A method and device for a mobile robot to get out of trouble, and a mobile robot

Abstract

This application discloses a method for escaping a mobile robot from a jam. The method includes: detecting, on the mobile robot side, whether there are dynamic obstacles with variable states in the current environment, detecting whether the mobile robot is in a trapped state, and if a dynamic obstacle is detected and the mobile robot is in a trapped state, virtualizing the area where the dynamic obstacle is located as an obstacle-free area, using the virtual obstacle-free area to adjust a planned path, and moving based on the adjusted planned path. This application achieves rapid escape from a jam in a dynamic environment.

Description

Technical Field

[0001] The present invention relates to the field of intelligent robots, and in particular, to a method and apparatus for a mobile robot to escape from being trapped, and a mobile robot. Background Art

[0002] With the intelligence of robots, more and more mobile robots can adjust the planned path when detecting obstacles to avoid being unable to move forward due to obstacles.

[0003] In the case of a large number of dense obstacles, it is inevitable that the mobile robot cannot reach the target position point, which will cause the robot to be trapped and unable to escape; or, it needs to go through multiple collisions to escape, with a low success rate of escaping and easy damage to the hardware. Summary of the Invention

[0004] The present invention provides a method for a mobile robot to escape from being trapped to improve the speed of the mobile robot to escape from being trapped.

[0005] A method for a mobile robot to escape from being trapped provided by the present invention includes:

[0006] On the side of the mobile robot,

[0007] Detect whether there are dynamic obstacles with variable states in the current environment, and detect whether the mobile robot is in a trapped state.

[0008] If a dynamic obstacle is detected and the mobile robot is in a trapped state, then virtualize the position area where the detected dynamic obstacle is located as an obstacle-free area, and use the virtual obstacle-free area to adjust the planned path.

[0009] Proceed based on the adjusted planned path.

[0010] Preferably, the method further includes:

[0011] Obtain the range of the trapped area of the mobile robot.

[0012] In the case where the range of the trapped area of the mobile robot is less than the set area threshold, adopt a random movement method to escape from being trapped.

[0013] In the case where the range of the trapped area of the mobile robot is not less than the set area threshold, execute the step of virtualizing the position area where the detected dynamic obstacle is located as an obstacle-free area and using the virtual obstacle-free area to adjust the planned path.

[0014] Preferably, the detecting whether there are dynamic obstacles with variable states in the current environment includes:

[0015] Obtaining navigation map information of the current environment, wherein the navigation map information includes semantic information,

[0016] identifying the dynamic obstacle based on semantic information having obstacle category information in the navigation map information;

[0017] The step of virtualizing the area where the dynamic obstacle is detected as an obstacle-free area and adjusting the planned path using the virtual obstacle-free area includes:

[0018] Based on the navigation map information, the planned path is adjusted according to the area determined if the detected dynamic obstacle is deemed not to exist.

[0019] Preferably, obtaining the range of the area where the mobile robot is trapped includes:

[0020] Based on the navigation map information,

[0021] Acquire first position information of the mobile robot being trapped, and acquire first edge contour information for characterizing the range of the trapped area based on the first position information.

[0022] The first edge contour information is used to determine the scope of the trapped area.

[0023] Preferably, adjusting the planned path in the area determined according to the detected dynamic obstacle being deemed to be absent includes:

[0024] Acquire second position information of the dynamic obstacle and second edge contour information of the dynamic obstacle, where the second edge contour information is used to characterize the spatial distribution of the dynamic obstacle.

[0025] According to the first edge contour information and the second edge contour information, it is determined whether there is a coincidence point between the first edge contour and the second edge contour or whether the number of coincidence points is greater than a set first number threshold,

[0026] If yes, the dynamic obstacle information in the navigation map information is shielded, and a route is planned according to the navigation map information after shielding the dynamic obstacle information.

[0027] Otherwise, it is determined whether there are any coincidence points between the first edge contour information and the existing planned path or whether the number of coincidence points is greater than a set second number threshold. If so, the dynamic obstacle information in the navigation map information is shielded, and a path is planned according to the navigation map information after shielding the dynamic obstacle information.

[0028] Repeatedly execute the step of determining whether there is a matching point between the first edge contour and the second edge contour or whether the number of matching points is greater than a set first quantity threshold according to the first edge contour information and the second edge contour information until the number of repeated executions reaches a set number threshold or the duration of the trapped state reaches a set time threshold.

[0029] Preferably, the obtaining of the navigation map information of the current environment includes:

[0030] Obtain image information and first map information in the current environment,

[0031] Perform object detection based on the image information to obtain an object detection result for classifying the objects in the image, where the object detection result at least includes the classification results of static objects and dynamic objects.

[0032] Use the object detection result to determine the semantic information of the map point cloud in the first map information, and obtain a navigation map including semantic information.

[0033] Preferably, the detecting whether the mobile robot is in a trapped state includes:

[0034] Plan the current path based on the navigation map information,

[0035] If a path from the current position to the target position cannot be planned, it is determined that the robot is in a trapped state;

[0036] The using the object detection result to determine the semantic information of the map point cloud in the first map information includes:

[0037] According to the first map information, use the external and internal camera parameters to project the currently obtained map points onto the image, and obtain the position information of the projected points of the map points in the pixel coordinate system.

[0038] According to the object box in the object detection result where the projected point position information is located, determine the classification result corresponding to the object box as the semantic information of the projected point.

[0039] The present invention also provides a processing device for the mobile robot to get out of trouble. The device includes a processor and a memory. The memory stores a computer program, and the processor is configured to execute the computer program to implement the steps of any one of the processing methods for the mobile robot to get out of trouble.

[0040] The present invention further provides a mobile robot, including a processor and a memory. The memory stores a computer program, and the processor is configured to execute the computer program to implement the steps of any one of the processing methods for the mobile robot to get out of trouble.

[0041] The present invention further provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the processing methods for the mobile robot to escape from trouble are implemented.

[0042] The processing method for the mobile robot to escape from trouble provided by the embodiment of the present application, through the detection of dynamic obstacles, virtualizes the position area where the detected dynamic obstacles are located as an obstacle-free area, and uses the virtual obstacle-free area to adjust the planned path, so that the current dynamic obstacles are processed as if there are no obstacles, and virtualizes the current dynamic obstacles as an obstacle-free passable area, thereby expanding the passable area for the cleaning robot in a trapped state. Planning a path based on the virtual expanded passable area is beneficial to improving the success rate of escaping from trouble, so that the cleaning robot can quickly escape from trouble, and realizes escaping from trouble in a dynamic environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a schematic flowchart of a processing method for the mobile robot to escape from trouble according to an embodiment of the present application;

[0044] Figure 2 is a schematic diagram of planning a path in a static obstacle environment;

[0045] Figure 3 is a schematic diagram in an environment with dynamic obstacles;

[0046] Figure 4 is a schematic flowchart of a processing method for the mobile robot to escape from trouble according to an embodiment of the present application;

[0047] Figure 5 is a schematic diagram of projecting map points into an image;

[0048] Figure 6 is a schematic diagram of obtaining first edge contour information and obstacle edge contour information;

[0049] Figure 7 is a schematic diagram of a processing device for the mobile robot to escape from trouble according to an embodiment of the present application;

[0050] Figure 8 is another schematic diagram of the processing device for the mobile robot to escape from trouble or the mobile robot according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] In order to make the purpose, technical means and advantages of the present application clearer, the following further describes the present application in detail with reference to the accompanying drawings.

[0052] The applicant's research has found that existing escape methods do not distinguish between the states of obstacles, treating both static and dynamic obstacles as static. However, dynamic obstacles, whose states can change, are unavoidable during mobile robot movement. This makes it easy for a mobile robot to become trapped in a certain area when following a pre-planned path for static obstacles. The dynamic changes of dynamic obstacles not only increase the probability of being trapped, but also make escape more difficult.

[0053] In view of this, an embodiment of the present application proposes a method for a mobile robot to escape from distress, which can realize escape processing in a dynamic environment.

[0054] See also Figure 1 As shown, Figure 1 This is a flow chart of a method for escaping a mobile robot according to an embodiment of the present application. The method includes: on the mobile robot side,

[0055] Step 101: Detect whether there are dynamic obstacles with variable states in the current environment, and detect whether the mobile robot is in a trapped state.

[0056] The dynamic obstacle with a variable state may be an obstacle with a variable position state, or an obstacle with a variable shape and / or size; the change may be real-time or non-real-time.

[0057] Step 102: If a dynamic obstacle is detected and the mobile robot is in a trapped state, the area where the dynamic obstacle is detected is virtualized as an obstacle-free area, and the planned path is adjusted using the virtual obstacle-free area.

[0058] Step 103: proceed based on the adjusted planned path.

[0059] Compared with the escape method that does not distinguish between dynamic and static obstacles, the embodiment of the present application can use the obstacle-free area virtualized by dynamic obstacles to adjust the planned path, expand the passable area in the trapped situation, increase the success rate of escape, and help to improve the speed of escape.

[0060] For ease of understanding, the following description will be given using a cleaning robot as an example. It should be understood that the present application is not limited to the application scenarios of cleaning robots, but is applicable to any mobile robot with mobility.

[0061] During operation, the cleaning robot records all detected obstacles in the navigation map. During navigation from the current location to the target location, the cleaning robot will use the historical obstacle map information to plan a path to avoid obstacles and reach the target location.

[0062] See also Figure 2As shown, Figure 2 This is a diagram of path planning in an environment with stationary obstacles. When the obstacles in the environment are stationary, the cleaning robot can easily plan a feasible path from the current location to the target location based on the detected obstacle information. However, when there are dynamic objects in the environment, the robot may encounter dynamic obstacles while traveling along the currently planned path. In this case, the dynamic obstacle information encountered will be recorded in the navigation map. In the event of a collision, the cleaning robot will replan the path based on the latest obstacle map. See Figure 3 As shown, Figure 3 This diagram shows an environment with dynamic obstacles. When there are many dynamic obstacles or the environment is narrow, the previously traversable area will be blocked by the recorded dynamic obstacles, making it impossible to plan a path from the current position to the target location. This can cause the cleaning robot to become trapped and the cleaning task to terminate abnormally.

[0063] As an example, the cleaning robot uses a single-line rotating laser radar and a monocular camera as sensors. The camera can obtain image information in the current environment so as to perform target detection based on the image information; the laser radar can obtain spatial point information in the current environment so as to establish a point cloud map including map points as a navigation map; through the external parameters installed by the camera and the laser radar, the map points are projected onto the image to obtain the semantic information of the corresponding point cloud, and then a navigation map containing semantic information is constructed, so that when the cleaning robot is in a trapped state, dynamic obstacles in the navigation map can be purposefully shielded to weaken the impact of dynamic obstacles on the planned path, thereby improving the passability of the cleaning robot and helping the cleaning robot to quickly escape.

[0064] See also Figure 4 As shown, Figure 4 This is a flow chart of a method for escaping a mobile robot according to an embodiment of the present application. The method includes:

[0065] Step 401: Obtain first map information of the current environment, perform positioning based on the first map information, and determine the current position of the cleaning robot.

[0066] As an example, the first map information may be pre-stored existing map information or real-time map information constructed using SLAM. Positioning based on the first map information may use machine vision to obtain current position information.

[0067] Step 402: Acquire current image information, and perform target detection based on the current image information to identify at least static obstacles and dynamic obstacles, and obtain a target detection result.

[0068] As an example, object detection can be to classify objects in an image using a machine learning model. The number of classifications can be determined according to the capabilities of the machine learning model. For example, it can be classified into static obstacles and dynamic obstacles, or it can be classified into multiple categories such as furniture, pets, and daily necessities. Then, according to the classification results, they are marked as static obstacles and dynamic obstacles. For example, common tables, chairs, cabinets, etc. are used as static obstacles, and shoes, pets, etc. are used as dynamic obstacles.

[0069] Step 403: Based on the current object detection results and the currently observed map point information, determine the semantic information of the map points in the first map information.

[0070] As an example, the cleaning robot projects the currently collected laser points as map points into the pixel coordinate system of the camera. Preferably, the laser points are clustered, and the clustered laser points are projected into the pixel coordinate system. Specifically,

[0071] Assume that the relative displacement between the lidar and the camera is t, and the relative rotation matrix is R. The map points currently obtained by the lidar are P = (p1, p2,...), where p i = (x i , y i , z i ), where x, y, and z are the coordinates of the current map point in the world coordinate system. The currently obtained image is where u and v are the coordinates of the current pixel point in the pixel coordinate system. Assume that the camera uses a pinhole model, and its internal parameters are K.

[0072] First, convert the map point coordinates to the camera coordinate system, which is expressed mathematically as:

[0073]

[0074] where, is the coordinate of the map point p i in the camera coordinate system;

[0075] Then, convert the spatial point in the camera coordinate system to the pixel coordinate system, which is expressed mathematically as:

[0076]

[0077] where, is the z coordinate of the map point p i in the camera coordinate system.

[0078] Based on the object detection results, the semantic information of the map points corresponding to the projection points located within the object bounding boxes can be determined. That is, according to the object bounding boxes in the object detection results where the projection point position information is located, the classification result corresponding to the object bounding box is determined as the semantic information of the projection point.

[0079] See Figure 5 as shown in Figure 5 Figure 7 is a schematic diagram of projecting map points onto an image. In the figure, the semantic information of the map points corresponding to the projection points located within the object bounding boxes is a shoe or a dynamic obstacle.

[0080] Through the above steps 401 to 403, a current navigation map including semantic information can be obtained. Moreover, through object detection performed on the acquired current image information, it can be detected whether there are dynamic obstacles.

[0081] As another example, if a navigation map including semantic information is obtained through other means, for example, by acquiring image information through a camera other than the cleaning robot, then it is also possible to detect whether there are dynamic obstacles based on the semantic information in the navigation map including semantic information.

[0082] Step 404, the cleaning robot detects whether it is in a trapped state. If so, step 405 is executed; otherwise, normal cleaning is performed.

[0083] As an example, path planning is performed based on the current position information and the current navigation map including semantic information. If a planned path from the current position to the target position cannot be planned according to the current navigation map, it is determined that the robot is in a trapped state.

[0084] Step 405, in the case of detecting a dynamic obstacle, the first edge contour information for characterizing the range of the trapped area, the second position information of the detected dynamic obstacle, and the second edge contour information of the detected dynamic obstacle for characterizing the spatial distribution of the dynamic obstacle are acquired.

[0085] As an example, the current position information is determined as the first position information of the trap. Based on the navigation map including semantic information, the second position information of the detected dynamic obstacle is determined, and the first edge contour information and the second edge contour information are searched. The first edge contour information characterizes the range of the trapped area.

[0086] To improve the accuracy of the edge contour information, clustering processing can be performed on the edge contour information.

[0087] See Figure 6 as shown in Figure 6A schematic diagram of obtaining first edge contour information and obstacle edge contour information. In the figure, the cleaning robot searches at its current location based on the navigation map to obtain first edge contour information and second edge contour information, such as the edge contour of the gray portion 601 and the edge contour of the dynamic obstacle 602 in the figure.

[0088] Step 406: Obtain the scope of the trapped area using the first edge contour information.

[0089] As an example, the area formed by the first edge contour is calculated according to the first edge contour information to serve as the range of the trapped area.

[0090] In the following steps, the planned path will be adjusted based on the navigation map information according to the areas determined in which the detected dynamic obstacles are considered to be absent.

[0091] As an example, step 407 determines whether the trapped area is smaller than the set area threshold. If so, it means that the current trapped area is small and the random movement escape method can be used to escape. Otherwise, it means that the current trapped area is large and step 408 is executed.

[0092] Step 408: Based on the first edge contour information and the second edge contour information, determine whether there is a coincidence point with the same position information between the first edge contour and the second edge contour.

[0093] If there are coincidence points or the number of coincidence points reaches a set first number threshold, the dynamic obstacle information corresponding to the second edge contour in the navigation map including the semantic information is shielded according to the second position information, that is, the dynamic obstacle information corresponding to the second edge contour in the navigation map is removed to regard the dynamic obstacle as not existing, and a path is planned based on the navigation map after removing the dynamic obstacle information, thereby adjusting the planned path, and traveling along the current planned path, returning to step 404 until the number of times of returning to step 404 reaches a set number threshold or the duration of the trapped state reaches a set time threshold.

[0094] If it does not exist, then execute step 409.

[0095] In this step, as an example, the intersection of the first edge contour and the second edge contour can be calculated to determine whether there is a matching point between the first edge contour and the second edge contour; or it can be determined directly through the edge contour information, that is, if the two edge contours have the same position information, then the same position part is the matching point.

[0096] Step 409: Based on the second edge contour information and the existing planned path information, determine whether there is a coincidence point between the second edge contour and the existing planned path.

[0097] If there are coincidence points or the number of coincidence points reaches a set second number threshold, the dynamic obstacle information corresponding to the second edge contour in the navigation map including the semantic information is shielded, that is, the dynamic obstacle information corresponding to the second edge contour in the navigation map is removed to regard the dynamic obstacle as not existing; and a path is planned based on the navigation map after removing the dynamic obstacle information, thereby adjusting the planned path, and traveling along the current planned path, returning to step 404 until the number of times returning to step 404 reaches a set number threshold or the duration of the trapped state reaches a set time threshold.

[0098] Otherwise, the process returns to step 404 until the number of times of returning to step 404 reaches a set number threshold or the duration of the trapped state reaches a set time threshold, and then reports the trapped information.

[0099] In this step, as an example, the intersection of the second edge contour and the existing planned path can be calculated to determine whether the first and second edge contours have a coincidence point. Alternatively, the determination can be made based on position information. That is, if the second edge contour and the planned path have the same position information, the portion at the same position is the coincidence point. The existing planned path can be a historical planned path or an original planned path.

[0100] This embodiment uses a navigation map marked with dynamic obstacle information to shield the current dynamic obstacle information in the trapped state, so that the current dynamic obstacle is processed as a form without obstacle, and the current dynamic obstacle is virtualized as a passable area without obstacle, thereby expanding the passable area for the cleaning robot in the trapped state. Planning a path based on the virtually expanded passable area is conducive to improving the success rate of escaping, so that the cleaning robot can quickly escape. Multiple combinations of escape logic can form an escape strategy, so that multiple planned paths can be tried, which is also conducive to improving the success rate of escaping. Compared with the method of treating dynamic obstacles as static obstacles during the escape process, this embodiment can make full use of the dynamic variability of the spatial position of dynamic obstacles, transform unfavorable factors into unfavorable factors, and thus increase the success rate of escaping.

[0101] See also Figure 7 As shown, Figure 7 This is a schematic diagram of a processing device for a mobile robot to escape from distress according to an embodiment of the present application. The device includes:

[0102] The detection module is used to detect whether there are dynamic obstacles with variable states in the current environment, and to detect whether the mobile robot is in a trapped state.

[0103] The escape processing module is used to detect a dynamic obstacle and the mobile robot is in a trapped state, then the area where the dynamic obstacle is detected is virtualized as an obstacle-free area, and the planned path is adjusted using the virtual obstacle-free area.

[0104] The mobile control module moves based on the adjusted planned path.

[0105] The apparatus further comprises:

[0106] The escape strategy selection module is used to obtain the range of the trapped area of the mobile robot. When the range of the trapped area of the mobile robot is smaller than the set area threshold, the escape strategy is performed by random movement. When the range of the trapped area of the mobile robot is not smaller than the set area threshold, the processing module is triggered to work.

[0107] The detection module includes:

[0108] a dynamic obstacle detection submodule, configured to identify the dynamic obstacle based on the acquired navigation map information of the current environment, wherein the navigation map information includes semantic information, and according to the semantic information having obstacle category information in the navigation map information;

[0109] The trapped state detection submodule is used to plan the current path based on the navigation map information. If the path from the current location to the target location cannot be planned, it is determined to be in a trapped state.

[0110] The escape processing module is configured to adjust the planned path according to the area determined when the detected dynamic obstacle is deemed not to exist based on the navigation map information.

[0111] The device also includes:

[0112] The navigation map acquisition module is used to obtain image information and first map information in the current environment, perform target detection based on the image information, and obtain target detection results for classifying targets in the image, wherein the target detection results include at least classification results of static targets and dynamic targets. Using the target detection results, the semantic information of the map point cloud in the first map information is determined to obtain a navigation map including the semantic information.

[0113] See also Figure 8 As shown, Figure 8 Another schematic diagram of a processing device for escaping a mobile robot or a mobile robot according to an embodiment of the present application, comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement any step of the processing method for escaping a mobile robot.

[0114] The memory may include a Random Access Memory (RAM), or may also include a Non-Volatile Memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0115] The aforementioned processor may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0116] An embodiment of the present invention also provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the steps of any of the processing methods for the mobile robot to escape from trouble are implemented.

[0117] For the device / network-side device / storage medium embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for the relevant parts, refer to the partial description of the method embodiment.

[0118] In this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0119] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A processing method for a mobile robot to escape from trouble, characterized in that, The method includes: On the mobile robot side, Obtaining navigation map information of the current environment, wherein the navigation map information includes semantic information, identifying dynamic obstacles based on the semantic information including obstacle category information in the navigation map information, and detecting whether there are dynamic obstacles with variable states in the current environment, Based on the navigation map information, first position information of the mobile robot being trapped is obtained, first edge contour information for characterizing the range of the trapped area is obtained according to the first position information, and the range of the trapped area is determined using the first edge contour information to detect whether the mobile robot is in a trapped state. If a dynamic obstacle is detected and the mobile robot is in a trapped state, the area where the dynamic obstacle is detected will be virtualized as an obstacle-free area, and the planned path will be adjusted using the virtual obstacle-free area. Travel based on the adjusted planned path; in, The step of virtualizing the area where the dynamic obstacle is detected as an obstacle-free area and adjusting the planned path using the virtual obstacle-free area includes: Based on the navigation map information, the planned path is adjusted according to the area determined if the detected dynamic obstacle is deemed not to exist.

2. The processing method according to claim 1, wherein, The method further comprises: Get the range of the area where the mobile robot is trapped, When the mobile robot is trapped in an area smaller than the set area threshold, it will use random movement to escape. When the mobile robot is trapped in an area not less than a set area threshold, the step of virtualizing the area where the dynamic obstacle is detected as an obstacle-free area and adjusting the planned path using the virtual obstacle-free area is performed.

3. The processing method according to claim 1, characterized in that, The adjusting the planned path according to the area determined to be free of detected dynamic obstacles includes: Acquire second position information of the dynamic obstacle and second edge contour information of the dynamic obstacle, where the second edge contour information is used to characterize the spatial distribution of the dynamic obstacle. According to the first edge contour information and the second edge contour information, it is determined whether there is a coincidence point between the first edge contour and the second edge contour or whether the number of coincidence points is greater than a set first number threshold, If yes, the dynamic obstacle information in the navigation map information is shielded, and the route is planned according to the navigation map information after shielding the dynamic obstacle information. Otherwise, it is determined whether there are any coincidence points between the first edge contour information and the existing planned path or whether the number of coincidence points is greater than a set second number threshold. If so, the dynamic obstacle information in the navigation map information is shielded, and a path is planned according to the navigation map information after shielding the dynamic obstacle information. Repeat the step of determining whether there are coincidence points between the first edge contour and the second edge contour or whether the number of coincidence points is greater than a set first number threshold based on the first edge contour information and the second edge contour information, until the number of repetitions reaches the set number threshold or the duration of the trapped state reaches the set time threshold.

4. The processing method according to claim 1, characterized in that, The obtaining of navigation map information of the current environment includes: Get the image information and first map information of the current environment, Performing target detection based on image information to obtain target detection results for classifying targets in the image, wherein the target detection results at least include classification results of static targets and dynamic targets, The target detection result is used to determine the semantic information of the map point cloud in the first map information, and obtain a navigation map including the semantic information.

5. The processing method according to claim 1, characterized in that, The detecting whether the mobile robot is in a trapped state comprises: Plan the current path based on navigation map information, If a path from the current location to the target location cannot be planned, it is determined to be in a trapped state.

6. The processing method according to claim 4, characterized in that The determining of semantic information of the map point cloud in the first map information by using the target detection result includes: According to the first map information, the camera external parameters and internal parameters are used to project the currently obtained map point into the image to obtain the projection point position information of the map point in the pixel coordinate system. According to the target frame in the target detection result where the projection point position information is located, the classification result corresponding to the target frame is determined as the semantic information of the projection point.

7. A processing device for a mobile robot to get out of trouble, characterized in that, The device includes a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement the steps of the processing method for escaping a mobile robot according to any one of claims 1 to 6.

8. A mobile robot, characterized in that, The method comprises a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement the steps of the method for escaping a mobile robot as claimed in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for escaping a mobile robot as claimed in any one of claims 1 to 6.

Images