Obstacle avoidance method, system and device for a disinfection robot

Abstract

The present invention relates to the field of robot control technology, and specifically to a method, system and device for obstacle avoidance of a pest control robot. Based on an obstacle image, the obstacle state type, obstacle avoidance space state type and obstacle impact type are judged, and the passage operation of the pest control robot is flexibly set, so that the pest control robot has an accurate and flexible obstacle avoidance function.

Description

Technical Field

[0001] The present invention relates to the field of robot control technology, and in particular to an obstacle avoidance method, system and equipment for a pest control robot. Background Art

[0002] Public health and safety has gradually become a hot topic of concern. Places with dense traffic such as hospitals, airports, train stations, and cinemas are prone to increase the risk of bacterial and virus transmission. Therefore, the adoption of germ disinfection measures in public places is particularly important for human life safety and health. Robots are widely used in disinfection work in public places and can replace humans to complete some specific tasks. The traditional manual disinfection method in public places consumes a lot of manpower and material resources, exposing cleaning staff to the virus environment, increasing the risk of infection, and this disinfection method has problems such as inadequate disinfection and sterilization, and sanitary dead corners. The disinfection robot can independently complete indoor disinfection work without supervision, which can not only effectively reduce the disinfection time, but also reduce the risk of cross infection through high-level disinfection mode.

[0003] However, in actual use, the disinfection robot still has limitations in its obstacle avoidance ability, and cannot make flexible obstacle avoidance decisions, resulting in low work reliability and efficiency. Summary of the invention

[0004] The purpose of the present invention is to provide a method, system and equipment for avoiding obstacles of a pest control robot.

[0005] The technical solution of the present invention is as follows:

[0006] A method for avoiding obstacles of a disinfection robot, applied to the disinfection robot, includes the following operations:

[0007] S1 The disinfection robot travels along a preset disinfection path, and if an obstacle is detected on the disinfection path, an obstacle image is acquired;

[0008] S2 obtains the obstacle state type based on the obstacle image;

[0009] If the obstacle state type is a dynamic obstacle, the disinfection robot will sound an alarm;

[0010] If the obstacle state type is a static obstacle, execute S3;

[0011] S3 obtains the obstacle avoidance space state type based on the obstacle image;

[0012] If the obstacle avoidance space state type is that there is an obstacle avoidance space, the disinfection robot passes through;

[0013] If the obstacle avoidance space state type is that there is no obstacle avoidance space, execute S4;

[0014] S4: Obstacle image is segmented and classified to obtain obstacle impact type;

[0015] If the obstacle impact type is a first-class obstacle, the disinfection robot passes through;

[0016] If the obstacle impact type is a second type of obstacle, the disinfection robot passes after replanning the disinfection path;

[0017] If the obstacle impact type is a third type obstacle, the disinfection robot will sound an alarm.

[0018] In the obstacle avoidance method for the disinfection robot as described above, the operation of obtaining the obstacle state type in S2 is specifically: obtaining a number of images within a preset acquisition period, comparing the obstacle positions in the several images, if the obstacle position changes, it is a dynamic obstacle, if the obstacle position does not change, it is a static obstacle.

[0019] In the obstacle avoidance method for the pest control robot as described above, the operation of obtaining the obstacle avoidance space state type in S3 is specifically: obtaining an image of the pest control robot, fusing it with the obstacle image after proportional conversion to obtain a fused image, in which, if the pest control robot is not in contact with the obstacle, the obstacle avoidance space exists; if the pest control robot is in contact with the obstacle, the obstacle avoidance space does not exist.

[0020] In the obstacle avoidance method of the disinfection robot as described above, the operation of obtaining the obstacle impact type in S4 is specifically as follows: the obstacle image is subjected to feature extraction to obtain low-level spatial features and high-level context features, the low-level spatial features and high-level context features are fused to obtain a segmented image, and the segmented image is classified to obtain the obstacle impact type.

[0021] After the operation of obtaining the segmented image, the method further includes loss reduction processing of the segmented image to obtain a fine segmented image, and performing the classification operation on the fine segmented image.

[0022] According to the obstacle avoidance method for the disinfection robot as described above, if the obstacle avoidance space state type is the existence of obstacle avoidance space in S3, the operation of the disinfection robot to pass is specifically as follows: if the obstacle avoidance space state type is the existence of obstacle avoidance space, the obstacle impact type is obtained based on the obstacle image; if the obstacle impact type is a first type of obstacle, the disinfection robot passes after the speed is reduced to the first speed; if the obstacle impact type is a second type of obstacle, the disinfection robot passes after the speed is reduced to the second speed; if the obstacle impact type is a third type of obstacle, the disinfection robot passes after the speed is reduced to the third speed.

[0023] In the obstacle avoidance method for the disinfection robot as described above, the operation of the disinfection robot finding an obstacle in S1 is achieved in the following manner: the disinfection robot detects along the disinfection path, and if the disinfection robot receives a reflected wave from an object, an obstacle is found; if the disinfection robot does not receive a reflected wave from an object, no obstacle is found.

[0024] A pest control robot obstacle avoidance system, comprising:

[0025] The obstacle state type generation module is used to obtain the obstacle image and obtain the obstacle state type;

[0026] An obstacle avoidance space state type generating module, used for obtaining an obstacle avoidance space state type based on the obstacle image;

[0027] An obstacle impact type generation module is used to segment and classify the obstacle image in sequence to obtain the obstacle impact type;

[0028] A disinfection robot motion control module, used to control the motion state of the disinfection robot;

[0029] An alarm module is used to realize the alarm of the disinfection robot.

[0030] A pest control robot obstacle avoidance device comprises a processor and a memory, wherein the processor implements the above-mentioned pest control robot obstacle avoidance method when executing a computer program stored in the memory.

[0031] A computer-readable storage medium is used to store a computer program, wherein the computer program, when executed by a processor, implements the above-mentioned obstacle avoidance method for a pest control robot.

[0032] The beneficial effects of the present invention are:

[0033] The present invention provides an obstacle avoidance method for a pest control robot. Based on an obstacle image, the obstacle state type, obstacle avoidance space state type and obstacle impact type are judged, and the passage operation of the pest control robot is flexibly set, so that the pest control robot has an accurate and flexible obstacle avoidance function.

[0034] The present invention provides an obstacle avoidance method for a pest control robot. The obstacle image is semantically segmented, followed by loss reduction processing, and finally classification processing to obtain a segmented image with clearer texture and contour, which facilitates accurate classification of obstacles and is beneficial to improving the obstacle avoidance function of the pest control robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] By reading the detailed description of the preferred embodiment below, the scheme and advantages of the present application will become clear to those skilled in the art. The accompanying drawings are only for the purpose of illustrating the preferred embodiment and are not to be considered as limiting the present invention.

[0036] In the attached picture:

[0037] Figure 1 Schematic diagram of the flow of the obstacle avoidance method in the embodiment.

[0038] Figure 2 Schematic diagram of the structure of the obstacle avoidance system in the embodiment.

[0039] Figure 3 Schematic diagram of the structure of the obstacle avoidance device in the embodiment. DETAILED DESCRIPTION

[0040] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings.

[0041] This embodiment provides a method for avoiding obstacles for a disinfection robot. Figure 1 , including the following operations:

[0042] S1 The disinfection robot travels along a preset disinfection path, and if an obstacle is detected on the disinfection path, an obstacle image is acquired;

[0043] S2 obtains the obstacle state type based on the obstacle image;

[0044] If the obstacle state type is a dynamic obstacle, the disinfection robot will sound an alarm;

[0045] If the obstacle state type is a static obstacle, execute S3;

[0046] S3 obtains the obstacle avoidance space state type based on the obstacle image;

[0047] If the obstacle avoidance space state type is that there is an obstacle avoidance space, the disinfection robot passes through;

[0048] If the obstacle avoidance space state type is that there is no obstacle avoidance space, execute S4;

[0049] S4: Obstacle image is segmented and classified to obtain obstacle impact type;

[0050] If the obstacle impact type is a first-class obstacle, the disinfection robot passes through;

[0051] If the obstacle impact type is a second type of obstacle, the disinfection robot passes after replanning the disinfection path;

[0052] If the obstacle impact type is a third type obstacle, the disinfection robot will sound an alarm.

[0053] The specific operation method is:

[0054] The S1 disinfection robot travels along a preset disinfection path. If an obstacle is detected on the disinfection path, an image of the obstacle is acquired.

[0055] First, a disinfection area map is established, and a disinfection path is planned within the disinfection area map. The disinfection area map and the disinfection path are input into the disinfection robot system for memory. Then, during the operation of the disinfection robot, if an obstacle is detected, the image of the obstacle is obtained, and the state type of the obstacle is determined.

[0056] When establishing a map of the disinfection area, first obtain the RGB image and depth map of the disinfection area, associate the ground objects in the RGB image with the corresponding three-dimensional point cloud in the depth map, and complete the information fusion of two-dimensional semantic information and three-dimensional point cloud data. Specifically, first perform a downsampling filtering operation on the single-frame point cloud in the acquired depth map, reduce the number of point clouds while retaining the structural features of the point cloud, and reduce the computing power required for subsequent operations. Then use the CPF algorithm to segment the point cloud, and obtain the spatial geometric relationship between the point clouds during the multi-plane segmentation process. During the implementation process, the RANSAC algorithm (Random sample consensus) is used to fit the point cloud to filter out the planes in the point cloud. Then, the Kd tree clustering algorithm (K-dimensiontree) is used to decompose the point cloud into point cloud clusters, and then the point cloud clusters are back-projected onto the RGB image to complete the fusion of the target rectangular detection box in the RGB image and the target rectangular box in the point cloud cluster, thereby obtaining a map with point cloud semantic information. The generated map is combined with the disinfection robot and obstacle distance information to realize the function of the map for local path planning and navigation. Complete 360° shooting of the disinfection area and repeat the above steps to obtain a map containing rich information.

[0057] When the rectangular detection box of the target object in the RGB image and the rectangular detection box of the target object in the point cloud group are fused, the intersection and union ratio between the rectangular detection box and the point cloud group matrix is ​​calculated, and the RGB image and the depth map are multiplied by the intersection and union ratio respectively and then fused to achieve similarity matching and obtain a semantic image with obstacle information (such as obstacle category, shape and position).

[0058] The operation of the disinfection robot finding obstacles in S1 is realized in the following way: the disinfection robot detects along the disinfection path. If the disinfection robot receives the reflected wave from the object, the obstacle is found; if the disinfection robot does not receive the reflected wave from the object, the obstacle is not found. The function of the disinfection robot to receive the reflected wave from the object is realized by radar. The principle is that the radar emits electromagnetic wave energy in a certain direction in space through the antenna. The object in this direction hits the electromagnetic wave and reflects it back. The radar antenna receives the reflected wave and sends it to the receiving device for processing to extract certain information about the object (the distance from the target object to the radar, the rate of change of distance or radial velocity, azimuth, altitude, etc.). The preset safety distance is 1-3m, and the preferred safety distance is 2m. The appropriate safety distance can give the disinfection robot a certain reaction time and facilitate the disinfection robot to obtain a complete obstacle image.

[0059] S2 obtains the obstacle state type based on the obstacle image; if the obstacle state type is a dynamic obstacle, the disinfection robot alarms; if the obstacle state type is a static obstacle, execute S3.

[0060] The operation of obtaining the obstacle state type in S2 is specifically: obtaining a number of images within a preset acquisition period, comparing the obstacle positions in the images, if the obstacle position changes, it is a dynamic obstacle, if the obstacle position does not change, it is a static obstacle. The preset acquisition period is 1-10s, preferably the preset acquisition period is 5s, by comparing the obstacle positions in the obstacle images within 5s, it is convenient to quickly distinguish the obstacle state type of the obstacle.

[0061] The above-mentioned acquisition of the obstacle image can be achieved by installing a visual sensor on the disinfection robot, and the visual sensor is preferably a Kinect camera, and the above-mentioned obstacle image is a depth image of the obstacle; the above-mentioned acquisition of the distance between the disinfection robot and the obstacle can be achieved by installing a distance sensor on the disinfection robot, and the distance sensor is preferably a lidar.

[0062] S3 obtains the obstacle avoidance space state type based on the obstacle image; if the obstacle avoidance space state type is that there is an obstacle avoidance space, the disinfection robot passes; if the obstacle avoidance space state type is that there is no obstacle avoidance space, execute S4.

[0063] The operation of obtaining the obstacle avoidance space state type in S3 is specifically as follows: obtaining the image of the disinfection robot, fusing it with the obstacle image after proportional conversion, and obtaining a fused image. In the fused image, if the disinfection robot does not contact the obstacle, there is an obstacle avoidance space; if the disinfection robot contacts the obstacle, there is no obstacle avoidance space.

[0064] Specifically, an image of the disinfection robot is acquired in advance, and the size of the disinfection robot image is converted into a corresponding proportion according to the detection distance between the disinfection robot and the obstacle, so as to obtain an image of the disinfection robot with the same size ratio as the obstacle image. Then, the obstacle image is used as the bottom layer, and the disinfection robot image is used as the top layer image and combined with the obstacle image to obtain a fused image. In the fused image, if the disinfection robot does not contact the obstacle, there is an obstacle avoidance space; if the disinfection robot contacts the obstacle, there is no obstacle avoidance space.

[0065] In S3, if the obstacle avoidance space state type is the existence of obstacle avoidance space, the operation of the disinfection robot passing is specifically as follows: if the obstacle avoidance space state type is the existence of obstacle avoidance space, based on the obstacle image, the obstacle impact type is obtained; if the obstacle impact type is the first type of obstacle, the disinfection robot speed is reduced to the first speed before passing; if the obstacle impact type is the second type of obstacle, the disinfection robot speed is reduced to the second speed before passing; if the obstacle impact type is the third type of obstacle, the disinfection robot speed is reduced to the third speed before passing. The first speed is 60%-100% of the original speed, preferably 80% of the original speed; the second speed is 30%-60% of the original speed, preferably 45% of the original speed; the third speed is 10%-30% of the original speed, preferably 20% of the original speed. After the obstacle avoidance space exists, the disinfection robot selects different passing speeds according to the obstacle impact type, which can improve the flexibility and safety of the disinfection robot.

[0066] The first, second and third types of obstacles are weak obstacles, general obstacles and dangerous obstacles respectively. Weak obstacles can be obstacles that are easy to move, such as paper scraps, water stains, cartons, plush toys, etc., general obstacles can be obstacles that are large, heavy and difficult to move, such as tables and chairs, and dangerous obstacles can be obstacles that affect disinfection and are inherently dangerous, such as trash cans, glass bottles, flammable and explosive items.

[0067] The operation of obtaining the obstacle impact type based on the obstacle image in S3 is the same as the operation in S4. To save space, the content of image segmentation and classification is described in detail in S4, and S3 is not repeated.

[0068] After the S4 obstacle image is segmented and classified, the obstacle impact type is obtained; if the obstacle impact type is the first type of obstacle, the disinfection robot can pass; if the obstacle impact type is the second type of obstacle, the disinfection robot can pass after the disinfection path is replanned; if the obstacle impact type is the third type of obstacle, the disinfection robot will alarm.

[0069] The operation of obtaining the obstacle impact type in S4 is specifically as follows: feature extraction is performed on the obstacle image to obtain low-level spatial features and high-level context features, the low-level spatial features and high-level context features are fused to obtain a segmented image, and the segmented image is classified to obtain the obstacle impact type.

[0070] Specifically, the BiSeNet network is used to semantically segment the obstacle image. In the BiSeNet network, the obstacle image is extracted in the spatial path with a small convolution step size to obtain low-level spatial features, and the obstacle image is extracted in the fast downsampling semantic path to obtain high-level context features. Then, the low-level spatial features and high-level context features are fused to obtain a segmented image, and then a classifier is used to classify and process the segmented image to obtain the obstacle impact type.

[0071] After the segmentation image is obtained, the segmentation image is further processed to obtain a fine segmentation image, and a classification operation is performed on the fine segmentation image. The loss reduction process is implemented through a cross entropy function to improve the segmentation accuracy and facilitate the acquisition of a fine segmentation image with clearer texture and contour.

[0072] Before formally classifying and processing the segmented images, a dataset containing the first, second, and third types of obstacles is used to pre-train the classifier, and then the pre-trained classifier is used to process the segmented images, so as to quickly and accurately obtain the obstacle impact type. The dataset is the NYU-DepthV2 dataset, which consists of video sequences of various indoor scenes. In this dataset, the color image and the depth image have been aligned, and dense pixel-level annotations are given to ensure the diversity of the training data samples.

[0073] If the obstacle impact type is the first type of obstacle, the disinfection robot will reduce its speed to the fourth speed before passing through. The fourth speed is 40%-60% of the original speed, preferably 50% of the original speed.

[0074] If the obstacle is of the second type, the position of the robot is taken as the new starting point, and the images around the robot are obtained. Based on the images around the robot, the A* algorithm is used to obtain the optimal path. The robot is decelerated to the fifth speed according to the optimal path. The A* algorithm is a mature existing technology, and it is not described here in detail to save space. The fifth speed is 10%-30% of the original speed, preferably 20% of the original speed.

[0075] This embodiment also provides a pest control robot obstacle avoidance system. Figure 2 ,include:

[0076] The obstacle state type generation module is used to obtain the obstacle image and obtain the obstacle state type;

[0077] An obstacle avoidance space state type generation module is used to obtain the obstacle avoidance space state type based on the obstacle image;

[0078] The obstacle impact type generation module is used to segment and classify the obstacle image in sequence to obtain the obstacle impact type;

[0079] The disinfection robot motion control module is used to control the motion state of the disinfection robot;

[0080] The alarm module is used to realize the alarm of the disinfection robot.

[0081] This embodiment also provides an obstacle avoidance device for the disinfection robot, see Figure 3 , including a processor and a memory, wherein the processor implements the above-mentioned obstacle avoidance method for a disinfection robot when executing a computer program stored in the memory.

[0082] This embodiment also provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the above-mentioned obstacle avoidance method for a pest control robot.

[0083] This embodiment provides an obstacle avoidance method for a pest control robot. Based on the obstacle image, the obstacle state type, obstacle avoidance space state type and obstacle impact type are judged, and the passage operation of the pest control robot is flexibly set, so that the pest control robot has accurate and flexible obstacle avoidance function.

[0084] This embodiment provides an obstacle avoidance method for a pest control robot. It semantically segments the obstacle image, then performs loss reduction processing, and finally performs classification processing to obtain a segmented image with clearer texture and contour, which facilitates accurate classification of obstacles and is beneficial to improving the obstacle avoidance function of the pest control robot.

Claims

1. A method for avoiding obstacles of a pest control robot, applied to the pest control robot, characterized in that: The following operations are included: S1 The disinfection robot travels along a preset disinfection path, and if an obstacle is detected on the disinfection path, an obstacle image is acquired; The disinfection route driving is based on the disinfection area map. The operation of establishing the disinfection area map is specifically as follows: obtaining the RGB image and depth map of the disinfection area, performing downsampling filtering operations on the single-frame point cloud in the depth map, reducing the number of point clouds while retaining the structural features of the point clouds, using the CPF algorithm to segment the point clouds, obtaining the spatial geometric relationship between the point clouds in the process of multi-plane segmentation, using the RANSAC algorithm to fit the point clouds, screening out the planes in the point clouds, using the Kd tree clustering algorithm to decompose the point clouds into point cloud clusters, back-projecting the point cloud clusters onto the RGB image, completing the fusion of the target object rectangular detection box in the RGB image and the target object rectangular box in the point cloud cluster, and obtaining a map with point cloud semantic information; completing 360° shooting of the disinfection area, repeating the above operations, and obtaining the disinfection area map; S2 obtains the obstacle state type based on the obstacle image; If the obstacle state type is a dynamic obstacle, the disinfection robot will sound an alarm; If the obstacle state type is a static obstacle, execute S3; S3 obtains the obstacle avoidance space state type based on the obstacle image; specifically, obtains the image of the disinfecting robot, fuses it with the obstacle image after proportional conversion, and obtains a fused image, in which if the disinfecting robot does not contact the obstacle, the obstacle avoidance space exists; if the disinfecting robot contacts the obstacle, the obstacle avoidance space does not exist; If the obstacle avoidance space state type is that there is an obstacle avoidance space, the disinfection robot passes; specifically: if the obstacle avoidance space state type is that there is an obstacle avoidance space, based on the obstacle image, the obstacle impact type is obtained; if the obstacle impact type is a first type of obstacle, the disinfection robot passes after the speed is reduced to the first speed; if the obstacle impact type is a second type of obstacle, the disinfection robot passes after the speed is reduced to the second speed; if the obstacle impact type is a third type of obstacle, the disinfection robot passes after the speed is reduced to the third speed; If the obstacle avoidance space state type is that there is no obstacle avoidance space, execute S4; S4: Obstacle image is segmented and classified to obtain obstacle impact type; Specifically, the obstacle image is subjected to feature extraction to obtain low-level spatial features and high-level context features, the low-level spatial features and high-level context features are fused to obtain a segmented image, and the segmented image is classified to obtain the obstacle impact type; After the operation of obtaining the segmented image, the process further includes loss reduction processing of the segmented image to obtain a fine segmented image, and performing the classification operation on the fine segmented image; If the obstacle impact type is a first-class obstacle, the disinfection robot passes; if the obstacle impact type is a second-class obstacle, the position of the disinfection robot is taken as a new starting point, images around the disinfection robot are obtained, and based on the images around the disinfection robot, the A* algorithm is used to obtain the optimal passage path, and the disinfection robot decelerates to the fifth speed according to the optimal passage path; if the obstacle impact type is a third-class obstacle, the disinfection robot alarms.

2. The obstacle avoidance method of the disinfection robot according to claim 1, characterized in that: The operation of obtaining the obstacle status type in S2 is specifically: Acquire several images within a preset acquisition period, compare the positions of obstacles in the several images, if the position of the obstacle changes, it is a dynamic obstacle, if the position of the obstacle does not change, it is a static obstacle.

3. The obstacle avoidance method of the disinfection robot according to claim 1, characterized in that: The operation of detecting obstacles on the disinfection path in S1 is implemented in the following manner: The disinfection robot detects along the disinfection path. If the disinfection robot receives the reflected wave from the object, it finds an obstacle; if the disinfection robot does not receive the reflected wave from the object, it does not find an obstacle.

4. A pest control robot obstacle avoidance system, used to implement the pest control robot obstacle avoidance method according to claim 1, characterized in that: include: The obstacle state type generation module is used to obtain the obstacle image and obtain the obstacle state type; The disinfection route driving is based on the disinfection area map. The operation of establishing the disinfection area map is specifically as follows: obtaining the RGB image and depth map of the disinfection area, performing downsampling filtering operations on the single-frame point cloud in the depth map, reducing the number of point clouds while retaining the structural features of the point clouds, using the CPF algorithm to segment the point clouds, obtaining the spatial geometric relationship between the point clouds in the process of multi-plane segmentation, using the RANSAC algorithm to fit the point clouds, screening out the planes in the point clouds, using the Kd tree clustering algorithm to decompose the point clouds into point cloud clusters, back-projecting the point cloud clusters onto the RGB image, completing the fusion of the target object rectangular detection box in the RGB image and the target object rectangular box in the point cloud cluster, and obtaining a map with point cloud semantic information; completing 360° shooting of the disinfection area, repeating the above operations, and obtaining the disinfection area map; The obstacle avoidance space state type generation module is used to obtain the obstacle avoidance space state type based on the obstacle image; specifically: obtain the image of the disinfecting robot, fuse it with the obstacle image after proportional conversion, and obtain a fused image, in which if the disinfecting robot does not contact the obstacle, the obstacle avoidance space exists; if the disinfecting robot contacts the obstacle, the obstacle avoidance space does not exist; The obstacle impact type generation module is used to obtain the obstacle impact type after segmenting and classifying the obstacle image in sequence; specifically: the obstacle image is subjected to feature extraction to obtain low-level spatial features and high-level context features, the low-level spatial features and high-level context features are fused to obtain a segmented image, and the segmented image is classified to obtain the obstacle impact type; After the operation of obtaining the segmented image, it also includes reducing the loss of the segmented image to obtain a fine segmented image, and performing the classification operation on the fine segmented image; if the obstacle impact type is a first-class obstacle, the disinfecting robot motion control module is executed; if the obstacle impact type is a second-class obstacle, the position of the disinfecting robot is taken as a new starting point, and the images around the killing robot are obtained. Based on the images around the disinfecting robot, the A* algorithm is used to obtain the optimal passage path, and the disinfecting robot motion control module is executed according to the optimal passage path; if the obstacle impact type is a third-class obstacle, the alarm module is executed; A disinfection robot motion control module, used to control the motion state of the disinfection robot; An alarm module is used to realize the alarm of the disinfection robot.

5. An obstacle avoidance device for a disinfection robot, characterized in that: It comprises a processor and a memory, wherein the processor implements the obstacle avoidance method of the disinfection robot described in any one of claims 1 to 3 when executing the computer program stored in the memory.

6. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein the computer program, when executed by a processor, implements the obstacle avoidance method for the disinfection robot described in any one of claims 1-3.

Images