generate the cleaning path with applied tilt angle safety constraints.

4. The modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to claim 1, characterized in that, The step of adjusting cleaning parameters in conjunction with the pollution distribution map, the cleaning execution unit type, and the cleaning mode to obtain an adaptive cleaning strategy includes: Based on the pollution distribution map, pollution areas are classified into levels to obtain first pollution level area marking information and second pollution level area marking information. The first pollution level area corresponds to the area where the pollution level score exceeds a set threshold; the second pollution level area corresponds to the area where the pollution level score is below the set threshold. Based on the cleaning execution unit type, parameter rule matching is performed to obtain cleaning parameter adjustment rules; Based on the area marking information of the first pollution level, a cleaning intensity calculation is performed to generate a first type of cleaning intensity instruction; Based on the area marking information of the second pollution level, a cleaning intensity calculation is performed to generate a second type of cleaning intensity instruction; Based on the cleaning mode, the cleaning instructions are modified to obtain the modified first type of cleaning intensity instructions and the modified second type of cleaning intensity instructions; The modified first type of cleaning intensity command, the modified second type of cleaning intensity command, and the cleaning parameter adjustment rules are merged and processed to generate the adaptive cleaning strategy.

5. A modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to claim 1, characterized in that, The step of controlling a preset differential motion unit to perform cleaning actions based on the cleaning path and the adaptive cleaning strategy, and generating a task execution state, includes: Based on the cleaning path, the motion command is decomposed to generate the target linear velocity, target angular velocity, and target velocities of the left and right tracks; Based on the adaptive cleaning strategy, the cleaning action is parsed to obtain the roller brush rotation speed command, water spray volume command, and moving speed correction command. Based on the target linear velocity, the target angular velocity, the target velocities of the left and right tracks, the roller brush rotation speed command, the water spray volume command, and the movement speed correction command, coordinated control is performed to drive the differential motion unit to complete the cleaning action; Real-time collection of cleaning area coverage data, abnormal triggering events, and robot status parameters; Based on the cleaning area coverage data, the abnormal triggering event, and the robot state parameters, the task state is synthesized to generate the task execution state.

6. The modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to claim 1, characterized in that, The step of calculating path boundaries and generating a safe passage area based on the local environment map, the robot's body dimensions, and the edge safety distance includes: Based on the local environment map, extract the boundaries of the photovoltaic modules and the boundaries of the passable areas; Calculate the robot's safety envelope area based on the robot's body dimensions and the edge safety distance; Within the boundary of the photovoltaic module, the path boundary is calculated based on the passable area boundary and the robot safety envelope area to obtain the safe passable area.

7. A modular autonomous cleaning machine control system suitable for high-tilt photovoltaic modules, characterized in that, The system includes: The task parsing module is used to receive external cleaning task instructions through a preset communication unit and parse the boundary of the area to be cleaned and the cleaning mode; obtain the robot body size and edge safety distance through a preset structural parameter acquisition unit; obtain photovoltaic panel image data through a preset image acquisition unit; obtain depth data through a preset depth vision unit; and identify passable areas based on the boundary of the area to be cleaned, the photovoltaic panel image data, and the depth data to generate a local environment map containing safe passable areas. The path planning module is used to perform path boundary calculation based on the local environment map, the robot body size, and the edge safety distance to generate a safe passage area; and to perform cleaning path generation processing based on the safe passage area and the cleaning mode to obtain a cleaning path with applied tilt angle safety constraints. The pollution assessment module is used to perform pollution feature analysis based on the photovoltaic panel image features acquired by the image acquisition unit to obtain pollution level classification results; and to mark polluted areas and generate a pollution distribution map based on the pollution level classification results. The strategy generation module is used to obtain the current cleaning execution unit type through a preset cleaning unit identification interface; and to adjust the cleaning parameters in a coordinated manner according to the pollution distribution map, the cleaning execution unit type, and the cleaning mode to obtain an adaptive cleaning strategy. The execution control module is used to control the preset differential motion unit to perform cleaning actions based on the cleaning path and the adaptive cleaning strategy, and generate a task execution status; and upload the task execution status to an external system through a preset status feedback unit.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the modular autonomous cleaning machine control method for high-tilt photovoltaic modules according to any one of claims 1 to 6.

Images