Automatic cleaning method based on environmental perception
By integrating the main control module and information collection module on the unmanned sweeper, identifying and counting garbage conditions and independently planning the cleaning mode and path, the problem that traditional unmanned sweepers cannot adjust the cleaning strategy in real time is solved, and efficient and high-quality are achieved Clean cleaning effect.
Patent Information
- Application Number
- CN202411058477.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-02
- Publication Date
- 2024-11-22
AI Technical Summary
Traditional unmanned sweepers failed to detect spam information in advance before starting their operations, and could not plan the optimal sweeping mode according to the real-time garbage volume, resulting in unchanged cleaning paths, making it difficult to ensure efficient sweeping, and the garbage location status cannot be updated in real time, affecting cleaning and cleaning. Rate.
Through the main control module and the information collection module, identify and count the ground garbage situation, independently plan the optimal cleaning mode, obtain the garbage status in real time, adjust the cleaning plan, mark the missing garbage and its location information, count the number of missing garbage, and re-sweep the missing garbage. Ensure the cleanliness of the operating area.
Improve the efficiency of cleaning operations, ensure the cleaning rate of patrol and sweeping, reduce the loss of sweeping, and ensure the cleanliness of the working area.
Smart Images

Figure CN119002479A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of automated cleaning, and in particular relates to an automated cleaning method based on environmental perception. Background Art
[0002] With the development of artificial intelligence algorithms and unmanned driving technology, the widespread application of intelligent unmanned sweepers has been promoted, which has liberated a large amount of human labor and solved the increasingly huge cleaning workload of urban public areas. Among the current related technologies, there are still some shortcomings in the automatic cleaning operations of traditional unmanned sweepers. Some unmanned sweepers do not detect garbage information and garbage volume in advance before starting the operation task, and cannot plan the optimal cleaning mode according to the real-time garbage volume. They perform indiscriminate coverage cleaning for operation areas with different garbage volume distribution at different times in different scenes, and the cleaning path does not change, which makes it difficult to ensure efficient cleaning; although some unmanned sweepers obtain target garbage feature information through visual sensors and visual algorithm recognition functions before starting the operation task, they can only perform fixed-point cleaning work for the location of the garbage, cannot plan the operation mode according to the garbage volume, cannot track the garbage status change information, and cannot update the garbage location status in real time for the task scene where the garbage status is constantly changing, adjust the global cleaning plan and local cleaning plan in real time, and record the remaining status of garbage cleaning, resulting in the inability to go back and forth to make up for the missed cleaning, affecting the cleaning cleanliness rate.
[0003] In response to the above-mentioned phenomena and problems, we conducted in-depth research and proposed an automated cleaning method based on environmental perception. The main control module cooperates with the information collection module to identify and count the ground garbage situation, and independently plan the optimal cleaning mode based on this to improve the cleaning efficiency; obtain the garbage status in real time, adjust the cleaning plan, and ensure the cleaning rate; mark the missed garbage and its location information, count the amount of missed garbage, and make up for the missed garbage to ensure the cleanliness rate of the operating area. Summary of the invention
[0004] In response to the above-mentioned existing problems, the present invention proposes an automated cleaning method based on environmental perception, which identifies and counts the ground garbage situation through the cooperation of the main control module and the information collection module, and independently plans the optimal cleaning mode based on this to improve the cleaning efficiency; obtains the garbage status in real time, adjusts the cleaning plan, and ensures the cleaning rate of the patrol; marks the missed garbage and its location information, counts the amount of missed garbage, and performs supplementary cleaning for the missed garbage to ensure the cleanliness rate of the operating area.
[0005] To achieve the above-mentioned purpose, the present invention provides an automated cleaning method based on environmental perception, including an unmanned cleaning vehicle, wherein the unmanned cleaning vehicle is provided with a main control module, an information collection module, a cleaning drive module and a sweeping brush cleaning module, wherein the main control module is used for signal reception integration and regulation of the unmanned cleaning vehicle, the information collection module includes a visual module and a communication module, and the main control module has a built-in garbage detection model. The automated cleaning method based on environmental perception specifically includes the following steps: S1, detection and determination of the working area environment: S11, the unmanned cleaning vehicle autonomously navigates into the working area, the sweeping brush cleaning module remains folded and closed, the visual module collects the global environmental image in the working area, and the communication module collects the information collected by the visual module. The collected global environment image is transmitted to the main control module; S12, the main control module compares the global environment image with the garbage detection model to obtain data information such as the location, type and quantity of garbage targets in the operating area; S13, the main control module calculates and analyzes the global garbage density in the operating area, and determines the garbage distribution type in the operating area. If the operating area is determined to be a global dense area, the unmanned sweeper performs the covering cleaning operation of step S2; if the operating area is determined to be a scattered distribution area, the unmanned sweeper performs the patrol cleaning operation of step S3; S2, covering cleaning operation: S21, the main control module plans a covering cleaning path for the global dense area to ensure that the unmanned sweeper is in the When the unmanned sweeper moves on the covered cleaning path, the cleaning range of the sweeping brush cleaning module completely covers the working area; S22, the unmanned sweeper moves along the covered cleaning path, and the cleaning drive module drives the sweeping brush cleaning module to be continuously lowered and started throughout the whole process to perform the cleaning operation; S23, when the unmanned sweeper reaches the end of the covered cleaning path, the cleaning drive module drives the sweeping brush cleaning module to be retracted and closed upwards; S3, patrol-type cleaning operation: S31, the main control module plans the inspection path for the scattered distribution areas to ensure that when the unmanned sweeper moves on the inspection path, the camera detection range of the visual module completely covers the working area; S32, the unmanned sweeper moves along the inspection path During patrol and cleaning, the visual module collects the path environment image in real time, the communication module transmits the path environment image to the main control module in real time, the main control module compares the path environment image with the garbage detection model, and obtains the garbage targets around the unmanned sweeper when it is traveling on the patrol path in real time; S33, the unmanned sweeper keeps the sweeping brush cleaning module in a folded and closed state during patrol and cleaning. When the main control module detects the existence of garbage targets around, the unmanned sweeper travels in the direction of the garbage target, and when it reaches the garbage target, the cleaning drive module drives the sweeping brush cleaning module to be lowered to start cleaning, and resets and closes after cleaning is completed. The unmanned sweeper returns to the patrol path and continues to travel until a single patrol and cleaning is completed.
[0006] Furthermore, the main control module sets the garbage target that has not been cleaned in a single time as a missed sweep mark, and the missed sweep mark includes the garbage quantity and location information of the garbage target, and sets a missed sweep mark threshold; after the unmanned sweeping vehicle completes step S33, a make-up sweep operation is set: S34, the main control module performs statistical comparison on all recorded missed sweep marks, if the number of missed sweep marks is greater than the missed sweep mark threshold, the unmanned sweeping vehicle turns around and returns to perform the make-up sweep operation of the missed sweep mark.
[0007] Furthermore, the main control module sets a threshold value for the number of supplementary sweeping operations for the supplementary sweeping operations, and the supplementary sweeping operations also include the following steps: S341. During the execution of the supplementary sweeping operation, the main control module detects the changes in the garbage targets in real time and updates the missed sweeping mark records; S342. After a single supplementary sweeping operation is completed, the main control module again performs a statistical comparison on all the missed sweeping marks recorded. If the number of missed sweeping marks is greater than the missed sweeping mark threshold, the unmanned sweeping vehicle performs the supplementary sweeping operation again; S343. The main control module performs a statistical record of the number of supplementary sweeping operations. If the number of supplementary sweeping operations of the unmanned sweeping vehicle exceeds the threshold value for the number of supplementary sweeping operations, and the main control module still determines that supplementary sweeping needs to be performed, the patrol sweeping task is forcibly terminated.
[0008] Furthermore, the main control module sets a garbage clustering level for the garbage target, the supplementary sweeping steps of the supplementary sweeping operation are the same as the patrol cleaning steps, and the end point of a single supplementary sweeping operation is the first missed sweep mark of its single supplementary sweeping path; the specific steps of the patrol cleaning in the step S33 are as follows: S331, the main control module sets a lateral offset range for the patrol inspection path, and the unmanned sweeper can move within the lateral offset range when passing through the patrol inspection path. If a single garbage target is detected within the lateral offset range, the unmanned sweeper approaches the single garbage target and cleans it; S332, if the lateral If two garbage targets with similar positions are detected within the offset range, and they are side by side with the unmanned sweeper, so that they cannot be cleaned in a single pass, they are cleaned from high to low according to the garbage clustering level; S333, if three or more garbage targets with similar positions are detected within the lateral offset range, and they are side by side with the unmanned sweeper, only the garbage targets with the highest garbage clustering level are cleaned, and the positions of the uncleaned garbage targets are recorded as missed sweep marks; S334, the unmanned sweeper completes the cleaning of the garbage targets within the lateral offset range, returns to the inspection path and continues to move forward until a single inspection and cleaning is completed.
[0009] Furthermore, in step S32, the main control module also sets up a garbage target re-inspection mechanism. If the main control module finds that a certain garbage target has changed its state during the garbage target re-inspection, the inspection path is re-planned according to steps S331 to S333. Specifically, the garbage target re-inspection mechanism is as follows: the visual module collects global environmental images in the operating area at a certain frequency, the main control module records the quantity and location information of each garbage target detected each time, and compares the state changes of the quantity and location information of each garbage target recorded last time with the previous time.
[0010] Furthermore, the garbage detection model is a garbage model algorithm with corresponding garbage model feature values obtained by collecting a large amount of garbage sample image data through a deep neural network architecture for garbage detection model training, and the visual module includes a calibrated visual camera and a lidar sensor; the specific method in the S13 step is: inputting the global environment image into the garbage detection model for feature extraction and matching, obtaining the pixel coordinates of the corresponding garbage in the global environment image, and converting the image pixel coordinates into laser map coordinates through the coordinate association relationship between the pre-calibrated visual camera and the lidar sensor to obtain data information such as the location, type and quantity of garbage in the working area.
[0011] Furthermore, a garbage quantity calibration value and a regional comparison critical value are set in the main control module, and the steps for judging the garbage distribution type in the working area in the step S13 are as follows: S131, the main control module divides the global environment image collected by the visual module into a number of equally divided sub-areas of equal area; S132, the main control module counts the amount of garbage in each sub-area, if the amount of garbage in the sub-area does not exceed the calibration value, then the sub-area is regarded as a non-garbage dense sub-area, otherwise it is a garbage dense sub-area; S133, the main control module counts the non-garbage dense sub-areas and the garbage dense sub-areas respectively, if the cumulative total number of garbage dense sub-areas is greater than the total number of non-garbage dense sub-areas and reaches the regional comparison critical value, it belongs to the global dense area, otherwise it belongs to the scattered distribution area.
[0012] Furthermore, when performing cleaning operations within the lateral offset range, the area between the current garbage target position and the nearest garbage target of the unmanned sweeper is the local walking area. If a new garbage target is detected in the local walking area, if the new garbage target is in the local walking area, the unmanned sweeper returns to the inspection path to continue the cleaning operation; if the new garbage target is not in the local walking area, the unmanned sweeper returns to the inspection path for cleaning operations and records the new garbage target as a missed cleaning mark.
[0013] Furthermore, the garbage clustering level includes a large amount of clustered garbage, regular clustered garbage, a small amount of clustered garbage and scattered garbage, and the judgment basis is: setting the garbage accumulation radius, the minimum garbage value, the intermediate garbage value and the maximum garbage value, performing statistical analysis on the garbage accumulation situation, taking a single garbage as the central reference point, and counting the amount of garbage within the garbage accumulation radius. If the counted garbage amount is equal to the minimum garbage value, it is determined to be a scattered garbage point; if the counted garbage amount is greater than the minimum garbage value but does not exceed the intermediate garbage value, it is determined to be a small amount of clustered garbage; if the counted garbage amount is greater than the intermediate garbage value but does not exceed the maximum garbage value, it is determined to be regular clustered garbage; if the counted garbage amount exceeds the maximum garbage value, it is determined to be a large amount of clustered garbage.
[0014] Furthermore, the state change includes the disappearance of garbage targets, the addition of garbage targets and the change of the position of garbage targets, and the judgment method is: when the unmanned sweeper is not cleaning, for the same garbage target, the main control module compares its record of the subsequent detection with the record of the previous detection. If it exists in the previous detection record but does not exist in the subsequent detection record, it is judged that the garbage target disappears, and the main control module clears the garbage target record; if it does not exist in the previous detection record but exists in the subsequent detection record, it is judged that the garbage target is added, and the main control module updates the record in real time. A new garbage target is added; if it exists in both previous and subsequent detection records, the initial position of the garbage target is used as the central reference point, a displacement determination radius is set, and the position information of the garbage target in the two records is judged and compared. If the position information of the garbage target in the later detection record is still within the determination radius relative to the previous detection record, it is determined that there is no change in the position of the garbage target, and the main control module keeps the garbage target record information unchanged; if its position information is not within the determination radius, it is determined that there is a change in the position of the garbage target, and the main control module clears the initial record of the garbage target, and updates the record of the newly added garbage target.
[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. An automated cleaning method based on environmental perception provided by the present invention can detect the global garbage location in the working area and count the amount of garbage by building a garbage detection model in the main control module and cooperating with the information collection module. The garbage distribution type of the working area is determined and confirmed according to the distribution of garbage targets in the working area. The optimal cleaning operation mode can be matched autonomously for different types of working areas. If the working area is determined to be a global dense area, a covering cleaning operation is performed. If the working area is determined to be a scattered distribution area, a patrol cleaning operation is performed, thereby improving the cleaning efficiency of the unmanned cleaning vehicle. 2. The patrol-type cleaning operation set up by the automated cleaning method based on environmental perception provided by the present invention can cope with the state changes of garbage targets caused by the outdoor environment, update garbage information in real time according to the state changes of garbage targets, and adjust the global cleaning plan and the local cleaning plan; perform garbage clustering level analysis on garbage targets to avoid or reduce the amount of garbage missed during a single patrol cleaning process; set the amount of garbage and location information of garbage targets that have not been cleaned in a single time as a missed cleaning mark, and determine whether to perform a supplementary cleaning operation by counting the missed cleaning marks and comparing them with the missed cleaning mark threshold, and adjust the number of round-trip supplementary cleaning operations, ultimately achieving the effect of improving the cleaning efficiency of the patrol-type cleaning operation and ensuring the cleanliness of the operation area. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A flow chart of an automatic cleaning method based on environmental perception according to the present invention; Figure 2 Flow chart of the garbage target review mechanism of the present invention. DETAILED DESCRIPTION
[0017] To further describe the present invention in detail, the following is an explanation with reference to the accompanying drawings. It is particularly noted that the embodiments described below are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0018] See also Figure 1 and Figure 2 The present embodiment provides an automated cleaning method based on environmental perception, including an unmanned cleaning vehicle, wherein the unmanned cleaning vehicle is provided with a main control module, an information collection module, a cleaning drive module and a sweeping brush cleaning module, wherein the information collection module includes a visual module and a communication module, and the main control module is used for signal reception integration and regulation of the unmanned cleaning vehicle.
[0019] The main control module has a built-in garbage detection model and a garbage target re-inspection mechanism; at the same time, the global environment image is divided into several equally divided sub-areas; a garbage amount calibration value, a regional comparison critical value and a garbage clustering level are set for the garbage target; a lateral offset range is set for the inspection path to prevent it from always detecting garbage in a certain direction and always offsetting in this direction, resulting in leaving the original inspection route, which in turn causes the visual module detection range to be unable to effectively cover the operating area; a threshold for the number of supplementary sweeping operations is set for the supplementary sweeping operation; a missed sweep mark is set for the garbage target that has not been cleaned once, and the missed sweep mark includes the amount and location information of the garbage target, and a missed sweep mark threshold is set; in this embodiment, the sub-area area is set to 1 square meter, the garbage amount calibration value is set to 5, the regional comparison critical value is set to 60%, and the garbage clustering level includes a large amount of clustered garbage, conventional clustered garbage, a small amount of clustered garbage and scattered garbage; the lateral offset range is within 3 meters to the left and right of the inspection path; the threshold for the number of supplementary sweeping operations is 3 times; and the missed sweep mark threshold is 3.
[0020] In this embodiment, the garbage detection model is a garbage model algorithm with corresponding garbage model feature values obtained by collecting a large amount of garbage sample image data through a deep neural network architecture for garbage detection model training. The visual module includes a calibrated visual camera and a lidar sensor. The specific method in the S13 step is: inputting the global environment image into the garbage detection model for feature extraction and matching, obtaining the pixel coordinates corresponding to the garbage in the global environment image, and converting the image pixel coordinates into laser map coordinates through the coordinate association relationship between the pre-calibrated visual camera and the lidar sensor to obtain data information such as the location, type and quantity of garbage in the working area.
[0021] In this embodiment, the garbage target re-inspection mechanism is specifically as follows: the visual module collects the global environment image in the operation area at a certain frequency, the main control module records the quantity and location information of the garbage targets detected each time, and compares the state changes of the quantity and location information of each garbage target recorded last time with the previous time; The state change includes the disappearance of garbage targets, the addition of garbage targets and the change of the position of garbage targets, and the judgment method is: when the unmanned sweeper does not perform cleaning, for the same garbage target, the main control module compares its record in the subsequent detection with the record in the previous detection. If it exists in the previous detection record but does not exist in the subsequent detection record, it is judged that the garbage target disappears, and the main control module clears the garbage target record; if it does not exist in the previous detection record but exists in the subsequent detection record, it is judged that the garbage target is newly added, and the main control module updates the record of the newly added garbage target in real time; if it exists in both the previous and subsequent detection records, the initial position of the garbage target is used as the central reference point, the displacement judgment radius is set, and the position information of the garbage target in the two records is judged and compared. If the position information of the garbage target in the subsequent detection record is still within the judgment radius range relative to the previous detection record, it is judged that there is no change in the position of the garbage target, and the main control module keeps the garbage target record information unchanged; if its position information is not within the judgment radius range, it is judged that there is a change in the position of the garbage target, and the main control module clears the initial record of the garbage target and updates the record of the newly added garbage target.
[0022] The garbage clustering level is determined based on: setting the garbage accumulation radius, the minimum garbage value, the intermediate garbage value and the maximum garbage value. In this embodiment, the garbage accumulation radius is set to 0.5 meters, the minimum garbage value is set to 1, the intermediate garbage value is set to 3, and the maximum garbage value is set to 6; a statistical analysis is performed on the garbage accumulation situation, and the garbage target is used as the central reference point. The amount of garbage is counted within a circular range with a radius of 0.5 meters around the garbage target. If the number of garbage counted is 1, it is determined to be a scattered garbage point; if the number of garbage counted is greater than 1 but not more than 3, it is determined to be a small amount of clustered garbage; if the number of garbage counted is greater than 3 but not more than 6, it is determined to be regular clustered garbage; if the number of garbage counted exceeds 6, it is determined to be a large amount of clustered garbage.
[0023] The automatic cleaning method based on environmental perception specifically includes the following steps: S1. Working area environment detection and determination: S11, the unmanned sweeper autonomously navigates into the working area, the sweeping brush cleaning module remains closed, the visual module collects the global environment image in the working area, and the communication module transmits the global environment image collected by the visual module to the main control module; S12, the main control module compares the global environment image with the garbage detection model to obtain data information such as the location, type and quantity of garbage targets in the operating area; S13, the main control module calculates and analyzes the global garbage density in the operating area, and determines the garbage distribution type in the operating area. The specific steps are as follows: S131, the main control module divides the global environment image collected by the visual module into sub-areas with an area of 1 square meter; S132, the main control module counts the amount of garbage in each sub-area. If the amount of garbage in the sub-area does not exceed 5, the sub-area is considered a non-garbage-intensive sub-area, otherwise it is a garbage-intensive sub-area; S133, the main control module counts the number of non-garbage dense sub-areas and garbage dense sub-areas respectively. If the cumulative total number of garbage dense sub-areas is greater than the total number of non-garbage dense sub-areas and reaches 60% of the total number, it belongs to the global dense area, otherwise it belongs to the scattered distribution area; If the operation area is determined to be a global dense area, the unmanned sweeper performs a covering sweeping operation in step S2; if the operation area is determined to be a scattered distribution area, the unmanned sweeper performs a patrol sweeping operation in step S3; S2. Covering cleaning operation: S21, the main control module plans a covering cleaning path for the global dense area to ensure that when the unmanned cleaning vehicle moves on the covering cleaning path, the cleaning range of the sweeping and brushing cleaning module completely covers the working area; S22, the unmanned cleaning vehicle moves along the covering cleaning path, and the cleaning driving module drives the sweeping and brushing cleaning module to be continuously lowered and started throughout the whole process to perform cleaning operations; S23, when the unmanned cleaning vehicle reaches the end of the covering cleaning path, the cleaning driving module drives the sweeping brush cleaning module to be retracted upward and closed; S3, Patrol cleaning operation: S31, the main control module plans an inspection path for the scattered distribution area, and the interval distance of the inspection path is sparser than that of the covering cleaning path, so as to ensure that when the unmanned cleaning vehicle moves on the inspection path, the camera detection range of the visual module completely covers the working area; S32, the unmanned sweeper performs patrol and cleaning along the inspection path. In this embodiment, the area between the current garbage target position and the nearest garbage target of the unmanned sweeper is a local walking area, the visual module collects the path environment image in real time, the communication module transmits the path environment image to the main control module in real time, the main control module compares the path environment image with the garbage detection model, and obtains the garbage targets around the unmanned sweeper when it is traveling along the inspection path in real time, and re-inspects the garbage targets at the same time. If a garbage target is found to have a state change during the garbage target re-inspection, the inspection path is re-planned according to steps S331 to S333; S33, the unmanned sweeper keeps the sweeping and brushing cleaning module in a closed state during patrol and cleaning. When the main control module detects that there are garbage targets within 3 meters on the left and right of the patrol path: S331, if a single garbage target is detected within the lateral offset range, the unmanned sweeper moves toward the garbage target, and when it reaches the garbage target, the cleaning drive module of the unmanned sweeper drives the sweeping brush cleaning module to lower and start cleaning, and resets and closes after cleaning is completed; S332: If two garbage targets with similar positions are detected within the lateral offset range, and they are in a side-by-side state with the unmanned sweeper, so that a single pass cannot guarantee the cleaning of all of them, then the garbage clustering levels are cleaned from high to low. In this embodiment, the side-by-side state means that the garbage targets are located on both sides of the traveling direction of the unmanned sweeper; S333: If three or more garbage targets with similar positions are detected within the lateral offset range, and they are in parallel with the unmanned sweeper, only the garbage target with the highest garbage clustering level is cleaned, and the position of the uncleaned garbage target is recorded as a missed sweep mark; S334, the unmanned sweeper completes the cleaning of the garbage target within the lateral offset range, returns to the inspection path and continues to move forward until a single inspection and cleaning is completed; in this embodiment, if a new garbage target is detected in the local walking area, if the new garbage target is in the local walking area, the unmanned sweeper returns to the inspection path and continues the cleaning operation; if the new garbage target is not in the local walking area, the unmanned sweeper returns to the inspection path and continues the cleaning operation and records the new garbage target as a missed sweep mark until a single inspection and cleaning is completed; S34, the main control module performs statistical comparison on all missed sweep marks recorded. If the number of missed sweep marks is greater than 3, the unmanned sweeper turns around and returns to perform the make-up sweep operation of the missed sweep marks. In this embodiment, the make-up sweeping steps of the make-up sweeping operation are the same as the patrol sweeping steps, and the end point of a single make-up sweeping operation is the first missed sweep mark of its single make-up sweeping path; S341, during the process of performing the supplementary sweeping operation, the main control module detects the changes of the garbage targets in real time and updates the missed sweeping mark record; S342, after a single supplementary sweeping operation is completed, the main control module again performs statistical comparison on all the missed sweeping marks recorded. If the number of missed sweeping marks is greater than 3, the unmanned sweeper performs the supplementary sweeping operation again; S343, the main control module makes a statistical record of the number of supplementary sweeping operations. If the number of supplementary sweeping operations of the unmanned sweeper exceeds 3 times, and the main control module still determines that supplementary sweeping is required, the patrol sweeping task is forcibly terminated.
[0024] In this embodiment, the present invention provides an automated cleaning method based on environmental perception. By building a garbage detection model in the main control module and cooperating with the information collection module, the global garbage location in the working area can be detected and the amount of garbage can be counted. The garbage distribution type of the working area can be determined and confirmed according to the distribution of garbage targets in the working area. The optimal cleaning operation mode can be matched autonomously for different types of working areas. If the working area is judged to be a global dense area, a covering cleaning operation is performed. If the working area is judged to be a scattered distribution area, a patrol-type cleaning operation is performed, thereby improving the cleaning efficiency of the unmanned cleaning vehicle. Among them, the patrol-type cleaning operation The industry can respond to the state changes of garbage targets caused by the outdoor environment, update garbage information in real time according to the state changes of garbage targets, and adjust the global cleaning plan and local cleaning plan; conduct garbage clustering level analysis on garbage targets to avoid or reduce the amount of garbage missed during a single patrol cleaning process; set the garbage quantity and location information of garbage targets that have not been cleaned in a single time as missed sweep marks, and determine whether to perform a make-up sweep operation by counting the missed sweep marks and comparing them with the missed sweep mark threshold, and adjust the number of round-trip make-up sweep operations, ultimately achieving the effect of improving the cleaning efficiency of patrol cleaning operations and ensuring the cleanliness of the operating area.
[0025] In this embodiment, the above is only an example and is not intended to limit the scope of protection of this application.
[0026] The present invention provides an automatic cleaning method based on environment perception.
[0027] The embodiments disclosed above are only for the purpose of describing the present invention in detail and cannot be used to limit the scope of the present invention. Therefore, simple modifications and changes made according to the scope of application of the claims of the present invention still fall within the scope of protection of the present invention.
[0028] The protection scope of the present invention shall be based on the defined scope. For those skilled in the art, several improvements and modifications may be made without departing from the spirit and scope of the present invention, and these and modifications shall also be regarded as the protection scope of the present invention.
Claims
1. An automated cleaning method based on environmental perception, comprising an unmanned cleaning vehicle, wherein the unmanned cleaning vehicle is provided with a main control module, an information collection module, a cleaning drive module and a sweeping brush cleaning module, wherein the main control module is used for signal reception and integration and regulation of the unmanned cleaning vehicle, and the information collection module comprises a visual module and a communication module. It is characterized in that The main control module has a built-in garbage detection model, and the automatic cleaning method based on environmental perception specifically includes the following steps: S1. Working area environment detection and determination: S11, the unmanned sweeper autonomously navigates into the working area, the sweeping brush cleaning module remains closed, the visual module collects the global environment image in the working area, and the communication module transmits the global environment image collected by the visual module to the main control module; S12, the main control module compares the global environment image with the garbage detection model to obtain data information such as the location, type and quantity of garbage targets in the operating area; S13, the main control module calculates and analyzes the global garbage density in the operation area, and determines the garbage distribution type in the operation area. If the operation area is determined to be a global dense area, the unmanned sweeper performs the covering cleaning operation of step S2; if the operation area is determined to be a scattered distribution area, the unmanned sweeper performs the patrol cleaning operation of step S3; S2. Covering cleaning operation: S21, the main control module plans a covering cleaning path for the global dense area to ensure that when the unmanned cleaning vehicle moves on the covering cleaning path, the cleaning range of the sweeping and brushing cleaning module completely covers the working area; S22, the unmanned cleaning vehicle moves along the covering cleaning path, and the cleaning driving module drives the sweeping and brushing cleaning module to be continuously lowered and started throughout the whole process to perform cleaning operations; S23, when the unmanned cleaning vehicle reaches the end of the covering cleaning path, the cleaning driving module drives the sweeping brush cleaning module to be retracted upward and closed; S3, Patrol cleaning operation: S31, the main control module plans the inspection path for the scattered distribution area to ensure that when the unmanned sweeper moves on the inspection path, the camera detection range of the visual module completely covers the working area; S32, the unmanned sweeper performs patrol and cleaning along the patrol route, the visual module collects the path environment image in real time, the communication module transmits the path environment image to the main control module in real time, the main control module compares the path environment image with the garbage detection model, and obtains the garbage targets around the unmanned sweeper when it is traveling along the patrol route in real time; S33, the unmanned sweeper keeps the sweeping brush cleaning module in the folded and closed state during patrol cleaning. When the main control module detects the presence of garbage targets in the surroundings, the unmanned sweeper moves towards the garbage targets. When reaching the garbage targets, the cleaning drive module drives the sweeping brush cleaning module to be lowered to start cleaning. After cleaning is completed, it is reset and closed. The unmanned sweeper returns to the patrol path and continues to move until a single patrol cleaning is completed.
2. The automatic cleaning method based on environmental perception according to claim 1, It is characterized in that The main control module sets the garbage target that has not been cleaned in a single time as a missed cleaning mark, wherein the missed cleaning mark includes the amount and location information of the garbage target, and sets a missed cleaning mark threshold; after the unmanned cleaning vehicle completes step S33, a supplementary cleaning operation is set: S34, the main control module performs statistical comparison on all the missed sweep marks recorded. If the number of missed sweep marks is greater than the missed sweep mark threshold, the unmanned sweeper turns around and returns to perform the missed sweep mark re-sweeping operation.
3. The automatic cleaning method based on environmental perception according to claim 2, It is characterized in that The main control module sets a threshold value of the number of times of supplementary sweeping operations for the supplementary sweeping operation, and the supplementary sweeping operation further comprises the following steps: S341, during the process of performing the supplementary sweeping operation, the main control module detects the changes of the garbage targets in real time and updates the missed sweeping mark record; S342, after a single supplementary sweeping operation is completed, the main control module again performs statistical comparison on all missed sweeping marks recorded, and if the number of missed sweeping marks is greater than the missed sweeping mark threshold, the unmanned sweeper performs the supplementary sweeping operation again; S343, the main control module makes a statistical record of the number of supplementary sweeping operations. If the number of supplementary sweeping operations of the unmanned sweeper exceeds the supplementary sweeping operation number threshold, and the main control module still determines that supplementary sweeping needs to be performed, the patrol sweeping task is forcibly terminated.
4. The automatic cleaning method based on environmental perception according to claim 3, It is characterized in that The main control module sets the garbage clustering level for the garbage target, the re-sweeping steps of the re-sweeping operation are the same as the patrol cleaning steps, and the end point of a single re-sweeping operation is the first missed sweep mark of its single re-sweeping path; the specific steps of the patrol cleaning in step S33 are as follows: S331, the main control module sets a lateral offset range for the inspection path. When the unmanned sweeper passes through the inspection path, it can move within the lateral offset range. If a single garbage target is detected within the lateral offset range, the unmanned sweeper approaches the single garbage target and cleans it. S332: If two garbage targets with similar positions are detected within the lateral offset range, and they are in a side-by-side state with the unmanned sweeper, so that a single pass cannot guarantee the cleaning of all of them, then the garbage is cleaned from high to low according to the garbage clustering level; S333: If three or more garbage targets with similar positions are detected within the lateral offset range, and they are in parallel with the unmanned sweeper, only the garbage target with the highest garbage clustering level is cleaned, and the position of the uncleaned garbage target is recorded as a missed sweep mark; S334: The unmanned sweeper completes the cleaning of garbage targets within the lateral offset range, returns to the inspection path and continues to move forward until a single inspection and cleaning is completed.
5. The automatic cleaning method based on environmental perception according to claim 4, It is characterized in that In the step S32, the main control module also sets up a garbage target re-inspection mechanism. If the main control module finds that a certain garbage target has changed its state during the garbage target re-inspection, the inspection path will be re-planned according to steps S331 to S333. The garbage target re-inspection mechanism is specifically: the visual module collects the global environment image in the working area at a certain frequency, and the main control module records the quantity and location information of each garbage target detected each time, and compares the state changes of the quantity and location information of each garbage target recorded last time with the previous time.
6. The automatic cleaning method based on environment perception according to claim 1, It is characterized in that The garbage detection model is a garbage model algorithm with corresponding garbage model feature values obtained by collecting a large amount of garbage sample image data through a deep neural network architecture for garbage detection model training. The visual module includes a calibrated visual camera and a lidar sensor. The specific method in the S13 step is: inputting the global environment image into the garbage detection model for feature extraction and matching, obtaining the pixel coordinates of the corresponding garbage in the global environment image, and converting the image pixel coordinates into laser map coordinates through the coordinate association relationship between the pre-calibrated visual camera and the lidar sensor to obtain data information such as the location, type and quantity of garbage in the working area.
7. The automatic cleaning method based on environment perception according to claim 1, It is characterized in that The main control module is provided with a garbage amount calibration value and a regional comparison critical value. The steps of determining the garbage distribution type of the operation area in step S13 are as follows: S131, the main control module divides the global environment image collected by the visual module into a number of equally divided sub-areas of equal area; S132, the main control module counts the amount of garbage in each sub-area. If the amount of garbage in the sub-area does not exceed the calibration value, the sub-area is regarded as a non-garbage-intensive sub-area, otherwise it is a garbage-intensive sub-area; S133, the main control module counts the number of non-garbage dense sub-areas and garbage dense sub-areas respectively. If the cumulative total number of garbage dense sub-areas is greater than the total number of non-garbage dense sub-areas and reaches the regional comparison critical value, it belongs to the global dense area, otherwise it belongs to the scattered distribution area.
8. The automatic cleaning method based on environment perception according to claim 4, It is characterized in that When performing cleaning operations within the lateral offset range, the area between the current garbage target position and the nearest garbage target of the unmanned sweeper is the local walking area. If a new garbage target is detected in the local walking area, if the new garbage target is in the local walking area, the unmanned sweeper returns to the inspection path to continue cleaning operations; if the new garbage target is not in the local walking area, the unmanned sweeper returns to the inspection path to clean operations and records the new garbage target as a missed cleaning mark.
9. The automatic cleaning method based on environment perception according to claim 4, It is characterized in that The garbage clustering level includes a large amount of clustered garbage, regular clustered garbage, a small amount of clustered garbage and scattered garbage, and the judgment basis is: setting the garbage accumulation radius, the minimum garbage value, the intermediate garbage value and the maximum garbage value, performing statistical analysis on the garbage accumulation situation, taking a single garbage as the central reference point, and counting the amount of garbage within the garbage accumulation radius. If the counted garbage amount is equal to the minimum garbage value, it is determined to be a scattered garbage point; If the number of statistical garbage is greater than the minimum garbage value but not greater than the intermediate garbage value, it is determined to be a small amount of clustered garbage; If the statistical garbage quantity is greater than the intermediate garbage value but does not exceed the maximum garbage value, it is determined to be regular clustered garbage; if the statistical garbage quantity exceeds the maximum garbage value, it is determined to be a large amount of clustered garbage.
10. The automatic cleaning method based on environment perception according to claim 1, It is characterized in that The state change includes the disappearance of garbage targets, the addition of garbage targets and the change of the position of garbage targets, and the judgment method is: when the unmanned sweeper does not perform cleaning, for the same garbage target, the main control module compares its record in the subsequent detection with the record in the previous detection. If it exists in the previous detection record but does not exist in the subsequent detection record, it is judged that the garbage target disappears, and the main control module clears the garbage target record; if it does not exist in the previous detection record but exists in the subsequent detection record, it is judged that the garbage target is added, and the main control module updates the record of the newly added garbage target in real time; if it exists in both the previous and subsequent detection records, the initial position of the garbage target is used as the central reference point, the displacement judgment radius is set, and the position information of the garbage target in the two records is judged and compared. If the position information of the garbage target in the subsequent detection record is still within the judgment radius relative to the previous detection record, it is judged that there is no change in the position of the garbage target, and the main control module keeps the garbage target record information unchanged; If the location information is not within the determination radius, it is determined that the location of the garbage target has changed, and the main control module clears the initial record of the garbage target and updates the record of the newly added garbage target.
Citation Information
Cited By
Intelligent carriage cleaning method
CN119705351A
Method and device for planning road surface cleaning scheme of urban road
CN120197788A
