A method and system for optimizing a path of a cleaning robot

Abstract

The application relates to the technical field of intelligent sensing, in particular to a path optimization method and system of a cleaning robot, which comprises the following steps: acquiring an environment profile of a to-be-cleaned area and generating an initial traversal path; controlling the cleaning robot to travel along the initial traversal path, extracting a current output torque and a corresponding actual displacement amount, and determining an abnormal site; performing envelope fitting on a plurality of adjacent abnormal sites to generate a low-adhesion boundary; generating an involute arc segment on the periphery of the low-adhesion boundary and connecting the initial traversal path to generate a corrected traversal path; and controlling the cleaning robot to travel along the corrected traversal path, so that the geometric center of the cleaning robot is kept outside the low-adhesion boundary, and the effective cleaning surface of the cleaning robot is cut into the inside of the low-adhesion boundary. The application achieves the technical effects of accurately balancing the safety avoidance of the main body and the deep coverage of the cleaning component, and eliminating the cleaning blind area of the low-adhesion area.

Description

Technical Field

[0001] This application relates to the field of intelligent sensing technology, specifically to a path optimization method and system for a cleaning robot. Background Technology

[0002] With the rapid development of smart home technology, cleaning robots have become an important part of modern homes. Cleaning robots achieve automated cleaning of indoor floors through autonomous path planning and environmental perception. However, existing technologies still have significant shortcomings when dealing with complex floor environments.

[0003] Currently, the mainstream path planning methods for cleaning robots mainly include zigzag traversal, spiral traversal, and random collision traversal. These methods can achieve good coverage on regular, flat surfaces, but in practical applications, indoor floors often have low-adhesion areas such as carpet edges, slippery areas, and polished floors. When cleaning robots enter these areas, they are prone to tire slippage, deviation, or even getting stuck, leading to reduced cleaning efficiency and missed areas.

[0004] Existing technologies typically use visual recognition or lidar scanning to pre-identify obstacles, but traditional sensors struggle to effectively detect changes in physical properties such as ground adhesion. Some solutions propose addressing this issue by adding dedicated ground material recognition sensors, but this significantly increases hardware costs and system complexity. Furthermore, even when low-adhesion areas are identified, existing path planning methods often employ simple detour strategies, avoiding these areas altogether. This results in ineffective cleaning of low-adhesion areas and their surroundings, creating cleaning blind spots.

[0005] On the other hand, the effective cleaning surface of a cleaning robot usually extends beyond the edge of the chassis. Theoretically, this allows the cleaning components to penetrate low-adhesion areas for cleaning while the robot body avoids obstacles. However, existing technologies lack precise path generation and attitude control methods, making it difficult to achieve a balance between "body avoidance" and "cleaning coverage." This results in either abandoning cleaning or forcing entry, which can easily lead to slippage. Summary of the Invention

[0006] In view of the aforementioned problems, this application is proposed to provide a path optimization method and system for a cleaning robot that overcomes or at least partially solves the aforementioned problems, comprising: A path optimization method for a cleaning robot, characterized by comprising the following steps: Obtain the environmental contour of the area to be cleaned, and generate an initial traversal path based on the environmental contour; The cleaning robot is controlled to travel along the initial traversal path, the current output torque and the corresponding actual displacement are extracted, and abnormal locations are determined based on the current output torque and the actual displacement. Envelope fitting is performed on multiple adjacent anomalous sites to generate low-adhesion boundaries; An involute arc segment is generated around the low adhesion boundary, and the involute arc segment is connected to the initial traversal path to generate a corrected traversal path. The cleaning robot is controlled to travel along the modified traversal path, so that the geometric center of the cleaning robot remains outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.

[0007] A path optimization system for a cleaning robot includes: The path generation module is used to obtain the environmental contour of the area to be cleaned and generate an initial traversal path based on the environmental contour. The location determination module is used to control the cleaning robot to travel along the initial traversal path, extract the current output torque and the corresponding actual displacement, and determine abnormal locations based on the current output torque and the actual displacement. The centroid extraction module is used to perform envelope fitting on multiple adjacent anomalous sites to generate low-adhesion boundaries. The path correction module is used to generate an involute arc segment on the periphery of the low adhesion boundary and connect the involute arc segment to the initial traversal path to generate a corrected traversal path. The driving control module is used to control the cleaning robot to travel along the modified traversal path, so that the geometric center of the cleaning robot is kept outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.

[0008] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a path optimization method for a cleaning robot as described above.

[0009] This application has the following advantages: In the embodiments of this application, addressing the shortcomings of existing cleaning robot path planning methods that rely on visual recognition or lidar scanning and cannot detect changes in physical characteristics such as ground adhesion differences, resulting in missing identification of low-adhesion areas, path planning employing only simple detour strategies, and the formation of cleaning blind spots in and around low-adhesion areas without precise path generation and attitude control methods to achieve a balance between subject avoidance and cleaning coverage, this application provides a path optimization method and system for cleaning robots, applicable to cleaning robots that need to perform autonomous cleaning operations in complex ground environments, including the steps of: acquiring the environmental contour of the area to be cleaned, and generating an initial path based on the environmental contour. The initial traversal path is established; the cleaning robot is controlled to travel along the initial traversal path, the current output torque and the corresponding actual displacement are extracted, and abnormal sites are determined based on the current output torque and the actual displacement; envelope fitting is performed on multiple adjacent abnormal sites to generate a low adhesion boundary; an involute arc segment is generated on the periphery of the low adhesion boundary, and the involute arc segment is connected to the initial traversal path to generate a corrected traversal path; the cleaning robot is controlled to travel along the corrected traversal path, so that the geometric center of the cleaning robot remains outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.By extracting the current output torque and corresponding actual displacement of the drive motor in real time during the initial traversal, and determining the abnormal sites based on the abnormal deviation relationship between the two, this approach solves the problems in existing technologies where traditional vision or lidar sensors cannot effectively detect differences in ground adhesion, while adding dedicated material recognition sensors significantly increases hardware costs and system complexity. This achieves the technical effect of dynamically perceiving low-adhesion areas using feedback from the robot's own drive system without adding extra hardware. Furthermore, by performing envelope fitting on multiple adjacent abnormal sites to generate spatially continuous low-adhesion boundaries, this addresses the problem in existing technologies where isolated abnormal points cannot reflect the overall contour of the low-adhesion area and are difficult to provide effective geometric constraints for subsequent path planning. This transforms discrete perception data into structured region boundaries, providing a basis for accurate path planning. The technical effects based on geometry are as follows: By generating involute arc segments around the low-adhesion boundary and connecting them to the initial traversal path, a corrected traversal path is generated. This solves the problem in existing technologies where path planning only uses simple detour strategies, resulting in ineffective cleaning of the low-adhesion area and its surrounding areas, creating cleaning blind spots. This achieves the technical effect of complete coverage of the area surrounding the low-adhesion boundary while ensuring driving safety. By controlling the cleaning robot to travel along the corrected traversal path, keeping the geometric center outside the low-adhesion boundary while allowing the effective cleaning surface to cut into the interior of the low-adhesion boundary, this solves the problem in existing technologies where there is a lack of precise path generation and attitude control methods, making it difficult to coordinate the body's avoidance and cleaning coverage. This achieves the technical effect of precisely balancing the body's safe avoidance and the deep coverage of the cleaning components, eliminating cleaning blind spots in the low-adhesion area. Attached Figure Description

[0010] To more clearly illustrate the technical solution of this application, the drawings used in the description of this application will be briefly introduced below. Obviously, the 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.

[0011] Figure 1 This is a flowchart illustrating the steps of a path optimization method for a cleaning robot according to an embodiment of this application; Figure 2 This is a structural block diagram of a path optimization system for a cleaning robot provided in one embodiment of this application; Figure 3 This is a schematic diagram of the structure of a computer electronic device provided in an embodiment of this application; 1. Computer electronic device; 2. External device; 3. Processing unit; 4. Bus; 5. Network adapter; 6. I / O interface; 7. Display; 8. Memory; 9. Random access memory; 10. Cache memory; 11. Storage system; 12. Program / utility; 13. Program module. Detailed Implementation

[0012] To make the objectives, features, and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0013] The inventors discovered through analysis of existing technologies that the existing cleaning robot path planning methods rely on visual recognition or lidar scanning and cannot detect changes in physical characteristics such as differences in ground adhesion. They have technical defects such as lack of low adhesion area identification, path planning only adopts a simple detour strategy, low adhesion areas and their surroundings form cleaning blind spots, and lack precise path generation and attitude control methods to achieve a balance between subject avoidance and cleaning coverage.

[0014] Reference Figure 1 This application illustrates a path optimization method for a cleaning robot according to an embodiment of the present application, including the following steps: S110. Obtain the environmental contour of the area to be cleaned, and generate an initial traversal path based on the environmental contour; S120. Control the cleaning robot to travel along the initial traversal path, extract the current output torque and the corresponding actual displacement, and determine the abnormal location based on the current output torque and the actual displacement. S130. Envelope fitting is performed on multiple adjacent abnormal sites to generate low-adhesion boundaries; S140. Generate an involute arc segment around the low adhesion boundary and connect the involute arc segment to the initial traversal path to generate a corrected traversal path. S150. Control the cleaning robot to travel along the modified traversal path, so that the geometric center of the cleaning robot remains outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.

[0015] First, some nouns or terms that appear in the description of this embodiment shall be interpreted as follows: A cleaning robot is an intelligent device that can autonomously perform cleaning tasks. It is typically equipped with a drive mechanism, a cleaning mechanism, sensors, and a control system. It can move along a preset or autonomously planned path and use its cleaning mechanism to clean the ground.

[0016] Environmental profiles refer to the geometric shape information of the external boundaries and internal obstacles of the area to be cleaned. They can usually be obtained through sensors, such as LiDAR and visual sensors, and are used to build a map model of the area to be cleaned.

[0017] The initial traversal path refers to the preset travel trajectory planned by the cleaning robot based on the environmental contours without considering differences in ground adhesion, aiming to cover the entire area to be cleaned. The initial traversal path can typically adopt a bow-shaped, spiral, or other regular coverage pattern.

[0018] The current output torque refers to the instantaneous torque applied to the drive wheels by the drive system of the cleaning robot during operation, which directly reflects the working state of the drive system and the force exerted on the ground.

[0019] Actual displacement refers to the actual spatial position change of a cleaning robot within a certain time period. Actual displacement can be measured by an odometer, inertial measurement unit, or other positioning sensors.

[0020] Anomalies refer to geographical locations where a cleaning robot's current output torque and actual displacement exhibit a significant mismatch during operation. This mismatch typically indicates insufficient traction between the drive wheels and the ground, leading to slippage or wheel spin.

[0021] Low adhesion boundary refers to the regional boundary formed by fitting multiple adjacent abnormal sites through a specific algorithm, which delineates the area where the ground adhesion is significantly lower than the normal level.

[0022] Envelope fitting is a geometric processing technique used to generate a continuous curve or polygon that encloses or approximately encloses a set of discrete points. In this embodiment, it is used to extract the contour of low-adhesion regions from anomalous sites.

[0023] An involute arc segment is a special type of curve, typically generated as the trajectory of a point on a straight line as it rolls without slipping on a circle. In this embodiment, the involute arc segment is used to construct a smooth avoidance path around a low-adhesion boundary.

[0024] The modified traversal path refers to a new travel path that is locally adjusted and optimized for low-adhesion areas by introducing involute arc segments, etc., based on the initial traversal path. The aim is to achieve effective cleaning of low-adhesion areas while preventing the robot body from slipping.

[0025] The geometric center refers to the geometric centroid of the cleaning robot chassis, which is usually the center point of the overall mass distribution of the cleaning robot. Its position is a key reference point for determining whether the cleaning robot enters or avoids a specific area.

[0026] The effective cleaning surface refers to the area covered by the components of a cleaning robot that actually perform the cleaning function, such as roller brushes, suction nozzles, and cloths. The effective cleaning surface typically extends beyond the geometric boundaries of the robot's chassis to facilitate cleaning edges and corners.

[0027] In the embodiments of this application, the acquisition of environmental contours can be achieved in various ways. For example, the static physical boundary information of the area to be cleaned can be pre-drawn manually or imported from CAD drawings and input into the control system of the cleaning robot. Based on this, classic path planning algorithms, such as the bow-shaped traversal algorithm or the spiral traversal algorithm, can be used to generate an initial traversal path covering the entire area. Another implementation method is that the cleaning robot scans the environment in initial exploration mode using its onboard LiDAR or depth camera to construct a two-dimensional or three-dimensional map of the area and extract the environmental contours from it. Subsequently, based on the constructed map information, an initial path capable of traversing all reachable areas is generated using a grid method or a topological graph method.

[0028] Furthermore, as the cleaning robot travels along its initial path, its drive system continuously outputs torque, which interacts with the ground through the wheels or tracks. The current output torque can be indirectly obtained by monitoring changes in the current or voltage of the drive motor. Simultaneously, the robot's internal odometer or inertial measurement unit can measure the robot's actual displacement in real time. When a significant increase in output torque is detected while the actual displacement remains relatively unchanged, this typically indicates that the drive wheels have slipped, and the robot's position at this point is marked as an anomaly. For example, a threshold can be set; when the ratio of torque to displacement exceeds this threshold, the current coordinates are recorded.

[0029] Furthermore, after collecting a sufficient number of anomalous sites, these discrete points need to be processed to identify the extent of the low-adhesion region. One approach is to project all anomalous sites onto a two-dimensional plane, then use adaptive concave envelope fitting to calculate the minimum convex polygon envelope of these points, and use this polygon as the low-adhesion boundary. Another approach is to perform cluster analysis on the anomalous sites, grouping spatially close anomalous sites into one class, and then performing polygon fitting on each class of anomalous sites, for example, using the Alpha Shape algorithm or boundary tracing algorithm, thereby generating one or more low-adhesion boundaries.

[0030] It should be noted that an adaptive concave envelope fitting is performed on each valid set of outliers. This is done by calculating the Alpha Shape based on the Delaunay triangulation combined with the dynamic rolling sphere radius, generating a concave polygon that accurately reflects the actual distribution contour of the point set, rather than using convex hull fitting. This is to avoid incorrectly including normal ground areas outside the low adhesion area within the boundary range.

[0031] Furthermore, to achieve cleaning while avoiding low-adhesion areas, a special path needs to be planned around them. One way to generate the involute arc segment is to construct a virtual base circle with a preset radius, centered on the geometric center of the low-adhesion boundary, for example, slightly larger than the radius of the cleaning robot's chassis. Then, starting from a point on the virtual base circle, an involute arc segment is generated by expanding along its tangent direction, allowing it to smoothly surround the low-adhesion boundary. Subsequently, the point in the initial traversal path closest to this involute arc segment is identified as the access point, and the initial path is connected to the involute arc segment by a straight line or a simple curve to form a modified traversal path.

[0032] Finally, precise control of the robot's attitude and position is required as it travels along the corrected traversal path. The robot's geometric center coordinates can be obtained in real time through its internal positioning system. Simultaneously, the edge position of its effective cleaning surface is determined based on the robot's structural parameters. During cleaning, the control system continuously calculates the distance between the geometric center and the low-adhesion boundary, ensuring this distance is always greater than zero, thus preventing the robot from entering the low-adhesion area. At the same time, by adjusting the robot's driving posture or fine-tuning the path, the edge of the effective cleaning surface can extend into the low-adhesion boundary, thereby achieving cleaning coverage of that area.

[0033] For example, as a specific implementation, suppose a cleaning robot is deployed at location A to perform cleaning operations. At this time, there is a low-adhesion area on the ground of location A due to long-term dampness or special material, which may be difficult to see directly with the naked eye.

[0034] First, after the cleaning robot starts, it acquires the environmental contour of the area to be cleaned and generates an initial traversal path based on this contour. Specifically, the robot scans location A using its onboard LiDAR to build an indoor map and identify static physical boundaries such as walls and furniture, thereby determining the environmental contour. Subsequently, the control system plans an initial bow-shaped traversal path covering the entire room based on the environmental contour, aiming to ensure that all cleanable areas are traversed by the robot.

[0035] At this point, the cleaning robot begins to travel along the initial traversal path. During travel, the current output torque and corresponding actual displacement are extracted in real time. For example, when the robot travels to a specific area at location A, its drive motor suddenly increases its output torque to maintain the preset speed, but the actual displacement reported by the wheel encoder is significantly less than expected, indicating that the wheel may be slipping. At this time, this location is marked as an anomaly, and its precise coordinates are recorded. Subsequently, the robot continues to travel along the initial path, and when it encounters a similar situation again, it will continue to record new anomaly locations. Through multiple journeys or data accumulation in exploration mode, multiple anomaly locations distributed within this low-adhesion area can be collected.

[0036] Furthermore, envelope fitting is performed on multiple adjacent anomalous sites to generate low-adhesion boundaries. Specifically, all collected anomalous sites are analyzed to identify spatially close groups of anomalous sites. For example, a clustering algorithm is used to classify the anomalous sites, and then geometric envelope fitting is performed on each group of sites to generate one or more polygonal contours. At this point, these generated polygonal contours can be identified as the low-adhesion boundaries of location A, clearly defining the area of ​​insufficient ground adhesion.

[0037] Based on this, an involute arc segment is generated around the low-adhesion boundary and connected to the initial traversal path to generate a modified traversal path. Specifically, a virtual base circle is constructed with the geometric centroid of the determined low-adhesion boundary as the center of the base circle and a radius that considers the robot chassis radius and safety buffer distance. Then, starting from the point of tangency between the virtual base circle and the low-adhesion boundary, a smooth involute arc segment is generated along its tangent direction, which can surround the entire low-adhesion boundary on the outside. At the same time, the path node in the initial traversal path closest to this involute arc segment is further identified as the connection reference point. By constructing introduction and exit transition curves, the involute arc segment is smoothly connected to the initial traversal path, thereby forming a modified traversal path that allows the geometric center of the cleaning robot to circumvent the low-adhesion boundary while guiding the effective cleaning surface to penetrate deeply.

[0038] Finally, the cleaning robot travels along the corrected traversal path. During travel, the control system monitors the distance between the robot's geometric center coordinates and the low-adhesion boundary in real time, ensuring that the robot's geometric center always remains outside the low-adhesion boundary, preventing the main body from entering the slippage area. Simultaneously, the system calculates the edge position of its effective cleaning surface based on the robot's structural parameters and adjusts the robot's driving posture so that the effective cleaning surface can penetrate into the low-adhesion boundary. For example, when the robot travels along the involute arc to the edge of the low-adhesion boundary, its chassis remains outside the boundary, but its side-extending brushes or suction nozzles can reach deep into the low-adhesion area for cleaning. In this way, the risk of the robot slipping in the low-adhesion area is avoided, while ensuring that the low-adhesion area is effectively cleaned, thus solving the problem of cleaning blind spots in traditional solutions.

[0039] Compared to existing technologies, this method first effectively and economically identifies low-adhesion areas on the ground by real-time monitoring of the cleaning robot's current output torque and actual displacement. In the example above, traditional methods struggle to directly detect visually imperceptible wet or low-friction areas on the ground at location A using vision or lidar. This method, however, directly captures the physical phenomenon of robot slippage by detecting the mismatch between the driving torque and actual displacement, thus identifying the abnormal location. This physical feedback-based detection mechanism avoids reliance on expensive dedicated sensors, thereby reducing system cost and complexity.

[0040] Secondly, this method generates a clear low-adhesion boundary by performing envelope fitting on the anomalous sites. Existing technologies often employ simple detour strategies after identifying low-adhesion areas, resulting in cleaning blind spots in these areas. However, this method accurately delineates the extent of low-adhesion areas through refined geometric fitting, providing precise geometric basis for subsequent path optimization.

[0041] Furthermore, this method introduces involute arc segments and incorporates them into the initial traversal path to generate a modified traversal path. In the example, the involute arc segments enable the cleaning robot to smoothly and controllably surround low-adhesion boundaries. This path planning approach provides superior trajectory continuity and driving stability compared to simple straight-line avoidance or coarse detours in existing technologies.

[0042] Most importantly, this application ensures that the geometric center of the cleaning robot remains outside the low-adhesion boundary while traveling along the modified traversal path, and the effective cleaning surface penetrates into the interior of the low-adhesion boundary. In the example, as the robot travels along the modified traversal path, its main body always safely avoids slippage areas, while the cleaning components can penetrate deep into the low-adhesion area for effective cleaning. This solves the problem of balancing "main body avoidance" and "cleaning coverage" in existing technologies, preventing cleaning omissions and thus improving cleaning efficiency and coverage. Overall, this method provides a cleaning robot path optimization scheme that balances safety, cleaning efficiency, and cost-effectiveness.

[0043] The path optimization method for a cleaning robot in this exemplary embodiment will be further described below.

[0044] In one embodiment of this application, the specific process of "obtaining the environmental contour of the area to be cleaned and generating an initial traversal path based on the environmental contour" in step S110 can be further explained in conjunction with the following description.

[0045] As described in the following steps Identify the static physical boundary of the environmental contour and extract the geometric centerline of the region where the environmental contour is located; Based on the static physical boundary, multi-level centripetal equidistant offsets are made towards the geometric central axis to generate nested distributed multi-ring topological equipotential lines; Obtain the tangential guide points between adjacent topological equipotential lines, and connect each loop of topological equipotential lines sequentially based on the tangential guide points to generate a continuous trajectory; The continuous trajectory is used as the initial traversal path.

[0046] It's important to note that static physical boundaries refer to the fixed outlines of obstacles or walls within a clean area, forming the basis of path planning. Geometric centerlines are the skeleton lines of the area, representing its central direction and helping to guide path generation. One approach is to scan the environment using LiDAR or a depth camera to construct a 2D or 3D map, then use image processing algorithms to identify the static physical boundaries. The geometric centerlines can then be calculated from the environmental outlines using skeleton extraction algorithms. Alternatively, static physical boundaries can be obtained from pre-imported CAD drawings or user-drawn area maps; the geometric centerlines can be calculated by connecting the farthest points within the area, taking the midpoint as the starting point of the centerline, and then extending outwards, or by using morphological refinement algorithms.

[0047] Furthermore, using the static physical boundary as a reference, multi-level centripetal equidistant offsets are made towards the geometric central axis to generate nested, multi-ring topological equipotential lines. The aim is to generate a series of pathlines similar in shape to the environmental boundary from the outside in, at uniform intervals. These generated topological equipotential lines can be understood as the potential trajectory along which the cleaning robot can travel, and the nested distribution ensures gradual coverage of the entire area. One implementation method is to use a distance transformation-based approach. First, the distance from each point within the area to the nearest static physical boundary is calculated. Then, based on a preset offset step size, a set of points with equal distance values ​​is extracted to form multi-level equipotential lines. Alternatively, a geometric offset algorithm can be used. Starting from the static physical boundary, continuous, equidistant offset operations are performed inwards. After each offset, the newly generated contour lines are smoothed to avoid sharp corners and self-intersections, thus obtaining a series of nested topological equipotential lines.

[0048] It should be noted that the offset should continue until the contour line shrinks to the point where it can no longer be offset. Therefore, to prevent missed detections, the offset step is an incremental amount. The static physical boundary is iteratively offset inward until the offset contour line shrinks to the point where it can no longer be offset, thus generating a multi-level initial contour line.

[0049] Furthermore, the rules for determining the offset step distance to prevent missed detection may include: taking the smaller value between the effective cleaning width of the cleaning robot and the drive wheel track as the offset step distance, so as to ensure that the distance between two adjacent topological equipotential lines does not exceed the effective cleaning width, while satisfying the geometric constraint that the drive wheels do not miss detection when the robot turns, thereby ensuring that the entire area is cleaned without omission.

[0050] Furthermore, tangential guide points are obtained between adjacent topological equipotential lines, and each loop of the topological equipotential lines is sequentially connected based on these tangential guide points to generate a continuous trajectory. This aims to ensure that the cleaning robot can smoothly transition from one equipotential line to another to form a complete cleaning path. By sequentially connecting the tangential guide points, a continuous trajectory covering the entire area can be constructed. One implementation method is to calculate the optimal connection points between adjacent topological equipotential lines based on the cleaning robot's turning capability and cleaning width. These connection points should allow the cleaning robot to switch from one equipotential line to the next with the smallest turning radius or the smoothest curve, while ensuring the continuity of cleaning coverage. For example, the closest point with the same normal direction between two equipotential lines can be found as the tangential guide point. Alternatively, points can be sampled evenly on each topological equipotential line, and then appropriate sampling points can be selected between adjacent equipotential lines for connection according to a preset connection strategy, such as spiral connection, bow connection, or S-shaped connection. When connecting the sampling points, a curve interpolation algorithm can be used to ensure the smoothness of the connection.

[0051] Finally, the continuous trajectory is used as the initial traversal path, aiming to determine the path of the cleaning robot's first traversal based on the continuous trajectory generated in the preceding steps. This initial traversal path is preliminarily optimized based on environmental geometry and robot dimensions, aiming to provide an efficient and comprehensive initial cleaning solution. One implementation method is to store the generated continuous trajectory data in the robot's path planning module as instructions for its subsequent driving control. Alternatively, the continuous trajectory can be converted into a sequence of instructions recognizable by the controller, such as a series of waypoints, speed commands, and turning commands, for the robot to execute.

[0052] For example, as a specific implementation, suppose a cleaning robot needs to clean an irregularly shaped room. First, the cleaning robot can scan the room using its onboard LiDAR to construct a two-dimensional point cloud map of the room. Using the point cloud data, static physical boundaries such as walls are identified through the RANSAC algorithm or Hough transform, and the contour lines of the static physical boundaries are extracted.

[0053] Furthermore, the geometric centerline of the room is obtained by calculating the skeleton lines of the room's outline. Based on this, multi-level centripetal equidistant offsets are made towards the geometric centerline inside the room, starting from the identified static physical boundary. For example, the offset step size can be set to half the effective cleaning width of the cleaning robot, and then multiple iterative offsets are performed to generate a series of topological equipotential lines that are similar in shape to the room boundary but gradually shrink inward. These topological equipotential lines are evenly distributed inside the room, like contour lines.

[0054] Finally, the system calculates and obtains tangential guide points between adjacent topological equipotential lines. For example, a point is found between the first and second equipotential lines so that after the robot travels to this point from the first circle, it can smoothly transition to the starting point of the second circle with an arc, and this arc maintains the tangential direction with both equipotential lines. Using methods such as B-spline curve interpolation, the tangential guide points are sequentially connected to the key points on each circle's topological equipotential line, forming a continuous trajectory from the outside in, spiral or arc-shaped. This continuous trajectory, for example, can consist of a series of ordered path points and corresponding heading angles, ultimately determining the initial traversal path for the cleaning robot's first cleaning operation.

[0055] The above technical solution identifies static physical boundaries and geometric centerlines, providing a precise benchmark for path planning. Multi-level centripetal equidistant offsets generate nested topological equipotential lines, ensuring comprehensive coverage of complex areas and avoiding cleaning blind spots. Obtaining tangential guide points and connecting the equipotential lines of each loop ensures good continuity and smoothness in the generated initial traversal path, reducing frequent turning and energy consumption during robot movement, thereby improving path execution efficiency and cleaning quality.

[0056] In one embodiment of this application, the specific process of "using the static physical boundary as a reference, performing multi-level centripetal equidistant offsets towards the geometric central axis to generate nested distributed multi-loop topological equipotential lines" can be further explained in conjunction with the following description.

[0057] As described in the following steps The effective cleaning width and drive wheel track of the cleaning robot are obtained, and the anti-missed detection offset step distance is determined based on the dimensional mapping relationship between the effective cleaning width and the drive wheel track. Using the anti-miss detection offset step size as the increment, the static physical boundary is iteratively offset inward to generate multi-level initial contour lines; The curvature radius of each node in the multi-level initial contour line is calculated through iteration, and sharp distortion areas with curvature radii smaller than a preset anti-slip steering threshold are extracted. The trajectory within the sharp distortion region is reconstructed by local curvature smoothing to generate nested, distributed multi-loop topological equipotential lines.

[0058] It should be noted that the effective cleaning width refers to the width of the area that a cleaning robot can effectively clean in one pass, usually determined by the physical dimensions of its cleaning brushes, suction nozzles, or wiping components. The drive wheel track refers to the distance between the two main drive wheels of the cleaning robot, directly affecting the robot's minimum turning radius and steering agility. The purpose of determining the anti-missed detection offset step distance is to ensure that the effective cleaning area completely covers the area to be cleaned when the cleaning robot travels along adjacent paths, avoiding cleaning blind spots or omissions. The anti-missed detection offset step distance can be determined based on empirical formulas, such as setting it as a percentage of the effective cleaning width to ensure a certain overlap between adjacent paths; alternatively, it can be calculated through simulation or theoretical analysis based on the cleaning robot's kinematic model, combined with its effective cleaning width and drive wheel track, to determine the maximum allowable offset distance while ensuring no missed cleaning.

[0059] Furthermore, inward iterative biasing refers to starting from the static physical boundary of the identified environmental contour and gradually generating nested contour lines inwards from the environment, using a previously determined anti-missed detection bias step size. These generated contour lines constitute a multi-level initial contour line, serving as the basis for generating topological equipotential lines. Methods for implementing inward iterative biasing can include: employing a distance-transform-based algorithm to generate equidistant lines by calculating the distance from each pixel to the boundary; or, offsetting each point on the static physical boundary along its inner normal direction, then connecting and smoothing the offset point set to generate a new inner contour line, repeating this process until the geometric centerline is reached or the entire area to be cleaned is covered.

[0060] Furthermore, the radius of curvature is a geometric quantity that measures the degree of curvature of a curve; the smaller the radius of curvature, the greater the curvature of the curve. For cleaning robots, an excessively small radius of curvature indicates that the path is too sharp, potentially exceeding its physical steering capability, leading to drive wheel slippage, path deviation, or inability to pass. The preset anti-slip steering threshold is the minimum acceptable radius of curvature value determined based on factors such as the cleaning robot's kinematic characteristics, drive system performance, and ground adhesion. Path segments below the anti-slip steering threshold are generally considered unsafe or inefficient to execute. The radius of curvature can be calculated by performing local circle fitting on discrete path nodes, or by using numerical methods such as the three-point normal intersection method or the finite difference method for estimation. Once the radius of curvature of each node is calculated, it can be compared with the preset anti-slip steering threshold to identify all path segments with a radius of curvature smaller than the anti-slip steering threshold; these path segments are the sharp distortion regions.

[0061] Finally, the purpose of local curvature smoothing reconstruction is to correct the path within the sharp distortion region, ensuring that its radius of curvature meets the turning requirements of the cleaning robot, thereby guaranteeing the path's executability. The reconstruction process should be performed locally to preserve the overall topology of the original path as much as possible. Common smoothing reconstruction methods include: refitting the path points within the sharp distortion region using B-spline curves, Bézier curves, or NURBS curves, adjusting control points to increase curve smoothness, and ensuring that the minimum radius of curvature of the reconstructed curve segment is greater than a preset anti-slip turning threshold; alternatively, signal processing methods such as Gaussian filtering and moving averages can be used to smooth the path point coordinates, and then reconnecting the smoothed points. After reconstruction, the smoothed path segments are connected to the undistorted segments of the original path, ultimately forming nested, multi-loop topological equipotential lines that satisfy the robot's kinematic constraints.

[0062] For example, as a specific implementation, consider a cleaning robot with an effective cleaning width of 0.4 meters and a drive wheel track of 0.25 meters. To ensure thorough cleaning, an anti-miss detection offset step distance of 0.35 meters can be empirically set, slightly smaller than the effective cleaning width, to guarantee a 0.05-meter overlap between adjacent paths. After obtaining the static physical boundary of the area to be cleaned, a distance transformation-based algorithm is used to generate multiple levels of initial contour lines in increments of 0.35 meters, for example, generating 10 nested contour lines.

[0063] At this point, the local radius of curvature of all discrete nodes on the 10 generated contour lines is calculated using the three-point circle method. Simultaneously, based on the kinematic model and driving performance of the cleaning robot, the preset anti-slip steering threshold is 0.2 meters, i.e., the minimum turning radius is 0.2 meters.

[0064] During the traversal calculation, if the radius of curvature of a path segment on a certain contour line is found to be less than 0.2 meters (e.g., a radius of curvature of 0.15 meters calculated at a corner), this path segment is marked as a sharp distortion region. For these marked sharp distortion regions, a cubic B-spline curve is used for local smoothing reconstruction. Specifically, the sharp distortion region and several path points on either side are selected as control points. By adjusting the parameters of the B-spline curve, a smooth curve segment is generated to replace the original sharp path segment. During generation, the system checks the minimum radius of curvature of the new curve segment in real time to ensure it is always greater than 0.2 meters, and that the new curve segment and the undistorted segment of the original path achieve a smooth G1 continuous connection. In this way, all sharp distortion regions are corrected into smooth curve segments, ultimately forming a nested distribution of multi-loop topological equipotential lines that satisfy the kinematic constraints of the cleaning robot.

[0065] Through the above technical solution, this application can effectively solve the problems of cleaning omissions and path inoperability in traditional equidistant offset methods. Specifically, by obtaining the effective cleaning width and drive wheel track of the cleaning robot, and determining the anti-miss detection offset step distance accordingly, it can be ensured that the generated path can achieve full coverage, thereby avoiding cleaning blind spots. At the same time, by traversing and calculating the radius of curvature of each node in the multi-level initial contour lines, and extracting sharp distortion areas with a radius of curvature smaller than the preset anti-slip steering threshold, path segments that pose obstacles to the movement of the cleaning robot can be identified. Furthermore, local curvature smoothing reconstruction is performed on the sharp distortion areas, so that the generated topological equipotential lines maintain cleaning coverage while their curvature meets the turning capability of the cleaning robot, thereby improving the executability of the path, the driving stability and operation efficiency of the cleaning robot, and effectively reducing problems such as slippage, jamming or incomplete cleaning caused by excessively sharp path turns.

[0066] In one embodiment of this application, the specific process of "performing envelope fitting on a plurality of adjacent anomalous sites to generate a low-adhesion boundary" in step S130 can be further described in conjunction with the following description.

[0067] As described in the following steps Extract the spatial distribution density of multiple abnormal sites, and cluster the abnormal sites that meet the preset connectivity threshold into a valid set of abnormal points; Using the chassis radius of the cleaning robot as the curvature constraint parameter, an adaptive concave envelope fitting is performed on the effective anomaly point set to generate a polygonal contour, and the polygonal contour is used as the low adhesion boundary. Calculate the initial topological center of the polygonal contour and verify the validity of the initial topological center; When the initial topological center is located outside the polygonal outline or within the area occupied by physical obstacles, the maximum inscribed circle inside the polygonal outline is calculated, and the center of the maximum inscribed circle is corrected to the geometric centroid of the low adhesion boundary. When the initial topological center is located inside the polygonal outline and does not fall into the area occupied by physical obstacles, the initial topological center is determined as the geometric centroid of the low adhesion boundary.

[0068] It should be noted that, firstly, for the multiple abnormal sites collected during the initial traversal path, spatial distribution density analysis is performed, and combined with a preset connectivity threshold, the abnormal sites are clustered into several valid abnormal point sets. This process can effectively filter out noise and isolated abnormal points, ensuring that the objects of subsequent processing are point groups that represent low-adhesion areas and have spatial continuity.

[0069] Furthermore, using the chassis radius of the cleaning robot as a key curvature constraint parameter, an adaptive concave envelope fitting is performed on the set of valid anomalies to generate a polygonal profile. This polygonal profile not only encloses all valid anomalies, but more importantly, the radius of curvature of its concave portion is always greater than or equal to the robot's chassis radius, geometrically ensuring that the cleaning robot can safely travel along its boundary without collision or jamming. At this point, the polygonal profile is determined as the low-adhesion boundary.

[0070] Furthermore, to determine a reliable navigation reference point within the low-adhesion boundary, the system calculates the initial topological center of the polygonal contour and verifies its validity. This verification process determines whether the initial topological center is located inside the polygonal contour and whether it overlaps with any known physical obstacle area. If the initial topological center does not meet these conditions (i.e., it is located outside the contour or within an obstacle area), the system further calculates the largest inscribed circle within the polygonal contour and corrects its center to the geometric centroid of the low-adhesion boundary. This is because the center of the largest inscribed circle, being the furthest from all boundaries, provides the safest internal reference point. Conversely, if the initial topological center passes the validity verification, it is directly determined as the geometric centroid of the low-adhesion boundary.

[0071] For example, as a specific implementation, suppose a cleaning robot is moving in a certain area, and the sensors detect a series of abnormal locations where the output torque is abnormal and the actual displacement deviates.

[0072] First, the detected anomalous sites are processed. For example, the DBSCAN clustering algorithm can be used, with a neighborhood radius of 0.5 meters and a minimum number of points of 5. Anomalous sites that are spatially close to each other and meet the density requirements are clustered to form multiple effective anomalous point sets, thereby filtering out unrepresentative scattered anomalous points.

[0073] Furthermore, for each set of valid anomalies, the system uses the chassis radius of the cleaning robot as a curvature constraint parameter to perform adaptive concave envelope fitting. Specifically, for example, if the chassis radius is 0.2 meters, the Alpha Shape algorithm can be used. By adjusting the Alpha parameter to match the chassis radius, a polygonal contour that can enclose all valid anomalies and whose concave curvature radius is not less than 0.2 meters can be generated. This polygonal contour can then be determined as the low adhesion boundary.

[0074] Furthermore, the initial topological center of the polygon outline is calculated. For example, the average coordinates of all vertices of the polygon can be calculated as the initial topological center, and then its validity is verified. If the initial topological center is determined to be outside the polygon outline by ray casting, or if it is found to fall within a known area occupied by a physical obstacle by querying preset map information, a correction mechanism is activated. In this case, the system calculates the largest inscribed circle inside the polygon outline and uses its center as the geometric centroid of the corrected low-adhesion boundary. For example, a grid search can be performed inside the polygon to calculate the shortest distance from each grid point to the polygon boundary, and the point with the largest distance can be used as the center. Conversely, if the initial topological center is inside the polygon outline and does not fall within an area occupied by a physical obstacle, then the initial topological center is directly determined as the geometric centroid of the low-adhesion boundary.

[0075] Through the above technical solution, this application can transform discrete, potentially noisy, abnormal sites detected during the movement of a cleaning robot into low-adhesion boundaries that take into account the robot's own size constraints, possess reliable geometric centroids, and have well-defined geometric shapes. This solves the problems in traditional methods where discrete abnormal points are difficult to use directly for path planning, or where boundaries formed by simple connections may be unsafe or inaccurate. Specifically, by performing spatial density clustering on abnormal sites, the true low-adhesion regions can be effectively identified and focused on, thus avoiding misjudgments of isolated noise points. By using the chassis radius of the cleaning robot as a curvature constraint parameter for adaptive concave envelope fitting, it can be ensured that the generated low-adhesion boundary is geometrically feasible, allowing the robot to safely move along its periphery without collisions or jamming. Furthermore, the calculation and validity verification of the geometric centroid of the low adhesion boundary, and the correction by the maximum inscribed circle when necessary, can ensure that the determined geometric centroid is always located within the effective interior of the low adhesion area and away from obstacles, providing a safe reference point for path correction. This allows the cleaning robot to plan its correction traversal path more safely, effectively avoiding the low adhesion area while ensuring thorough cleaning of the surrounding area, thus improving the efficiency and safety of cleaning operations.

[0076] In one embodiment of this application, the specific process of "using the chassis radius of the cleaning robot as the curvature constraint parameter, performing adaptive concave envelope fitting on the effective anomaly point set to generate a polygonal contour, and using the polygonal contour as the low adhesion boundary" can be further explained in conjunction with the following description.

[0077] As described in the following steps Based on the spatial coordinates of the effective anomaly point set, a spatial triangulation network containing multiple edge triangles is constructed; The dynamic rolling ball radius is generated by adding the chassis radius of the cleaning robot to the preset safety buffer distance. The circumcircle radius of all edge triangles in the spatial triangulation is calculated by traversing the network, and the circumcircle radius is then threshold-filtered using the dynamic rolling ball radius. Remove the edge triangles whose circumcircle radius is greater than the radius of the dynamic rolling ball, and extract the topological extension edges of the remaining edge triangles; The sequentially connected topological extension edges are subjected to smooth interpolation to generate a polygonal profile in which the global concave curvature radius is greater than the chassis radius.

[0078] It should be noted that by converting the discrete set of valid outliers into a spatial triangulation network, the topological relationships and spatial distribution between these points can be effectively represented. The methods for constructing the spatial triangulation network can include the Delaunay triangulation algorithm or the Voronoi diagram dual algorithm. These algorithms ensure that the generated triangles have good geometric properties, such as maximizing the minimum angle, thereby avoiding the generation of elongated or distorted triangles, laying the foundation for subsequent edge recognition and curvature calculation.

[0079] Furthermore, the dynamic rolling ball radius is used to consider the actual size and operational safety of the cleaning robot during the envelope fitting process. The chassis radius represents the minimum external dimensions of the cleaning robot body, while the preset safety buffer distance provides additional safety margins to cope with positioning errors, motion inertia, or environmental uncertainties. The generation of the dynamic rolling ball radius ensures that the subsequently generated low-adhesion boundary is geometrically capable of accommodating the cleaning robot's safe passage and effectively cutting into it during cleaning operations.

[0080] Furthermore, an edge triangle is a triangle that has at least one side that is not shared with other triangles, and it is usually located on the outer contour of a point set. Calculating the circumcircle radius of an edge triangle and comparing it with the radius of a dynamic rolling ball can effectively filter out triangles with excessively large circumcircle radii. An excessively large circumcircle radius usually indicates that the area covered by the triangle is relatively flat or too open, making it unsuitable as part of a concave envelope, or that its curvature does not meet the robot's passage requirements.

[0081] Furthermore, by eliminating edge triangles that do not meet the criteria, areas that may lead to overly smooth or inaccurate boundaries can be removed. The topological extension edges of the remaining edge triangles refer to the edges of these triangles that are located outside the point set. These topological extension edges together constitute the preliminary outline of the effective set of outlier points. Since they have passed the preliminary curvature screening, their compatibility with the size of the cleaning robot can be ensured.

[0082] Furthermore, smoothing interpolation can be performed using methods such as B-spline curves, NURBS curves, or Catmull-Rom splines to connect a series of discrete line segments into a smooth curve. This process eliminates jagged edges or sharp corners caused by discrete points and line segments, resulting in a more rounded low-adhesion boundary. Simultaneously, by ensuring that the global concave curvature radius is greater than the chassis radius of the cleaning robot, it is ensured that the robot's chassis will not interfere with the boundary when traveling along it, thus avoiding collisions and improving navigation safety.

[0083] For example, as a specific implementation, after obtaining the spatial coordinates of the effective anomaly set, the Delaunay triangulation algorithm can be used to construct a spatial triangulation network, generating triangles with the minimum angles maximized, thereby optimizing the mesh quality. For instance, if the chassis radius of the cleaning robot is 0.2 meters and the preset safety buffer distance is 0.05 meters, the dynamic rolling ball radius can be set to 0.25 meters.

[0084] Furthermore, when traversing the edge triangles in the spatial triangulation, the circumcircle radius of each edge triangle is calculated. If the circumcircle radius of an edge triangle is greater than 0.25 meters, the region represented by the edge triangle is considered too open or its curvature does not meet the requirements, and it is discarded.

[0085] Based on this, the topological extension edges are extracted from the remaining edge triangles. These topological extension edges collectively delineate the initial boundary of the low-adhesion region. To make the boundary smoother and meet curvature requirements, cubic B-spline curves can be used to smooth and interpolate these topological extension edges. During the interpolation process, iterative optimization or parameter adjustment ensures that the radius of curvature of any concave portion of the final generated polygonal contour is greater than 0.2 meters, which is greater than the chassis radius of the cleaning robot. This allows the cleaning robot's chassis to safely avoid the concave areas of the boundary even when operating near the low-adhesion boundary.

[0086] Through the above technical solutions, the scheme in this embodiment can fully consider the actual physical size and operational safety requirements of the cleaning robot when generating low-adhesion boundaries. Specifically, by constructing a spatial triangular mesh and combining it with dynamic rolling ball radius for curvature filtering, the generation of low-adhesion boundaries containing sharp depressions or excessively small curvature can be effectively avoided. Smooth interpolation further ensures the continuity and roundness of the boundary, making the generated low-adhesion boundary more geometrically reasonable and safe. This not only improves the accuracy of the low-adhesion boundary but also ensures that the chassis of the cleaning robot can always safely avoid the concave part of the low-adhesion boundary when traveling along the corrected traversal path, thereby reducing the risk of collision between the robot and the edge of the low-adhesion area, improving the reliability of path planning and the safety of cleaning operations.

[0087] In one embodiment of this application, the specific process following the step of "smoothing interpolation of the sequentially connected topological extension edges to generate a polygonal profile in which the global concave curvature radius is greater than the chassis radius" can be further described in conjunction with the following description.

[0088] As described in the following steps The concave curvature radius of each curve segment in the polygonal contour is calculated through iteration, and local concave segments with concave curvature radii smaller than the chassis radius are extracted. Extend the two endpoints of the local concave segment outward to construct a transition arc tangent to the radius of the chassis outside the polygonal outline; The transition arc bridges the two side contour lines of the local concave segment, generating a corrected contour segment with an inward curvature radius greater than the chassis radius. The low-adhesion boundary is updated by replacing the local concave segment in the polygonal contour with the corrected contour segment.

[0089] It should be noted that calculating the concave curvature radius of each curve segment in the polygonal contour involves analyzing the generated polygonal contour and performing local curve fitting on multiple consecutive points along the contour, such as using circular arc fitting or spline curve fitting, to calculate the curvature radius of these local curves. This step aims to identify areas within the contour that may pose a potential risk to the cleaning robot's movement. Extracting local concave segments with a concave curvature radius smaller than the chassis radius involves comparing the calculated curvature radius with the cleaning robot's chassis radius. When a local area is found to have a concave curvature radius smaller than the cleaning robot's chassis radius, this area is marked as a local concave segment, indicating that the curvature in this area is too severe, and the cleaning robot's chassis may not be able to pass smoothly, posing a collision risk.

[0090] Furthermore, extending the two endpoints of the local concave segment outwards means that after identifying the local concave segment, in order to provide sufficient space for smoothing, the starting and ending points of the concave segment are appropriately displaced outwards along their respective normal or tangent directions of the polygonal contour. This extension operation creates the geometric conditions for constructing the transition arc. Constructing a transition arc tangent to the chassis radius outside the polygonal contour means using the extended endpoints as a reference to generate an arc in the outer region of the polygonal contour. The radius of this arc is set to be equal to or slightly larger than the chassis radius of the cleaning robot, and this arc needs to be tangent to the contour segment where the extended endpoints are located. Its function is to provide a smooth and safe transition path, ensuring that the chassis of the cleaning robot can pass through the area without obstruction.

[0091] Furthermore, bridging the two sides of the local concave segment with a transition arc means using the constructed transition arc as a new contour segment to connect the unmodified contour segments on both sides of the original local concave segment. In this way, the transition arc can effectively replace the original sharp concave segment, forming a continuous and smooth contour. Generating a corrected contour segment with a concave radius of curvature greater than the chassis radius means that after bridging with the transition arc, the new contour segment's concave radius of curvature will meet the requirement of being greater than the chassis radius of the cleaning robot, thus ensuring that the corrected contour line has sufficient curvature at any concave point, preventing the cleaning robot from getting stuck or colliding during operation.

[0092] Finally, replacing the locally concave segments in the polygonal contour with modified contour segments means removing the identified locally concave segments from the geometric data and replacing them with modified contour segments that have a larger concave radius of curvature. Updating the low-adhesion boundary means that after completing the above replacement operations, the new, locally smoothed polygonal contour is used as the final low-adhesion boundary. This better guides the cleaning robot's path planning and ensures its safe movement along the edges of low-adhesion areas.

[0093] For example, as a specific implementation, suppose that after performing envelope fitting on the set of valid outliers, a polygonal contour is generated, which exhibits a relatively sharp "V"-shaped indentation in a certain region.

[0094] First, a discrete point method can be used to sample a series of points along the path of the polygonal contour at a preset step size. For every three consecutive sampling points, an arc can be fitted and its radius of curvature calculated, thereby traversing and calculating the concave radius of curvature of the "V"-shaped recessed area. If the concave radius of curvature of the "V"-shaped recess is found to be smaller than the chassis radius of the cleaning robot, it is identified as a local recessed segment.

[0095] Furthermore, the two endpoints of the local concave segment can be extended outwards by a certain distance along their respective normal directions, for example, by a predetermined safety margin. Then, using the chassis radius of the cleaning robot as the radius, a smooth transition arc is constructed on the outer side of the polygonal contour between the extended endpoints. This transition arc is tangent to the contour segment where the extended endpoints are located. For example, the center positions determined by the two extended endpoints and the chassis radius can be calculated, and then the transition arc can be drawn.

[0096] Finally, the newly generated transition arc replaces the original "V"-shaped recess, thus updating the low-adhesion boundary. After the above processing, the original sharp "V"-shaped recess is replaced by a smooth arc, ensuring that the corrected low-adhesion boundary has sufficient curvature in any concave area, allowing the cleaning robot to safely bypass the area.

[0097] Through the above technical solution, this application can effectively solve the problem that the radius of curvature of local concavities may be smaller than the radius of the cleaning robot chassis when generating low-adhesion boundaries. Based on the initial generation of the polygonal contour, the contour lines are further inspected to identify and correct all sharp concave areas that could lead to collisions with the cleaning robot chassis. By extending the endpoints of the local concave segments outward and constructing transition arcs tangent to the chassis radius for bridging, it can be ensured that the corrected low-adhesion boundary has sufficient curvature at any concave location. This avoids the risk of the cleaning robot's chassis scraping or colliding with the low-adhesion boundary while traveling along the corrected path, thus improving the safety and path planning reliability of the cleaning robot when operating at the edges of complex low-adhesion areas.

[0098] In one embodiment of this application, the specific process of step S140, "generating an involute arc segment on the periphery of the low adhesion boundary and connecting the involute arc segment to the initial traversal path to generate a corrected traversal path," can be further explained in conjunction with the following description.

[0099] As described in the following steps An involute base circle is constructed with the geometric centroid as the center of the base circle and the sum of the chassis radius of the cleaning robot and the preset safety buffer distance as the base circle radius. Extract the tangent point or the nearest contact point between the involute base circle and the low adhesion boundary contour line, and take the tangent point or the nearest contact point as the starting point, expand along the tangent direction of the involute base circle at the starting point to generate an involute arc segment surrounding the low adhesion boundary. Identify the path node in the initial traversal path that is closest to the low adhesion boundary, and use the path node as the access reference point; Construct an introduction transition curve from the access reference point to the starting point of the involute arc segment, and an exit transition curve for the subsequent path from the ending point of the involute arc segment to the access reference point; The introduced transition curve, the involute arc segment, and the exit transition curve are connected sequentially, and combined with the initial traversal path, to generate the corrected traversal path.

[0100] It should be noted that the center of the involute base circle is usually the geometric center of the low-adhesion boundary. This ensures the central symmetry of the path planning or is considered in terms of the center of gravity. Its radius combines the chassis radius and safety buffer distance of the cleaning robot, thus ensuring that the robot's chassis maintains a safe distance from the low-adhesion boundary when traveling along the involute curve, avoiding collisions or getting stuck. The involute base circle can be constructed by defining its center coordinates and radius in two-dimensional or three-dimensional space using software algorithms, or it can be directly calculated based on preset geometric parameters.

[0101] Furthermore, the tangent points between the involute base circle and the low-adhesion boundary contour line are first extracted. When a tangent relationship exists, this tangent point is directly taken as the unfolding starting point; when there is no tangent point, the point on the involute base circle closest to the low-adhesion boundary contour line is taken as the unfolding starting point. After determining the starting point, the tangent direction of the involute base circle at this starting point is used as the initial unfolding direction, and involute arc segments are generated outwards, so that the involute arc segments can surround the entire low-adhesion boundary on the outside. Among them, the involute has unique geometric characteristics: the normal line at any point on it is tangent to the base circle, that is, the tangent line of the involute at any point is perpendicular to the line connecting that point to the corresponding tangent point on the base circle. This allows the cleaning robot traveling along the involute to maintain a certain geometric relationship with the base circle throughout the entire arc segment, thereby achieving a smooth and continuous bypass trajectory. The generation of involute arc segments can be achieved through the parametric equation method. That is, with the center of the base circle as the origin and the radius of the base circle as the parameter, the coordinates of discrete points on the arc segment are calculated according to the standard parametric equation of the involute. Then, these discrete points are connected in sequence to form a complete involute arc segment.

[0102] Furthermore, identifying the path node closest to the low-adhesion boundary in the initial traversal path and using this path node as the access reference point is to determine the connection point between the corrected path and the initial path. Selecting the closest path node as the access reference point minimizes the scope of path correction, maintains the integrity of the initial traversal path, and reduces the complexity of path correction. The identification of path nodes can be achieved by calculating the minimum distance from all nodes on the initial traversal path to the low-adhesion boundary and selecting the node with the smallest distance.

[0103] Furthermore, transition curves can avoid sharp turns or discontinuities on the path, thus ensuring the stability and safety of the cleaning robot during its movement. Both introducing and deriving transition curves can be implemented using various curve types, such as Bézier curves, B-spline curves, or cubic spline curves. These curves can achieve smooth connections through control points or nodes.

[0104] Finally, by connecting the aforementioned curve segments in a logical order, a complete and continuous corrected path can be formed. This, combined with the initial traversal path, means that in low-adhesion boundary areas, the original path is replaced by the corrected path, while in other areas, the initial traversal path is used. This combination ensures that the cleaning robot can efficiently complete the cleaning task of the entire area while safely handling low-adhesion areas.

[0105] For example, as a specific implementation, assume that the cleaning robot has identified a low-adhesion boundary in a certain area and determined its geometric centroid coordinates to be (X1, Y1). The chassis radius of the cleaning robot is R, and the preset safety buffer distance is S.

[0106] First, the system constructs an involute base circle with (X1, Y1) as the center and R+S as the radius. Then, the system calculates the point of tangency between the contour line of the low adhesion boundary and this involute base circle, such as point P.

[0107] Based on this, starting from point P, an involute arc segment is generated through numerical calculation along its outward normal direction. This arc segment can completely enclose the low-adhesion boundary. At the same time, the system will traverse all path nodes on the initial traversal path, calculate the minimum distance from these path nodes to the low-adhesion boundary, and select the node P1 with the smallest distance as the access reference point.

[0108] The system then constructs an introductory transition curve from P1 to the starting point of the involute arc segment, such as a cubic Bézier curve, to ensure a smooth path. Similarly, it constructs an outtroductory transition curve from the ending point of the involute arc segment to the subsequent path from P1.

[0109] Finally, the portion of the initial traversal path that overlaps with the low adhesion boundary is replaced with a combination of an introduction transition curve, an involute arc segment, and an exit transition curve, thereby generating a complete modified traversal path.

[0110] Through the aforementioned technical solutions, the cleaning robot can perform cleaning operations around low-adhesion areas in a geometrically precise and dynamically safe manner. Specifically, the modified traversal path ensures that the geometric center of the cleaning robot always maintains a safe distance from the low-adhesion boundary, effectively avoiding potential slippage or getting stuck risks. Simultaneously, the involute arc design allows the effective cleaning surface of the cleaning robot to penetrate into the interior of the low-adhesion boundary, thus achieving thorough cleaning of the edge of this area and avoiding the creation of cleaning blind spots. Furthermore, the introduction and exit of transition curves ensure the smoothness and continuity of the path, effectively improving the cleaning robot's driving stability and operational efficiency in complex environments.

[0111] In one embodiment of this application, the specific process of step S150, "controlling the cleaning robot to travel along the modified traversal path, so that the geometric center of the cleaning robot remains outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary," can be further described in conjunction with the following description.

[0112] As described in the following steps Obtain the coordinates of the chassis geometric center of the cleaning robot and the coordinates of the edge position of the effective cleaning surface; Calculate the shortest distance from the geometric center of the chassis to the contour line of the low adhesion boundary, as the center avoidance margin; Calculate the penetration depth from the edge position of the effective cleaning surface to the low adhesion boundary contour line, as the cleaning cut-in amount; The center avoidance margin and the cleaning cut-in amount are monitored in real time. When the center avoidance margin is less than a preset safety threshold or the cleaning cut-in amount is less than a preset effective threshold, the driving posture or driving trajectory of the cleaning robot is adjusted. Maintain the center clearance margin greater than zero and the cleaning cut-in amount greater than the preset effective threshold until the cleaning operation of the area surrounding the low adhesion boundary is completed.

[0113] It's important to clarify that obtaining the geometric center coordinates of the cleaning robot's chassis and the edge coordinates of the effective cleaning surface refers to acquiring the position of the geometric center point of the robot's main structure in two-dimensional or three-dimensional space, as well as the spatial position of the outermost boundary point of the robot's components that are in contact with or near the ground during actual cleaning operations. These coordinates form the basis for geometric relationship calculations, used to assess the robot's position relative to low-adhesion boundaries and cleaning coverage. For example, the robot's global pose can be acquired in real time through its internal positioning system, such as SLAM algorithms combined with LiDAR, vision sensors, and IMUs, and then the chassis geometric center and the edge position of the effective cleaning surface can be calculated based on the robot's structural model. Alternatively, accurate coordinate information can be obtained by pre-calibrating the robot's structural parameters, combining data from encoders, odometers, and other sensors for trajectory calculation, and periodically calibrating with external positioning benchmarks such as UWB and GPS.

[0114] Furthermore, the center avoidance margin is used to quantify the safe distance between the robot body and the danger zone, ensuring that the robot will not slip or get stuck when the chassis enters a low-adhesion area. For example, a geometric algorithm, such as a shortest distance algorithm from a point to a polygon or curve, can be used to calculate the distance from the geometric center point of the chassis to all points on the low-adhesion boundary contour line under the robot's current pose, and then take the minimum value. Alternatively, a distance field or potential field method can be used to pre-calculate the distance information around the low-adhesion boundary, and the robot only needs to query the distance value at the corresponding position during runtime.

[0115] Furthermore, the cleaning penetration depth is used to evaluate the effectiveness of the cleaning operation, ensuring that the cleaning surface can fully cover and clean the edges of the low-adhesion area, avoiding cleaning blind spots. For example, a geometric intersection detection algorithm can be used to determine the intersection points of the edge segments of the effective cleaning surface and the contour lines of the low-adhesion boundary, and calculate the length of the edge segments extending beyond the boundary. Alternatively, Boolean operations or polygon clipping algorithms can be used to calculate the intersection area of ​​the effective cleaning surface and the low-adhesion boundary, and the penetration depth can be indirectly calculated based on the ratio of the intersection area to the area of ​​the effective cleaning surface.

[0116] Furthermore, by continuously monitoring key parameters, potential deviations from the robot can be detected and corrected in a timely manner, ensuring both safety and cleaning effectiveness. For example, the aforementioned calculations can be performed periodically, and the results compared with preset thresholds. Once a condition is triggered, new speed commands, steering angles, or local path plans can be generated based on the magnitude and direction of the deviation to adjust the robot's posture or trajectory. Alternatively, advanced control strategies such as model predictive control or fuzzy control can be employed. Using center avoidance margin and cleaning cut-in amount as inputs to the controller, combined with existing robot dynamics models, the robot's state over a future period can be predicted, and the control output optimized to minimize deviations and meet constraints.

[0117] It should be noted that the preset safety threshold should be greater than zero so that the control system can trigger attitude adjustment before the geometric center of the cleaning robot reaches the low adhesion boundary, leaving sufficient response margin for the control system and avoiding the geometric center from crossing the low adhesion boundary and entering the slipping area due to control delay.

[0118] Finally, maintaining a center avoidance margin greater than zero and a cleaning cut-in depth greater than a preset effective threshold until the cleaning operation around the low-adhesion boundary is completed is the ultimate goal and continuous requirement of the entire control process, ensuring that the robot remains in a safe and effective cleaning state throughout the entire cleaning operation. For example, the control system will continuously execute the above monitoring and adjustment process until the robot completes the cleaning task around the low-adhesion boundary. At this point, the cleaning robot has completely cleaned the area along the corrected traversal path, and all relevant areas are effectively covered. Alternatively, this control strategy can be activated at the start of the cleaning task and deactivated at the end of the task through the task management module. The determination of task completion can be based on path completion, area coverage, or user commands.

[0119] For example, as a specific implementation, suppose the cleaning robot is a floor cleaning robot with a circular chassis and a front-mounted rotating brush. This floor cleaning robot is equipped with a LiDAR for environmental perception and localization, an inertial measurement unit for attitude perception, and a wheel encoder for odometry. During the cleaning process, the floor cleaning robot first obtains its global pose, including X and Y coordinates and heading angle, using the LiDAR and SLAM algorithm. Based on a pre-stored robot structure model, it calculates the geometric center coordinates of the chassis and the coordinates of the edge of the rotating brush in the current pose.

[0120] At this point, the system uses geometric algorithms, such as calculating the shortest distance from a point to a polygon, to determine the shortest distance between the chassis's geometric center and the identified low-adhesion boundary, using this distance as the center avoidance margin. Simultaneously, by calculating the overlap or penetration length between the edge of the rotating brush and the low-adhesion boundary contour, the cleaning cut-in amount can be obtained.

[0121] Based on this, the system monitors these two values ​​in real time at a frequency of 10 times per second. If the center avoidance margin is less than the preset safety threshold, such as 5 cm, or the cleaning cut-in is less than the preset effective threshold, such as 2 cm, the robot control system will activate the local path planner, such as the dynamic window method (DWA), to generate a small local trajectory correction, such as slightly adjusting the steering angle or making a lateral offset, so as to increase the center avoidance margin or achieve the required cleaning cut-in.

[0122] During the cleaning process, the ground cleaning robot will continuously perform the above monitoring and adjustments until the cleaning work around the low adhesion boundary is completed, ensuring safety and cleaning effectiveness throughout the process.

[0123] Through the above technical solution, the cleaning robot can perceive and precisely control its relative position to the low-adhesion boundary in real time, effectively avoiding safety issues such as slippage and getting stuck caused by the robot chassis entering the low-adhesion area due to path execution deviation. Simultaneously, the solution in this embodiment ensures that the effective cleaning surface of the cleaning robot can always fully penetrate the interior of the low-adhesion boundary, thereby thoroughly removing stains from the boundary area and avoiding blind spots and incomplete cleaning. This dynamic adaptive control strategy significantly improves the safety, reliability, and efficiency of cleaning operations, enabling the cleaning robot to complete tasks stably and efficiently even in complex and changing environments.

[0124] As the system implementation is basically similar to the method implementation, it is described in a relatively simple way. For relevant details, please refer to the description of the method implementation.

[0125] Reference Figure 2 This application illustrates a path optimization system for a cleaning robot according to an embodiment of the present application, which specifically includes the following modules: The path generation module 210 is used to obtain the environmental contour of the area to be cleaned and generate an initial traversal path based on the environmental contour. The location determination module 220 is used to control the cleaning robot to travel along the initial traversal path, extract the current output torque and the corresponding actual displacement, and determine abnormal locations based on the current output torque and the actual displacement. Centroid extraction module 230 is used to perform envelope fitting on multiple adjacent anomalous sites to generate low-adhesion boundaries; The path correction module 240 is used to generate an involute arc segment on the periphery of the low adhesion boundary and connect the involute arc segment to the initial traversal path to generate a corrected traversal path. The driving control module 250 is used to control the cleaning robot to travel along the modified traversal path, so that the geometric center of the cleaning robot is kept outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.

[0126] Reference Figure 3 The illustration shows a computer electronic device for implementing a path optimization method for a cleaning robot according to this application, which may specifically include the following: The aforementioned computer electronic device 1 is manifested in the form of a general-purpose computing device. The components of the computer electronic device 1 may include, but are not limited to: one or more processors or processing units 3, memory 8, and a bus 4 connecting different system components (including memory 8 and processing unit 3).

[0127] Bus 4 represents one or more of several bus architectures, including memory buses or memory controllers, peripheral buses, graphics acceleration ports, processors, or local buses using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Audio / Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0128] Computer electronic device 1 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by computer electronic device 1, including volatile and non-volatile media, removable and non-removable media.

[0129] Memory 8 may include computer system readable media in the form of volatile memory, such as random access memory 9 and / or cache memory 10. Computer electronic device 1 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 11 may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as a "hard disk drive"). Although Figure 3As not shown, a disk drive for reading and writing to a removable non-volatile disk (such as a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 4 via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules 13 configured to perform the functions of the embodiments of this application.

[0130] A program / utility 12 having a set (at least one) of program modules 13 may be stored, for example, in memory. Such program modules 13 include—but are not limited to—an operating system, one or more application programs, other program modules 13, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 13 typically perform the functions and / or methods described in the embodiments of this application.

[0131] The computer electronic device 1 can also communicate with one or more external devices 2 (e.g., keyboard, pointing device, display 7, camera, etc.), and with one or more devices that enable an operator to interact with the computer electronic device 1, and / or with any device that enables the computer electronic device 1 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through the I / O interface 6. Furthermore, the computer electronic device 1 can also communicate with one or more networks (e.g., local area network (LAN)), wide area network (WAN), and / or public networks (e.g., the Internet) through the network adapter 5. Figure 3 As shown, network adapter 5 communicates with other modules of computer electronic device 1 via bus 4. It should be understood that, although... Figure 3 Not shown, it may be combined with other hardware and / or software modules, including but not limited to: microcode, device drivers, redundant processing unit 3, external disk drive array, RAID system, tape drive and data backup storage system 11, etc.

[0132] The processing unit 3 executes various functional applications and data processing by running programs stored in memory 8, such as implementing a path optimization method for a cleaning robot provided in the embodiments of this application.

[0133] That is, when the processing unit 3 executes the above program, it achieves the following: obtaining the environmental contour of the area to be cleaned, and generating an initial traversal path based on the environmental contour; controlling the cleaning robot to travel along the initial traversal path, extracting the current output torque and the corresponding actual displacement, and determining abnormal sites based on the current output torque and the actual displacement; performing envelope fitting on multiple adjacent abnormal sites to generate a low adhesion boundary; generating an involute arc segment on the periphery of the low adhesion boundary, and connecting the involute arc segment to the initial traversal path to generate a corrected traversal path; controlling the cleaning robot to travel along the corrected traversal path, so that the geometric center of the cleaning robot remains outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.

[0134] In this application embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a path optimization method for a cleaning robot as provided in all embodiments of this application.

[0135] That is, when the program is executed by the processor, it performs the following: acquiring the environmental contour of the area to be cleaned, and generating an initial traversal path based on the environmental contour; controlling the cleaning robot to travel along the initial traversal path, extracting the current output torque and the corresponding actual displacement, and determining abnormal sites based on the current output torque and the actual displacement; performing envelope fitting on multiple adjacent abnormal sites to generate a low-adhesion boundary; generating an involute arc segment on the periphery of the low-adhesion boundary, and connecting the involute arc segment to the initial traversal path to generate a corrected traversal path; controlling the cleaning robot to travel along the corrected traversal path, so that the geometric center of the cleaning robot remains outside the low-adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low-adhesion boundary.

[0136] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that includes or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0137] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0138] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the operator's computer, partially on the operator's computer, as a standalone software package, partially on the operator's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the operator's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider). The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably.

[0139] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0140] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0141] The path optimization method and system for a cleaning robot provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A path optimization method for a cleaning robot, characterized in that, The method includes the following steps: Obtain the environmental contour of the area to be cleaned, and generate an initial traversal path based on the environmental contour; The cleaning robot is controlled to travel along the initial traversal path, the current output torque and the corresponding actual displacement are extracted, and abnormal locations are determined based on the current output torque and the actual displacement. Envelope fitting is performed on multiple adjacent anomalous sites to generate low-adhesion boundaries; An involute arc segment is generated around the low adhesion boundary, and the involute arc segment is connected to the initial traversal path to generate a corrected traversal path. The cleaning robot is controlled to travel along the modified traversal path, so that the geometric center of the cleaning robot remains outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.

2. The path optimization method for a cleaning robot according to claim 1, characterized in that, The step of obtaining the environmental contour of the area to be cleaned and generating an initial traversal path based on the environmental contour specifically includes: Identify the static physical boundary of the environmental contour and extract the geometric centerline of the region where the environmental contour is located; Based on the static physical boundary, multi-level centripetal equidistant offsets are made towards the geometric central axis to generate nested distributed multi-ring topological equipotential lines; Obtain the tangential guide points between adjacent topological equipotential lines, and connect each loop of topological equipotential lines sequentially based on the tangential guide points to generate a continuous trajectory; The continuous trajectory is used as the initial traversal path.

3. The path optimization method for a cleaning robot according to claim 2, characterized in that, The step of generating nested, distributed multi-loop topological equipotential lines by performing multi-level centripetal equidistant offsets towards the geometric central axis based on the static physical boundary specifically includes: The effective cleaning width and drive wheel track of the cleaning robot are obtained, and the anti-missed detection offset step distance is determined based on the dimensional mapping relationship between the effective cleaning width and the drive wheel track. Using the anti-miss detection offset step size as the increment, the static physical boundary is iteratively offset inward to generate multi-level initial contour lines; The curvature radius of each node in the multi-level initial contour line is calculated through iteration, and sharp distortion areas with curvature radii smaller than a preset anti-slip steering threshold are extracted. The trajectory within the sharp distortion region is reconstructed by local curvature smoothing to generate nested, distributed multi-loop topological equipotential lines.

4. The path optimization method for a cleaning robot according to claim 1, characterized in that, The step of performing envelope fitting on multiple adjacent anomalous sites to generate low-adhesion boundaries specifically includes: Extract the spatial distribution density of multiple abnormal sites, and cluster the abnormal sites that meet the preset connectivity threshold into a valid set of abnormal points; Using the chassis radius of the cleaning robot as the curvature constraint parameter, an adaptive concave envelope fitting is performed on the effective anomaly point set to generate a polygonal contour, and the polygonal contour is used as the low adhesion boundary. Calculate the initial topological center of the polygonal contour and verify the validity of the initial topological center; When the initial topological center is located outside the polygonal outline or within the area occupied by physical obstacles, the maximum inscribed circle inside the polygonal outline is calculated, and the center of the maximum inscribed circle is corrected to the geometric centroid of the low adhesion boundary. When the initial topological center is located inside the polygonal outline and does not fall into the area occupied by physical obstacles, the initial topological center is determined as the geometric centroid of the low adhesion boundary.

5. The path optimization method for a cleaning robot according to claim 4, characterized in that, The step of using the chassis radius of the cleaning robot as a curvature constraint parameter to perform adaptive concave envelope fitting on the effective anomaly point set to generate a polygonal contour, and using the polygonal contour as the low adhesion boundary, specifically includes: Based on the spatial coordinates of the effective anomaly point set, a spatial triangulation network containing multiple edge triangles is constructed; The dynamic rolling ball radius is generated by adding the chassis radius of the cleaning robot to the preset safety buffer distance. The circumcircle radius of all edge triangles in the spatial triangulation is calculated by traversing the network, and the circumcircle radius is then threshold-filtered using the dynamic rolling ball radius. Remove the edge triangles whose circumcircle radius is greater than the radius of the dynamic rolling ball, and extract the topological extension edges of the remaining edge triangles; The sequentially connected topological extension edges are subjected to smooth interpolation to generate a polygonal profile in which the global concave curvature radius is greater than the chassis radius.

6. The path optimization method for a cleaning robot according to claim 5, characterized in that, After the step of performing smooth interpolation on the sequentially connected topological extension edges to generate the polygonal contour where the global concave curvature radius is greater than the chassis radius, the method further includes: The concave curvature radius of each curve segment in the polygonal contour is calculated through iteration, and local concave segments with concave curvature radii smaller than the chassis radius are extracted. Extend the two endpoints of the local concave segment outward to construct a transition arc tangent to the radius of the chassis outside the polygonal outline; The transition arc bridges the two side contour lines of the local concave segment, generating a corrected contour segment with an inward curvature radius greater than the chassis radius. The low-adhesion boundary is updated by replacing the local concave segment in the polygonal contour with the corrected contour segment.

7. The path optimization method for a cleaning robot according to claim 4, characterized in that, The step of generating an involute arc segment around the low adhesion boundary and connecting the involute arc segment to the initial traversal path to generate a corrected traversal path specifically includes: An involute base circle is constructed with the geometric centroid as the center of the base circle and the sum of the chassis radius of the cleaning robot and the preset safety buffer distance as the base circle radius. Extract the tangent point or the nearest contact point between the involute base circle and the low adhesion boundary contour line, and take the tangent point or the nearest contact point as the starting point, expand along the tangent direction of the involute base circle at the starting point to generate an involute arc segment surrounding the low adhesion boundary. Identify the path node in the initial traversal path that is closest to the low adhesion boundary, and use the path node as the access reference point; Construct an introduction transition curve from the access reference point to the starting point of the involute arc segment, and an exit transition curve for the subsequent path from the ending point of the involute arc segment to the access reference point; The introduced transition curve, the involute arc segment, and the exit transition curve are connected sequentially, and combined with the initial traversal path, to generate the corrected traversal path.

8. The path optimization method for a cleaning robot according to claim 1, characterized in that, The step of controlling the cleaning robot to travel along the corrected traversal path, keeping the geometric center of the cleaning robot outside the low adhesion boundary, and ensuring that the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary, specifically includes: Obtain the coordinates of the chassis geometric center of the cleaning robot and the coordinates of the edge position of the effective cleaning surface; Calculate the shortest distance from the geometric center of the chassis to the contour line of the low adhesion boundary, as the center avoidance margin; Calculate the penetration depth from the edge position of the effective cleaning surface to the low adhesion boundary contour line, as the cleaning cut-in amount; The center avoidance margin and the cleaning cut-in amount are monitored in real time. When the center avoidance margin is less than a preset safety threshold or the cleaning cut-in amount is less than a preset effective threshold, the driving posture or driving trajectory of the cleaning robot is adjusted. Maintain the center clearance margin greater than zero and the cleaning cut-in amount greater than the preset effective threshold until the cleaning operation of the area surrounding the low adhesion boundary is completed.

9. A path optimization system for a cleaning robot, characterized in that, include: The path generation module is used to obtain the environmental contour of the area to be cleaned and generate an initial traversal path based on the environmental contour. The location determination module is used to control the cleaning robot to travel along the initial traversal path, extract the current output torque and the corresponding actual displacement, and determine abnormal locations based on the current output torque and the actual displacement. The centroid extraction module is used to perform envelope fitting on multiple adjacent anomalous sites to generate low-adhesion boundaries. The path correction module is used to generate an involute arc segment on the periphery of the low adhesion boundary and connect the involute arc segment to the initial traversal path to generate a corrected traversal path. The driving control module is used to control the cleaning robot to travel along the modified traversal path, so that the geometric center of the cleaning robot is kept outside the low adhesion boundary, and the effective cleaning surface of the cleaning robot cuts into the interior of the low adhesion boundary.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the method as described in any one of claims 1 to 8.

Images