A garbage recognition-based self-adaptive cleaning method and system for a sweeping robot

Abstract

This invention belongs to the field of robotic vacuum cleaner technology, specifically relating to an adaptive cleaning method and system for robotic vacuum cleaners based on debris recognition. The invention includes generating a deviation cleaning path from an initial cleaning path. Specifically, this involves acquiring images of debris within the cleaning area, determining the distribution location information of the debris based on the images, generating an initial cleaning path based on the distribution location information, identifying the intersection area between the initial cleaning path and the debris distribution location information, and generating a deviation cleaning path to cover the debris corresponding to the intersection area. By generating a deviation cleaning path, this invention achieves precise cleaning of the debris's distribution location information, thereby improving the cleaning task coverage and ensuring thorough cleaning of the debris area. Simultaneously, the closed-loop optimization process established in this invention enables the robotic vacuum cleaner to possess adaptive learning and continuous optimization capabilities, enhancing its autonomy and robustness in complex environments.

Description

Technical Field

[0001] This invention belongs to the field of robotic vacuum cleaner technology, specifically relating to an adaptive cleaning method and system for robotic vacuum cleaners based on garbage recognition. Background Technology

[0002] As a self-service mobile cleaning device, robotic vacuum cleaners have become an important part of smart homes. They use various sensors to sense and identify the environment and utilize path planning algorithms to achieve autonomous navigation and comprehensive cleaning of indoor spaces, thereby freeing people from daily floor cleaning work and improving their quality of life.

[0003] However, while pursuing efficient path planning, existing technologies often overlook the issue of secondary pollution. Traditional path planning algorithms mostly focus on geometric coverage, but when a robotic vacuum cleaner approaches or passes through areas of accumulated debris along a predetermined path, its high-speed rotating brushes can easily knock up or blow away lighter, fibrous debris such as dust, hair, and paper scraps, causing them to fall into already cleaned areas and resulting in secondary pollution. This not only reduces the cleanliness of a single cleaning cycle but may also force the robot to perform repeated cleaning or require manual cleaning, leading to unnecessary energy consumption and impacting the automatic cleaning experience. Furthermore, the functional logic of existing technologies is relatively rigid, lacking the ability to adapt to cleaning tasks with fine precision. This prevents the robot from adjusting its speed, brush rotation, or suction power based on the type, shape, or density of debris, resulting in weaknesses in situational awareness and intelligent decision-making. Consequently, the actual cleaning performance of robotic vacuum cleaners in complex environments is often unsatisfactory.

[0004] To address the aforementioned issues, this invention provides an adaptive cleaning method and system for a sweeping robot based on waste recognition. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive cleaning method and system for a sweeping robot based on garbage recognition, in order to solve the problem in the prior art where the garbage distribution overlaps with the cleaning path, causing the garbage to be re-excited by the cleaning brush or cleaning vent, thus causing secondary pollution to the area already cleaned by the sweeping robot.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: An adaptive cleaning method for robotic vacuum cleaners based on garbage recognition, including... The process of generating a deviated cleaning path from the initial cleaning path includes: acquiring images of trash within the cleaning area and determining the distribution location information of the trash based on the trash images; generating an initial cleaning path based on the distribution location information; identifying the intersection area between the initial cleaning path and the distribution location information of the trash; and generating a deviated cleaning path to cover the trash corresponding to the intersection area based on the intersection area. After the robot vacuum completes a cleaning task based on the initial cleaning path and any deviations from the initial cleaning path, it executes the following: Evaluate the cleaning performance when deviating from the cleaning path to generate deviation parameters; Based on the deviation parameters, the initial cleaning path is corrected to generate an optimized cleaning path for the next cleaning task.

[0007] Preferably, the step of determining the distribution location information of garbage based on garbage images includes: The garbage image is input into a preset recognition model, which then outputs information about the type and location of the garbage.

[0008] Preferably, the step of outputting the distribution location information of the waste by the recognition model includes: The recognition model outputs at least one labeled point for each garbage image; Based on at least one labeled point, calculate the smallest closed shape that covers all labeled points as the distribution boundary; The closed area enclosed by the distribution boundary is defined as the location information of the waste distribution.

[0009] Preferably, the step of evaluating the cleaning performance deviating from the cleaning path to generate deviation parameters includes: Obtain the time period corresponding to the deviation from the cleaning path; Determine whether the deviation from the cleaning path completely covers the corresponding distribution location information within the deviation time period; If the deviation from the cleaning path does not completely cover the corresponding distribution location information, the deviation time period will be marked as an extended segment. Deviation parameters are generated based on the extended segment.

[0010] Preferably, the step of generating deviation parameters based on the extended segment includes: For each deviation time period marked as an extension segment, calculate the corresponding extension segment offset magnitude; Among them, the offset amplitude of the extended segment is a preset quantitative index used to characterize the degree of deviation from the cleaning path and its corresponding distribution location information; The maximum value among all calculated offset amplitudes of the extended segments is determined as the deviation parameter.

[0011] Preferably, the step of correcting the initial cleaning path based on the deviation parameter to generate an optimized cleaning path for the next cleaning task includes: The initial cleaning path is divided into multiple path segments according to a preset unit time. Identify the path segments in time that correspond to the deviation time period corresponding to the deviation parameter; Based on the deviation parameters, the coordinates of the identified path segments are corrected to generate corrected segments. The corrected segments and the remaining path segments together constitute the optimized cleaning path.

[0012] This invention also discloses an adaptive cleaning system for a robotic vacuum cleaner based on garbage recognition, comprising: The waste identification module is used to acquire images of waste within the cleaning area and determine the distribution location information of the waste; The cleaning path optimization module works in response to the distribution location information determined by the waste identification module. It is used to generate an initial cleaning path and also to receive deviation parameters generated by the cleaning effect evaluation module to correct the initial cleaning path. The cleaning execution module is used to instruct the robot vacuum cleaner to execute the cleaning path generated by the cleaning path optimization module. The cleaning effect evaluation module is used to monitor the execution process of the cleaning execution module, evaluate the cleaning effect to generate deviation parameters, and feed the deviation parameters back to the cleaning path optimization module.

[0013] Preferably, the waste identification module is configured to: acquire waste images in the cleaning area through the image acquisition unit, input the waste images into a preset identification model, and output the type of waste and at least one labeling point for calibrating the waste; based on at least one labeling point, determine the smallest closed shape covering all labeling points through a geometric calculation procedure, and define the area enclosed by the smallest closed shape as the distribution location information of the waste at that location.

[0014] Preferably, the cleaning path optimization module is configured to: generate an initial cleaning path based on the cleaning area information; identify the intersection area between the initial cleaning path and the distribution location information of the garbage, and generate an off-path cleaning path for each intersection area; after receiving the deviation parameter, divide the initial cleaning path into multiple path segments according to a preset unit time, identify the path segments that correspond to the deviation time period corresponding to the deviation parameter in time, and perform coordinate correction on the identified path segments based on the deviation parameter to generate corrected segments, and the corrected segments and the remaining path segments constitute the optimized cleaning path.

[0015] Preferably, the cleaning execution module is configured to: receive the cleaning path generated by the cleaning path optimization module, and parse the cleaning path into low-level control instructions for the robot vacuum's execution components, so as to instruct the robot vacuum to move along the planned path and perform cleaning actions; at the same time, record the robot vacuum's real-time position, posture and timestamp status information during the execution process.

[0016] Beneficial effects 1. This invention obtains images of garbage and coordinates of the garbage distribution area to determine the garbage distribution location information, and generates an initial cleaning path based on the garbage distribution location information. It then identifies the intersection area between the initial cleaning path and the garbage distribution location information, and generates a dedicated deviation cleaning path for the intersection area to control the robot vacuum cleaner to perform cleaning. This achieves precise and targeted cleaning of the garbage area. By generating deviation paths that cover the intersection area, it effectively solves the problem of easily missing the edges of garbage or irregularly distributed garbage in the traditional fixed path cleaning mode, thereby improving the coverage of a single cleaning task and achieving thorough cleaning of garbage.

[0017] 2. This invention utilizes a preset recognition model to process garbage images and output at least one marked point. Based on all marked points, it calculates the smallest closed shape covering them as the distribution boundary, thereby accurately defining the distribution location of garbage. By calculating the degree of overlap between different distribution boundaries, it achieves an objective determination of garbage type. This realizes the combination of machine vision and geometric calculation-based recognition and positioning, providing high-precision input data for subsequent path planning. This ensures that the formulation of cleaning strategies is based on accurate perception and recognition of garbage distribution, thus fundamentally avoiding cleaning omissions caused by recognition errors or positioning deviations.

[0018] 3. This invention evaluates the actual performance of each deviation from the cleaning path, i.e., whether it completely covers the corresponding distribution location information, thereby generating quantified deviation parameters. These parameters are used to correct the coordinates of the initial cleaning path, thereby generating an optimized cleaning path. This optimized path is then used as the initial path for the next cleaning task. This gives the robot vacuum cleaner the ability to learn adaptively and continuously optimize. As a result, it can automatically correct and iterate its cleaning strategy based on feedback from historical cleaning results, gradually eliminating systematic deviations. This continuously improves cleaning efficiency and path rationality during long-term operation, enhancing its robustness and autonomy in complex and changing environments. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method provided by the present invention; Figure 2 This is a system module diagram provided by the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of protection of the invention.

[0021] Example 1 See Figure 1 This embodiment provides an adaptive cleaning method for robotic vacuum cleaners based on garbage recognition. It is applied to robotic vacuum cleaners equipped with image acquisition and processing capabilities, environmental perception capabilities, and path planning and execution control capabilities. The method automatically generates and continuously optimizes the cleaning path based on the actual garbage distribution. The specific steps of this method are as follows: The robot vacuum cleaner activates its onboard image acquisition unit, such as a wide-angle camera, and spatial positioning unit, such as a lidar or inertial measurement unit, to scan the currently set cleaning area to obtain images of the garbage on the ground. Simultaneously, it records and constructs a series of spatial reference coordinates for the cleaning area. These spatial reference coordinates are a set of fixed coordinate points established within the cleaning area, providing a unified reference system for the calculation and calibration of all subsequent position information.

[0022] The acquired waste images are processed to identify the specific type of waste and accurately determine its physical location. This process is completed through a preset image feature comparison and classification rule set, which internally stores visual feature data of various typical waste samples. The visual feature data specifically includes the outline and texture features of paper scraps, the linear features of hair, or the color and distribution density features of granular stains. By calculating the similarity between the real-time acquired image features and the sample features in this database, waste identification and classification are achieved.

[0023] When processing the acquired garbage images, the local image features and image features in the collected garbage images are compared with the sample features in the classification rule set for structured comparison and similarity calculation. When the similarity between the image features of a certain area and the sample features of a specific type of garbage exceeds the preset judgment threshold, it is determined that there is garbage of that type in that area. After determining that the garbage exists, multiple annotation points are automatically generated in the image area to mark the outline or centroid of the garbage, that is, multiple coordinate points used to mathematically mark the outline or centroid of the garbage.

[0024] By using a predetermined geometric calculation procedure, the smallest closed shape that can cover all marked points is determined. The smallest closed shape is preferably the smallest bounding rectangle or convex hull. The geographical area defined by the closed shape is defined as the distribution location information of the garbage at that location, representing the data of the precise geographical range occupied by a single identified garbage in the cleaning area.

[0025] Based on all the known distribution location information, one or more initial cleaning paths are planned and generated. That is, based on all the known waste distribution location information, one or more baseline cleaning trajectories are planned and generated to efficiently connect and cover these locations.

[0026] The logic for generating the initial cleaning path is to ensure that each path can connect and cover the distribution information of one or more garbage locations in the most efficient way. At the same time, each initial cleaning path is spatially independent and does not overlap, so as to avoid ineffective repetitive work. When generating the initial cleaning path, strategies such as prioritizing the cleaning of the farthest area or advancing from the edge of the area to the center are adopted to ensure the overall orderliness of the cleaning task.

[0027] The initial cleaning path is finely adjusted to generate a deviation cleaning path. The generated deviation cleaning path is a fine-tuned execution path generated based on the initial cleaning path and after local trajectory fine-tuning based on the specific garbage distribution location information in the vicinity. Its goal is to ensure that the cleaning components can completely cover the garbage area.

[0028] Specifically, the spatial relationship between the initial cleaning path and its neighboring distribution location information is analyzed. When the trajectory of an initial cleaning path intersects with the distribution location information of a certain piece of garbage or the distance is less than the preset effective cleaning radius, a local deviation from the cleaning path with a precisely adjusted trajectory will be generated based on the path sub-segment. The purpose is to ensure that the cleaning components of the sweeping robot can completely cover the corresponding distribution location information. The preset effective cleaning radius represents the width of the effective range of the sweeping robot's cleaning components in removing garbage.

[0029] The robot vacuum is instructed to strictly follow the generated cleaning path deviation to perform the cleaning task. During or after the cleaning process, the cleaning effectiveness is evaluated in real time. The evaluation process is as follows: Record the time interval of each deviation from the cleaning path that the robot takes, which is the actual time that the robot takes to perform a single deviation from the cleaning path. After the deviation period ends, the image acquisition unit is immediately invoked to re-image the area where the original distribution information corresponding to the deviation from the cleaning path is located. By comparing the image data before and after cleaning, it is determined whether the garbage in the area has been completely removed, and thus whether the deviation from the cleaning path has achieved complete coverage of the predetermined target.

[0030] Deviation parameters for path optimization are generated based on the evaluation results. This data packet is generated based on the evaluation results of the execution effect of a single deviation from the cleaning path. It contains specific instructions for subsequent path optimization, such as correction instructions or confirmation instructions.

[0031] If the evaluation result is determined to be that the deviation from the cleaning path fails to completely cover its corresponding distribution location information, it means that there is still residual garbage after cleaning. In this case, the deviation time period associated with the failed cleaning will be marked as an extended segment, and a deviation parameter containing specific correction instructions will be generated simultaneously. The specific correction instructions may be to expand the coverage of the deviation path by a preset distance unit in a specific direction, or to increase the number of reciprocating cleanings in the area. It should be noted that the extended segment refers to a specific marker placed on the corresponding deviation time period when a deviation from the cleaning task is evaluated as not completely removing garbage, in order to trigger subsequent path correction operations.

[0032] If the evaluation result is complete coverage, the corresponding deviation time period is marked as a normal deviation time period, and a confirmation parameter is generated to confirm that the current path setting is valid and does not need to be corrected.

[0033] All deviation parameters generated during this cleaning task are summarized, and the initial cleaning path is corrected based on these parameters to generate an optimized cleaning path that is more suitable for the current environment.

[0034] The above correction process includes: identifying the path segment of the initial cleaning path that is spatially closest to the deviation cleaning path that generates the correction instruction; and performing a geometric transformation on the path segment according to the correction instructions contained in the deviation parameters, such as coordinate offset or path extension instructions, to generate a correction segment, that is, a new path segment generated by performing a geometric transformation on a specific segment in the initial cleaning path according to the correction instructions in the deviation parameters, which is used to replace the original path segment. Replace the original path segment with the corrected segment, and then concatenate all path segments with all corrected segments in their original order to form the final optimized cleaning path.

[0035] The optimized cleaning path generated during the completion of this cleaning task is stored and set as the default initial cleaning path for the next task in the same area. In this way, the robot vacuum cleaner can automatically adjust its cleaning path based on feedback from each actual cleaning result, continuously adapting to the dynamic changes in the distribution of debris on the ground to improve cleaning efficiency and coverage.

[0036] Example 2 See Figure 2This embodiment discloses an adaptive cleaning system for a robotic vacuum cleaner based on garbage recognition. It can evaluate the cleaning performance and adaptively correct and optimize the cleaning path to improve cleaning efficiency and coverage. The system can be embedded in a robotic vacuum cleaner, which includes, but is not limited to, a processor, memory, LiDAR and infrared sensors for navigation and obstacle avoidance, as well as, but is not limited to, a camera and other visual sensors for image acquisition, and power and cleaning components for performing cleaning.

[0037] The system specifically includes the following modules: The waste identification module is configured to acquire images of waste within the cleaning area and determine the location information of the waste distribution.

[0038] In the specific execution process, before or during the cleaning task, the camera on the robot vacuum cleaner takes a picture of the ground in front of its path as a garbage image; the garbage image is input into a preset recognition model and the position of the garbage in the image is output. The recognition model can be a deep learning network in a convolutional neural network trained with a large amount of image data, which can identify the types of garbage such as paper scraps, hair, and food scraps. Specifically, the recognition model outputs at least one annotation point for each piece of trash identified in the image. Based on the multiple annotation points output by the model, a minimum closed shape that can cover all annotation points is obtained through established geometric calculation procedures such as the convex hull algorithm or the minimum bounding rectangle algorithm. The outline of the minimum closed shape is defined as the distribution boundary. The closed area surrounded by the distribution boundary is determined as the distribution location of the trash and passed to the cleaning path optimization module.

[0039] The core task of the cleaning path optimization module is to generate an initial cleaning path and correct it after receiving deviation parameters in order to generate an optimized cleaning path.

[0040] Furthermore, at the start of a cleaning task, an initial cleaning path covering the entire area is generated based on the map information of the cleaning area, such as a bow-shaped path; the garbage distribution location information provided by the garbage identification module is received, and the intersection areas between the initial cleaning path and the garbage distribution location information are identified, that is, the areas where the initial path will pass through or be adjacent to the garbage. For each intersection area, a dedicated off-track cleaning path is generated to efficiently and completely cover the corresponding garbage distribution location information.

[0041] Preferably, a small-scale reciprocating or spiral cleaning is performed within this area. After a cleaning task including the initial cleaning path and deviations from the cleaning path is completed, deviation parameters generated by the cleaning effect evaluation module are received to correct the execution path. The original initial cleaning path is divided into multiple path segments according to a preset unit time, preferably 1 second; Based on the deviation time period information associated with the deviation parameter, the corresponding path segment is located in time. Based on the specific value of the deviation parameter, the coordinates of the identified path segments are corrected. Specifically, if the deviation parameter indicates a systematic deviation of 2 cm to the right in a certain road segment, the coordinates of the corresponding path segment are shifted 2 cm to the left to generate a corrected segment. The corrected segment, together with other unmodified path segments, forms a new optimized cleaning path for the next cleaning task.

[0042] The cleaning execution module is responsible for translating the path plan generated by the cleaning path optimization module into the actual physical actions of the robot vacuum. It also receives the complete cleaning path generated by the optimization module and parses it into low-level control commands for the robot's hub motors, side brushes, roller brushes, and fans. The complete cleaning path includes the initial path and all deviations from it, ensuring the robot moves strictly according to the planned path to perform cleaning actions within the designated area. During the cleaning process, the module continuously records the robot's real-time position, posture, and timestamp status information for use by the cleaning performance evaluation module.

[0043] The cleaning effect evaluation module is used to monitor the execution process of the cleaning execution module after the cleaning task is completed, evaluate the cleaning execution effect, and generate deviation parameters.

[0044] Specifically, this module focuses on the actual trajectory of the sweeping robot when it deviates from the cleaning path, obtains the deviation time period corresponding to each deviation from the cleaning path, and compares the actual movement trajectory of the robot recorded by the cleaning execution module with the garbage distribution location information that it should cover within the deviation time period. By comparison, it determines whether the deviation from the cleaning path completely covers the corresponding distribution location information within the deviation time period. If the actual trajectory fails to completely cover the target area due to systematic deviations such as wheel slippage and gyroscope drift, this deviation time period is marked as an extended segment. For each deviation time period marked as an extended segment, the corresponding extended segment offset amplitude is calculated. The extended segment offset amplitude is a preset quantitative index used to characterize the degree of deviation from the cleaning path and its corresponding distribution location information. It can be further defined as the maximum vertical distance between the actual trajectory boundary and the target distribution boundary. After calculating the offset of all extension segments, the maximum value among all calculated offsets is determined as the final deviation parameter for this cleaning task. This module also feeds it back to the cleaning path optimization module to guide the next path correction.

[0045] This embodiment, based on actively identifying and specifically cleaning localized garbage areas, achieves the goal of gradually eliminating systematic path deviations caused by robot hardware wear, changes in ground material, or sensor errors by quantitatively evaluating the cleaning performance and continuously learning and correcting the initial cleaning path, thereby improving the accuracy and reliability of long-term cleaning.

Claims

1. An adaptive cleaning method for a robotic vacuum cleaner based on garbage recognition, characterized in that, The process of generating a deviated cleaning path from the initial cleaning path includes: acquiring images of trash within the cleaning area and determining the distribution location information of the trash based on the trash images; generating an initial cleaning path based on the distribution location information; identifying the intersection area between the initial cleaning path and the distribution location information of the trash; and generating a deviated cleaning path to cover the trash corresponding to the intersection area based on the intersection area. After the robot vacuum completes a cleaning task based on the initial cleaning path and the deviations generated from the initial cleaning path, it performs the following: evaluates the cleaning performance of the deviations to generate deviation parameters, and corrects the initial cleaning path based on the deviation parameters to generate an optimized cleaning path for the next cleaning task.

2. The adaptive cleaning method for a sweeping robot based on garbage recognition according to claim 1, characterized in that, The steps for determining the distribution location information of garbage based on garbage images include: The garbage image is input into a preset recognition model, which then outputs information about the type and location of the garbage.

3. The adaptive cleaning method for a sweeping robot based on garbage recognition according to claim 2, characterized in that, The steps for the identification model to output information on the distribution location of waste include: The recognition model outputs at least one labeled point for each garbage image; Based on at least one labeled point, calculate the smallest closed shape that covers all labeled points as the distribution boundary; The closed area enclosed by the distribution boundary is defined as the location information of the waste distribution.

4. The adaptive cleaning method for a sweeping robot based on garbage recognition according to claim 1, characterized in that, The steps for evaluating the cleaning performance deviating from the cleaning path to generate deviation parameters include: Obtain the time period corresponding to the deviation from the cleaning path; Determine whether the deviation from the cleaning path completely covers the corresponding distribution location information within the deviation time period; If the deviation from the cleaning path does not completely cover the corresponding distribution location information, the deviation time period will be marked as an extended segment. Deviation parameters are generated based on the extended segment.

5. The adaptive cleaning method for a sweeping robot based on garbage recognition according to claim 4, characterized in that, The steps for generating deviation parameters based on the extended segment include: For each deviation time period marked as an extension segment, calculate the corresponding extension segment offset magnitude; Among them, the offset amplitude of the extended segment is a preset quantitative index used to characterize the degree of deviation from the cleaning path and its corresponding distribution location information; The maximum value among all calculated offset amplitudes of the extended segments is determined as the deviation parameter.

6. The adaptive cleaning method for a sweeping robot based on garbage recognition according to claim 1, characterized in that, The steps for refining the initial cleaning path based on deviation parameters to generate an optimized cleaning path for the next cleaning task include: The initial cleaning path is divided into multiple path segments according to a preset unit time. Identify the path segments in time that correspond to the deviation time period corresponding to the deviation parameter; Based on the deviation parameters, the coordinates of the identified path segments are corrected to generate corrected segments. The corrected segments and the remaining path segments together constitute the optimized cleaning path.

7. An adaptive cleaning system for a robotic vacuum cleaner based on garbage recognition, characterized in that, include: The waste identification module is used to acquire images of waste within the cleaning area and determine the distribution location information of the waste; The cleaning path optimization module works in response to the distribution location information determined by the waste identification module. It is used to generate an initial cleaning path and also to receive deviation parameters generated by the cleaning effect evaluation module to correct the initial cleaning path. The cleaning execution module is used to instruct the robot vacuum cleaner to execute the cleaning path generated by the cleaning path optimization module. The cleaning effect evaluation module is used to monitor the execution process of the cleaning execution module, evaluate the cleaning effect to generate deviation parameters, and feed the deviation parameters back to the cleaning path optimization module.

8. The adaptive cleaning system for a sweeping robot based on garbage recognition according to claim 7, characterized in that, The waste identification module is configured to: acquire images of waste in the cleaning area through the image acquisition unit, input the waste images into the preset identification model, and output the type of waste and at least one labeling point for identifying the waste; Based on at least one marked point, a minimum closed shape covering all marked points is determined through geometric calculation procedures, and the area enclosed by this minimum closed shape is defined as the distribution location information of the waste at that location.

9. The adaptive cleaning system for a sweeping robot based on garbage recognition according to claim 7, characterized in that, The cleaning path optimization module is configured to: generate an initial cleaning path based on the cleaning area information; identify the intersection area between the initial cleaning path and the distribution location information of the garbage, and generate an off-path cleaning path for each intersection area; after receiving the deviation parameter, divide the initial cleaning path into multiple path segments according to a preset unit time, identify the path segments that correspond to the deviation time period corresponding to the deviation parameter in time, and perform coordinate correction on the identified path segments based on the deviation parameter to generate corrected segments, and the corrected segments and the remaining path segments constitute the optimized cleaning path.

10. The adaptive cleaning system for a sweeping robot based on garbage recognition according to claim 7, characterized in that, The cleaning execution module is configured to: receive the cleaning path generated by the cleaning path optimization module, and parse the cleaning path into low-level control instructions for the robot vacuum's execution components, so as to instruct the robot vacuum to move along the planned path and perform cleaning actions; at the same time, record the robot vacuum's real-time position, posture and timestamp status information during the execution process.

Images