Dynamic obstacle avoidance method and system for sensing robot

Abstract

The invention provides a dynamic obstacle avoidance method and system for a sensing robot, and relates to the technical field of robot obstacle avoidance, and the method comprises the steps: obtaining a historical cleaning record; performing cleaning demand analysis, and dividing the target parking lot into a plurality of cleaning subareas; a plurality of sensing robots are configured; interactively obtaining real-time parking information, and constructing a plurality of static obstacle maps; performing cleaning route fitting according to the plurality of cleaning cycle configurations and the plurality of sensing robots; and in the process of controlling the multiple sensing robots to execute the cleaning tasks of the multiple cleaning subareas, the dynamic obstacles are recognized in real time, and dynamic obstacle avoidance optimization is conducted on the multiple standard obstacle avoidance cleaning tracks. According to the invention, the technical problem that the cleaning efficiency of the parking lot is further affected due to the delay of obstacle identification and path planning caused by the high change frequency of the dynamic obstacle in the prior art can be solved, and the automation and management efficiency of the parking lot is improved and the environment of the parking lot is optimized through real-time sensing and dynamic obstacle avoidance.

Description

Technical Field

[0001] The present application relates to the technical field of robot obstacle avoidance, and in particular to a dynamic obstacle avoidance method and system for a perception robot. Background Art

[0002] With the acceleration of urbanization and the development of intelligent technology, large parking lots have become an important infrastructure for urban transportation. In order to meet the cleaning needs of the parking lot environment, the cleaning perception robot integrates multiple sensors (such as lidar, cameras, ultrasonic sensors, etc.) and advanced path planning algorithms to realize the automated cleaning of the parking lot. However, due to the complexity of the parking lot environment, especially the frequent changes of dynamic obstacles, the perception robot needs to have efficient obstacle avoidance capabilities to ensure the smooth completion of the cleaning task. There are a large number of dynamic obstacles in the parking lot environment, such as pedestrians, moving vehicles or other cleaning robots, and the trajectories of these obstacles are irregular and change frequently. Existing perception robots mainly rely on pre-planned static paths, which makes it easy for cleaning robots to pause or miss cleaning when facing frequently changing dynamic environments, affecting the cleaning efficiency of the parking lot.

[0003] In summary, the prior art has a technical problem that the high frequency of changes in dynamic obstacles in the parking lot environment causes delays in obstacle recognition and path planning, further affecting the cleaning efficiency of the parking lot. Summary of the invention

[0004] The purpose of this application is to provide a dynamic obstacle avoidance method and system for a perception robot, so as to solve the technical problem in the prior art that the high frequency of changes of dynamic obstacles in the parking lot environment leads to delays in obstacle recognition and path planning, further affecting the cleaning efficiency of the parking lot.

[0005] In view of the above problems, the present application provides a dynamic obstacle avoidance method and system for a perception robot.

[0006] In a first aspect, the present application provides a dynamic obstacle avoidance method for a perception robot, which is implemented by a dynamic obstacle avoidance system for a perception robot, wherein the dynamic obstacle avoidance method for a perception robot includes: interactively obtaining historical cleaning records of a target parking lot; performing cleaning demand analysis based on the historical cleaning records, and dividing the target parking lot into multiple cleaning sub-areas based on the analysis results, wherein the multiple cleaning sub-areas have multiple cleaning cycle configurations; configuring multiple perception robots in the multiple cleaning sub-areas; interactively obtaining real-time parking information, and constructing multiple static obstacle maps for the multiple cleaning sub-areas based on the real-time parking information; performing cleaning route fitting in the multiple static obstacle maps based on the multiple cleaning cycle configurations and multiple equipment performance information of multiple perception robots to obtain multiple standard obstacle avoidance cleaning trajectories, wherein the multiple standard obstacle avoidance cleaning trajectories have multiple driving speed identifiers; in the process of controlling the multiple perception robots to perform cleaning tasks in the multiple cleaning sub-areas with the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, real-time identifying dynamic obstacles based on local environment perception, and dynamically optimizing the multiple standard obstacle avoidance cleaning trajectories based on the identification results.

[0007] In the second aspect, the present application also provides a dynamic obstacle avoidance system of a perception robot, which is used to execute a dynamic obstacle avoidance method of a perception robot as described in the first aspect, wherein the dynamic obstacle avoidance system of the perception robot includes: a cleaning record acquisition module, which is used to interactively obtain historical cleaning records of a target parking lot; a cleaning area division module, which is used to perform cleaning demand analysis based on the historical cleaning records, and divide the target parking lot into multiple cleaning sub-areas based on the analysis results, wherein the multiple cleaning sub-areas have multiple cleaning cycle configurations; a robot configuration module, which is used to configure multiple perception robots in the multiple cleaning sub-areas; an obstacle map construction module, which is used to interactively obtain historical cleaning records of a target parking lot; a cleaning area division module, which is used to perform cleaning demand analysis based on the historical cleaning records, and divide the target parking lot into multiple cleaning sub-areas based on the analysis results, wherein the multiple cleaning sub-areas have multiple cleaning cycle configurations; The invention relates to a method for controlling the plurality of perception robots to perform cleaning tasks in the plurality of cleaning sub-areas by controlling the plurality of perception robots to perform cleaning tasks in the plurality of cleaning sub-areas with the plurality of driving speed identifiers and the plurality of standard obstacle avoidance cleaning trajectories. The method comprises: obtaining real-time parking information from each other, and constructing multiple static obstacle maps of the plurality of cleaning sub-areas according to the real-time parking information; a route fitting module, wherein the route fitting module is used to perform cleaning route fitting in the plurality of static obstacle maps according to the plurality of cleaning cycle configurations and the plurality of equipment performance information of the plurality of perception robots, and obtain a plurality of standard obstacle avoidance cleaning trajectories, wherein the plurality of standard obstacle avoidance cleaning trajectories have a plurality of driving speed identifiers; an obstacle recognition module, wherein the obstacle recognition module is used to identify dynamic obstacles in real time based on local environment perception during the process of controlling the plurality of perception robots to perform cleaning tasks in the plurality of cleaning sub-areas with the plurality of driving speed identifiers and the plurality of standard obstacle avoidance cleaning trajectories, and dynamically optimize the plurality of standard obstacle avoidance cleaning trajectories according to the recognition results.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] Obtain historical cleaning records of a target parking lot through interaction; perform cleaning demand analysis based on the historical cleaning records, and divide the target parking lot into multiple cleaning sub-areas based on the analysis results, wherein the multiple cleaning sub-areas have multiple cleaning cycle configurations; configure multiple perception robots in the multiple cleaning sub-areas; obtain real-time parking information interactively, and construct multiple static obstacle maps for the multiple cleaning sub-areas based on the real-time parking information; perform cleaning route fitting on the multiple static obstacle maps based on the multiple cleaning cycle configurations and multiple equipment performance information of the multiple perception robots, and obtain multiple standard obstacle avoidance cleaning trajectories, wherein the multiple standard obstacle avoidance cleaning trajectories have multiple driving speed identifiers; in the process of controlling the multiple perception robots to perform cleaning tasks in the multiple cleaning sub-areas with the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, identify dynamic obstacles in real time based on local environment perception, and perform dynamic obstacle avoidance optimization on the multiple standard obstacle avoidance cleaning trajectories based on the identification results. In other words, a static obstacle map of the parking lot (such as fixed vehicles and pillar positions) is constructed through real-time parking information, the standard obstacle avoidance cleaning trajectory is dynamically optimized, and the robot path is adjusted in real time to bypass dynamic obstacles. This improves the real-time nature of obstacle recognition and the flexibility of path planning, improves the automation and management efficiency of the parking lot, and optimizes the parking lot environment.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1 A schematic diagram of a flow chart of a dynamic obstacle avoidance method of a perception robot in this application;

[0013] Figure 2 This is a schematic diagram of the structure of a dynamic obstacle avoidance system of a perception robot in this application.

[0014] Explanation of reference numerals: cleaning record acquisition module 11 , cleaning area division module 12 , robot configuration module 13 , obstacle map construction module 14 , route fitting module 15 , obstacle identification module 16 . DETAILED DESCRIPTION

[0015] This application provides a dynamic obstacle avoidance method and system for a perception robot, which solves the technical problem in the prior art that the high frequency of changes in dynamic obstacles in the parking lot environment leads to delays in obstacle recognition and path planning, further affecting the cleaning efficiency of the parking lot. By building a static obstacle map of the parking lot (such as fixed vehicles and pillar positions) through real-time parking information, the standard obstacle avoidance cleaning trajectory is dynamically optimized, and the robot path is adjusted in real time to bypass dynamic obstacles, which improves the real-time nature of obstacle recognition and the flexibility of path planning, improves the automation and management efficiency of the parking lot, and optimizes the parking lot environment.

[0016] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0017] For example, please refer to the attached Figure 1 The present application provides a dynamic obstacle avoidance method for a perception robot, wherein the dynamic obstacle avoidance method for a perception robot is applied to a dynamic obstacle avoidance system for a perception robot, and the dynamic obstacle avoidance method for a perception robot specifically comprises the following steps:

[0018] Step 1: Interactively obtain the historical cleaning records of the target parking lot.

[0019] Specifically, interacting with the parking management system means obtaining data from the parking management system, including information on cleaning activities of the target parking lot in the past period of time, such as the date, time, cleaning area, cleaning method used, cleaning completion rate, etc. Interaction refers to the process of exchanging data with external data sources or users through database queries, API calls, or manual input. Establish a communication channel with the parking management system, make a request to the parking management system to obtain historical cleaning records, and clarify the time range and specific content of the required records. The parking management system provides historical cleaning records in the form of paper documents, spreadsheets, or special reports based on the request. For example, suppose the parking management system provides cleaning records for the past year, and the records show that comprehensive cleaning is carried out twice a week, using environmentally friendly cleaning agents, and each cleaning is completed by two staff members and takes 3 hours. By obtaining historical cleaning records, understanding the frequency and effect of cleaning can help to allocate cleaning resources and robots more reasonably.

[0020] Step 2: Perform cleaning demand analysis based on the historical cleaning records, and divide the target parking lot into multiple cleaning sub-areas based on the analysis results, wherein the multiple cleaning sub-areas have multiple cleaning cycle configurations.

[0021] Specifically, the cleaning needs of different areas of the parking lot are analyzed based on historical cleaning records, and information such as the cleaning time, frequency, and cleaning content of each area is obtained. The relationship between the cleaning frequency and usage of each area is analyzed to identify which areas require more cleaning work due to heavy traffic or long periods of no cleaning. According to the area, traffic volume, frequency of regional use, pollution level, etc. of the target parking lot, a regional division grid is set to divide the target parking lot into multiple cleaning areas. According to the traffic information of each area and the historical cleaning records, the cleaning cycle required for each area is determined. Areas with heavy traffic may be divided into separate sub-areas, while areas with less cleaning needs may be merged into one sub-area. Each sub-area will have an independent cleaning cycle configuration. For areas with heavy traffic, the cleaning cycle is shorter, which may be cleaned every 2 hours; while for areas with light traffic and less pollution, the cleaning cycle may be longer, which may be cleaned every 12 hours.

[0022] After the parking lot is divided into multiple cleaning sub-areas, different cleaning cycles are configured according to the actual cleaning needs of each area, and they are flexibly adjusted according to the actual situation. Adjacent sample cleaning areas with similar cleaning cycle constraints are merged to form multiple cleaning sub-areas. Each cleaning sub-area has a cleaning cycle configuration, that is, how long it takes to clean the sub-area. Through cleaning demand analysis, the cleaning sub-areas are reasonably divided, and the appropriate cleaning cycle is configured for each sub-area, which effectively avoids over-cleaning or under-cleaning and improves the execution efficiency of cleaning tasks.

[0023] Step three: deploy multiple perception robots in the multiple cleaning sub-areas.

[0024] Specifically, the number of perception robots required is determined based on the area, cleaning frequency, and cleaning cycle of each cleaning sub-area. Each perception robot is responsible for the tasks of a cleaning sub-area. When configuring the number of robots, larger cleaning sub-areas may require multiple perception robots to work in parallel to improve cleaning efficiency. The cleaning tasks in areas with high traffic volume and frequent use are complex, and multiple robots may be required to work in turns or in parallel to reduce cleaning time. A perception robot is a robot used to perform automated cleaning tasks. It is equipped with various sensors (such as lidar, cameras, ultrasonic sensors, etc.), which can perceive the surrounding environment and perform path planning, automatically avoid obstacles, and perform cleaning tasks. Reasonably allocate multiple perception robots to different cleaning sub-areas to ensure that the cleaning tasks of each cleaning sub-area can be completed efficiently. When configuring, you need to select the appropriate number and type of robots based on the area, cleaning requirements, and cycle of each sub-area.

[0025] Different cleaning sub-areas may require different types of perception robots. For example, some areas require efficient cleaning (such as heavy traffic and oil pollution), so robots with strong cleaning capabilities are needed; while for areas with less pollution, robots with relatively weaker cleaning capabilities can be selected. According to the environmental complexity of the cleaning sub-area (for example, whether there are many dynamic obstacles), select robots equipped with different sensors. For example, robots with a combination of lidar and visual sensors are suitable for complex environments. After configuring the robots, it is necessary to ensure that the robots can work together and be scheduled according to the predetermined cleaning cycles and tasks. According to the area and cleaning requirements of the cleaning sub-area, the robots are assigned to specific tasks. The robots enter the corresponding cleaning sub-areas according to the assigned tasks. If there are multiple robots working in the same sub-area, it is necessary to ensure that the multiple robots can avoid mutual interference and clean collaboratively. By reasonably configuring multiple perception robots to different cleaning sub-areas, it can be ensured that the cleaning tasks in each area are completed on time and the backlog of cleaning tasks can be avoided.

[0026] Step 4: interactively obtain real-time parking information, and construct multiple static obstacle maps of the multiple cleaning sub-areas according to the real-time parking information.

[0027] Specifically, by interacting with the parking lot management system, real-time parking information is obtained, including the dynamic distribution information of vehicles in the parking lot, specifically the occupancy of each parking space at the current time, the specific location of the vehicle (two-dimensional or three-dimensional coordinates), whether the lane is occupied, etc., which is collected in real time through the parking lot monitoring system and intelligent sensors (such as geomagnetic sensors and cameras). The real-time parking information is dynamically loaded into the existing regional layout model, and the static obstacle map content is updated. The static obstacle map includes fixed obstacles (such as pillars, walls, etc.) and dynamic obstacles (such as vehicles), which dynamically reflect the parking situation of vehicles in the parking lot.

[0028] The regional layout model of the parking lot is constructed based on the regional layout data and architectural design information of the parking lot, and it is divided into multiple static obstacle maps according to multiple cleaning sub-areas. Combined with the real-time parking information of each area, the final multiple static obstacle maps are obtained. Every time new parking data is obtained, the static obstacle map of the corresponding cleaning sub-area is updated. The static obstacle map of each sub-area is uploaded to the robot navigation system to guide it to avoid all obstacles (fixed and dynamic) and perform cleaning tasks. The robot will re-plan the path according to the dynamic changes of the map to ensure safety and cleaning efficiency. The integration of real-time parking information and fixed obstacle information enables the robot to accurately avoid all obstacles (including vehicles), greatly improving the efficiency and safety of task completion.

[0029] Step 5: Cleaning route fitting is performed on the multiple static obstacle maps according to the multiple cleaning cycle configurations and multiple equipment performance information of the multiple perception robots to obtain multiple standard obstacle avoidance cleaning trajectories, wherein the multiple standard obstacle avoidance cleaning trajectories have multiple driving speed identifiers.

[0030] Specifically, multiple equipment performance information of multiple perception robots is obtained, that is, technical parameters related to the cleaning perception robots, such as maximum driving speed, battery life, cleaning width, cleaning efficiency, etc. Through previous perception and sensor data, a static obstacle map of the target parking lot is obtained, and all unchanged static obstacles are marked. According to the multiple cleaning cycle configurations of multiple cleaning sub-areas and the multiple equipment performance information of multiple perception robots, different cleaning cycles correspond to different cleaning requirements, and the equipment performance information determines the maximum cleaning capacity of the robot. Route fitting is performed on multiple static obstacle maps, including a U-shaped path and a serpentine path, and initial path planning and supplementary path planning are performed on multiple static obstacle maps respectively. Based on the fitted cleaning route, a standard obstacle avoidance cleaning trajectory is generated, which has taken into account the static obstacles and the robot's equipment performance to ensure that the robot can smoothly perform the cleaning task according to the optimal path.

[0031] According to the complexity of multiple cleaning cycle configurations and standard obstacle avoidance cleaning trajectories, different driving speeds are set at each key node of the trajectory. The generated standard obstacle avoidance trajectory is optimized to ensure that each cleaning area is fully covered and there is no repeated cleaning. The driving speed mark refers to the driving speed set by the robot on each cleaning trajectory according to the cleaning requirements of different areas, the density of obstacles, and the performance of the robot. By marking the speed on the cleaning path, the robot can make adaptive adjustments according to different task requirements. The generated standard obstacle avoidance cleaning trajectory is loaded into each perception robot, and the robot's driving is controlled according to the speed mark of the trajectory. When performing a cleaning task, the robot adjusts its driving speed according to the speed mark of each path point. Through route fitting based on cleaning cycle configuration and equipment performance information, combined with static obstacle data, multiple standard obstacle avoidance cleaning trajectories are generated, and specific driving speed marks are assigned to each trajectory, ensuring that the cleaning robot can adjust the cleaning route and speed according to different environments and equipment status, thereby improving cleaning efficiency and work quality, while avoiding interference from obstacles.

[0032] Step six: In the process of controlling the multiple perception robots to perform the cleaning tasks of the multiple cleaning sub-areas using the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, dynamic obstacles are identified in real time based on local environment perception, and dynamic obstacle avoidance optimization is performed on the multiple standard obstacle avoidance cleaning trajectories based on the identification results.

[0033] Specifically, according to multiple driving speed identifiers and corresponding multiple standard obstacle avoidance cleaning trajectories, multiple perception robots are controlled in multiple cleaning sub-areas to enable multiple perception robots to perform cleaning tasks. The local environment is perceived in real time according to the dynamic obstacle perception architecture built into each perception robot, and the real-time perceived data is transmitted to the K dynamic obstacle perception models of the dynamic obstacle perception architecture to identify the spatial characteristics of dynamic obstacles and determine the dynamic obstacles and their position coordinates of the cleaning sub-area, that is, objects that constantly change position or state in the environment, such as moving people, vehicles, other robots, etc.

[0034] By calculating the distance and relative speed between the dynamic obstacles in each cleaning sub-area and the current path of the perception robot in the current cleaning sub-area, it is determined whether obstacle avoidance is required. If the distance between the obstacle and the path is less than a certain threshold, or the speed and direction of the obstacle may cause a collision, it is determined to be a dynamic obstacle avoidance requirement. Dynamic obstacle avoidance optimization is performed on multiple standard obstacle avoidance cleaning trajectories according to the corresponding dynamic obstacle avoidance requirements. In other words, the original cleaning trajectory of the robot is adjusted or optimized in real time according to the perceived dynamic obstacle information, ensuring that the robot can avoid obstacles and continue to perform tasks smoothly without compromising cleaning efficiency. Through real-time perception and dynamic obstacle recognition, the robot can effectively identify and avoid dynamic obstacles, thereby optimizing the cleaning trajectory, which is not only the adjustment of the path, but also includes subtle adjustments to the speed and trajectory according to the real-time status of the obstacle, ensuring that the robot can smoothly perform cleaning tasks without being disturbed by dynamic obstacles.

[0035] Further, step 2 of this application includes:

[0036] A pre-set area division grid is provided; the target parking lot is divided into H sample cleaning areas by using the area division grid; historical traffic flow information of the target parking lot is interactively obtained; cleaning demand analysis is performed based on the historical traffic flow information and historical cleaning records to obtain H cleaning cycle constraints; adjacent areas of the H sample cleaning areas are merged based on the H cleaning cycle constraints to obtain the multiple cleaning sub-areas.

[0037] Specifically, the area division grid refers to dividing the target parking lot into regular grid units, each unit is called a grid, which usually has a fixed shape and size (such as a square or rectangle). According to the area of ​​the target parking lot, traffic flow, frequency of area use, degree of pollution, etc., a suitable grid size is preset to affect the division strength of the cleaning area. The target parking lot is divided into H sample cleaning areas using the area division grid, and the characteristics of each sample cleaning area are recorded, such as location, area, facility distribution, etc. H is an integer greater than or equal to 1, depending on the size of the parking lot and the resolution of the grid.

[0038] Interact with the parking lot management system to obtain the historical traffic information of the target parking lot, including vehicle entry and exit time, length of stay, frequency, etc. Analyze the cleaning needs of each sample cleaning area based on the historical traffic information and cleaning records. Perform a usage frequency distribution analysis on the historical traffic information, count the vehicle usage frequency distribution of each sample area (such as the number of vehicle passes or stay time), and obtain H sample usage frequencies of H sample cleaning areas. Extract the cleaning frequency of each sample area from the historical cleaning records, that is, the number of cleaning times for each sample cleaning area, and obtain H sample cleaning frequencies of H sample cleaning areas.

[0039] Perform cleaning demand analysis on the H sample usage frequencies and H sample cleaning frequencies, analyze the cleaning needs of the H sample cleaning areas, and obtain H cleaning cycle constraints for the H sample cleaning areas, that is, the cleaning frequency and time requirements for each sample cleaning area. If the cleaning cycle constraints of adjacent grids (sample cleaning areas) are similar, they are merged into one cleaning sub-area to form multiple cleaning sub-areas. The larger unit formed by merging adjacent sample cleaning areas (based on the similarity of cleaning cycle constraints) is used for the actual allocation and execution of cleaning tasks. By analyzing traffic information and cleaning records, the cleaning cycle of each cleaning sub-area is customized. Merging adjacent areas can reduce the number of times the cleaning equipment is moved and improve cleaning efficiency.

[0040] Furthermore, the present application also includes the following steps:

[0041] A usage frequency distribution analysis is performed based on the historical traffic flow information to obtain H sample usage frequencies of the H sample cleaning areas; the H sample cleaning frequencies of the H sample cleaning areas are obtained by calling the historical cleaning records; a historical usage frequency set, a historical cleaning frequency set and a historical cleaning cycle set are interactively obtained; a multivariate regression analysis is performed on the historical usage frequency set, the historical cleaning frequency set and the historical cleaning cycle set, and a cleaning cycle calculation model is output; the H sample usage frequencies and the H sample cleaning frequencies are loaded into the cleaning cycle calculation model to solve the cleaning interval cycle, and the H cleaning cycle constraints are obtained.

[0042] Specifically, the usage frequency distribution analysis is performed on the historical traffic information, and the vehicle usage frequency distribution of each sample area is counted (such as the number of vehicle passes or the length of stay), and the H sample usage frequencies of the H sample cleaning areas are obtained, and the high-frequency and low-frequency usage areas of each area in the parking lot are identified. The cleaning frequency of each sample area is extracted from the historical cleaning records, that is, the number of cleaning times of each sample cleaning area, and the H sample cleaning frequencies of the H sample cleaning areas are obtained.

[0043] By interacting with the parking lot management system, we obtain the historical usage frequency set, historical cleaning frequency set, and historical cleaning cycle set, which include the usage frequency, cleaning frequency, and cleaning cycle of all sample cleaning areas (the cleaning cycle is calculated through the cleaning frequency and event records). We perform multiple regression analysis on the historical usage frequency set, historical cleaning frequency set, and historical cleaning cycle set to analyze the relationship between multiple independent variables (such as usage frequency and cleaning frequency) and dependent variables (such as cleaning cycle).

[0044] The historical usage frequency set, historical cleaning frequency set and historical cleaning cycle set are standardized or normalized so that data of different dimensions can be compared within the same range. A cleaning cycle calculation model is established using multiple regression analysis, and the historical usage frequency set, historical cleaning frequency set and historical cleaning cycle set are input into the cleaning cycle calculation model. The model is continuously optimized, that is, the coefficients of the multiple regression equation are adjusted, and finally a cleaning cycle calculation model with high prediction accuracy is obtained. H sample usage frequencies and H sample cleaning frequencies are input into the cleaning cycle calculation model to solve the cleaning interval cycle, and the cleaning interval cycle of each sample cleaning area is obtained. H cleaning cycle constraints are obtained, that is, the recommended cleaning frequency of each sample cleaning area. By using frequency distribution analysis and multiple regression analysis, a cleaning cycle calculation model is established, and the cleaning cycle is predicted or calculated according to the usage frequency and cleaning frequency of each sample cleaning area, which helps to realize the personalized customization of cleaning tasks, improve cleaning efficiency, and ensure the cleaning quality of the parking lot.

[0045] Further, step 4 of this application includes:

[0046] Interactively obtain the regional layout data and architectural design information of the target parking lot; construct a regional space model according to the architectural design information; load the regional layout data into the regional space model to obtain a regional layout model; preset a cleaning cross-scale; expand the multiple cleaning sub-areas according to the cleaning cross-scale to obtain multiple actual cleaning ranges; and divide the regional layout model into the multiple static obstacle maps according to the multiple actual cleaning ranges.

[0047] Specifically, by interacting with the parking lot management system, the regional layout data and architectural design information of the target parking lot are collected. Regional layout data refers to the spatial distribution information of each area inside the parking lot, including the geometric position and size of structural elements such as lanes, entrances and exits, parking spaces, column positions, walls, stairs, etc. The architectural design information includes the building structure, floor height, wall structure, slope design, size, materials and other building-related details of the parking lot. Based on the architectural design information, the regional spatial model of the parking lot is constructed using building information modeling (BIM) software or geographic information system (GIS) tools to reflect its spatial limitations and dimensions. The regional spatial model can be a two-dimensional plan or a three-dimensional stereo model, depending on the cleaning robot's need to understand the environment, and it reflects the physical structure, spatial relationships and geometric characteristics of each area of ​​the parking lot in detail.

[0048] The regional spatial model refers to a three-dimensional or two-dimensional spatial representation model constructed based on architectural design information, which reflects the physical structure and spatial relationship of the parking lot. The actual regional layout data (such as vehicle parking positions, column positions, entrances and exits, etc.) is loaded into the constructed regional spatial model to generate the final regional layout model, ensuring that the spatial model not only contains the static information of the architectural design, but also reflects the actual use of the parking lot. The regional layout model is a comprehensive model generated after the regional layout data is loaded into the regional spatial model, reflecting the spatial structure and functional layout of the parking lot.

[0049] Preset cleaning cross scales to define the cross or overlap range between cleaning sub-areas. The cross scale ensures that the robot does not miss cleaning areas or repeat cleaning when switching between different sub-areas. The preset cleaning cross scale ensures that the robot can seamlessly switch between different sub-areas, avoiding blind spots or repeated cleaning.

[0050] Using the preset cleaning cross-scale, the multiple divided cleaning sub-areas are expanded to generate the actual cleaning range. By expanding the preset cross-scale to the surrounding area based on the original sub-area, an actual cleaning area covering the entire parking lot is formed. By expanding the cleaning sub-areas, it is ensured that the actual cleaning range covers the entire parking lot to avoid missing any area. According to the actual cleaning range after expansion, the regional layout model is divided into multiple static obstacle maps. Each static obstacle map corresponds to an actual cleaning range, and the location and characteristics of fixed obstacles within the range are marked. The static obstacle map is used to guide the robot to perform path planning and obstacle avoidance during the cleaning process, ensuring that the robot can avoid fixed structural obstacles (such as pillars and walls) and efficiently complete the cleaning task. Through the static obstacle map, the robot can accurately identify and avoid fixed obstacles to avoid collisions and cleaning interruptions.

[0051] The static obstacle map is a map generated based on the regional layout model and the actual cleaning range, marking the fixed obstacles in the parking lot (such as pillars, walls, fixed equipment, etc.). Through the regional layout data and architectural design information, an accurate regional spatial model is constructed, and the cleaning sub-area division is optimized based on the cleaning cross-scale to ensure that the cleaning tasks of each area are clear and efficient. The fixed obstacles within the actual cleaning range are accurately marked in the static obstacle map, so that the robot can quickly identify and avoid obstacles, reduce the risk of collision, and improve cleaning efficiency and safety.

[0052] Further, step five of this application includes:

[0053] A second cleaning width is obtained by extracting the second device performance information of the second perceptual robot; a second grid scale is constructed based on the second cleaning width; after selecting the static obstacle distribution in the second static obstacle map, the second static obstacle map is divided into a second cleaning grid array using the second grid scale; grid connection of the second cleaning grid array is performed based on a U-shaped path to obtain a second initial path and multiple discrete cleaning grids; grid connection of the multiple discrete cleaning grids is performed based on a serpentine path to obtain multiple supplementary paths; the multiple supplementary paths are fitted to the second initial path to generate a second standard obstacle avoidance cleaning trajectory; a second driving speed is obtained according to the second cleaning cycle configuration and the trajectory degree of the second standard obstacle avoidance cleaning trajectory, and the second driving speed is used to identify the second standard obstacle avoidance cleaning trajectory.

[0054] Specifically, a perception robot is arbitrarily selected from a plurality of perception robots as the second perception robot. Second device performance information of the second perception robot is obtained, that is, specific parameters related to the cleaning ability of the second perception robot, including cleaning width, cleaning speed, cleaning efficiency, etc. The second cleaning width is extracted from the second device performance information. The cleaning width represents the maximum effective width of the robot for a single cleaning and is an important basis for path planning.

[0055] On the second static obstacle map, select the distribution area of ​​static obstacles. Construct the second grid scale according to the second cleaning width, that is, the size of each grid unit in the static obstacle map. The larger the grid scale, the higher the computational efficiency of the cleaning path planning, but the precision may be reduced; vice versa. The grid scale should match the cleaning width to ensure that each grid can be effectively cleaned by the robot. In the second static obstacle map, the division is based on the grid scale, and the state of each grid unit (obstacle / cleanable) is determined by the static obstacle map. Use the second grid scale to divide the static obstacle map into multiple grids to form a second cleaning grid array.

[0056] The U-shaped path is a path planning mode. The robot connects the grid units in sequence in a rectangular or spiral manner. The robot moves in a rectangular or U-shaped route within the cleaning area, usually gradually covering from the outer edge to the inside. The grids of the second cleaning grid array are connected using a U-shaped path (i.e., moving in a spiral from the outside to the inside or from the inside to the outside) to form a second initial path. The second initial path is the first path obtained by U-shaped path planning. It is the preliminary movement trajectory of the cleaning robot and is used to cover most of the cleaning area. Discrete cleaning grids are grids that are not covered by the U-shaped path. They are usually not cleaned due to obstacles, boundaries or path planning problems, so additional path connections are required.

[0057] To cover multiple discrete cleaning grids, a serpentine path is used for supplementation. In the serpentine path, the robot will use a Z-shaped pattern to shuttle back and forth between each row, connecting and cleaning those grids that are not covered by the U-shaped path. The cleaning area is supplemented by serpentine path planning, filling the discrete grid parts not covered by the U-shaped path, ensuring that all areas are fully covered, and obtaining multiple supplementary paths. Multiple supplementary paths are fitted to the second initial path to generate a complete second standard obstacle avoidance cleaning trajectory, which not only covers all cleaning areas but also avoids static obstacles.

[0058] The second form speed is calculated according to the second cleaning cycle configuration of the second cleaning sub-area and the trajectory complexity of the second standard obstacle avoidance cleaning trajectory. The cleaning cycle configuration takes into account the cleaning frequency and cleaning time, while the trajectory degree takes into account the number of turns and cleaning difficulty of the path. The trajectory degree is used to describe the complexity of the cleaning trajectory, including the turning frequency, path length, direction change, etc., which directly affects the optimization of the robot's driving speed. The calculated second driving speed is used to identify the second standard obstacle avoidance cleaning trajectory to ensure that the robot travels at an appropriate speed during the cleaning process. The second driving speed is the optimal cleaning speed calculated according to the trajectory degree, which ensures cleaning efficiency and stability while avoiding deviations caused by excessive speed in complex trajectory areas. By analogy, the above steps are performed for multiple perception robots and multiple cleaning sub-areas to obtain multiple standard roadblock cleaning trajectories corresponding to multiple perception robots. By extracting the cleaning width from the equipment performance information and constructing a grid division and path planning model, efficient cleaning of complex areas in the parking lot is achieved. Combining trajectory optimization with speed control significantly improves the adaptability and efficiency of the cleaning robot in a dynamic environment.

[0059] Furthermore, step six of this application includes:

[0060] Interactively obtain multiple sample recognition characteristics of multiple sample dynamic obstacles; obtain K types of sample sensors and K groups of sample dynamic obstacles by aggregating the multiple sample recognition characteristics; configure multiple obstacle avoidance sensing arrays in the multiple perception robots according to the K types of sample sensors, wherein each obstacle avoidance sensing array includes K sensor entities; with the K types of perception data of the K types of sample sensors as constraints, obtain K groups of sample perception data sets and K groups of sample perception space feature sets of the K groups of sample dynamic obstacles through network calls; use the K groups of sample perception data sets and the K groups of sample perception space feature sets as training data to construct K dynamic obstacle perception models, wherein the K dynamic obstacle perception models include K groups of dynamic obstacle perception branches connected in parallel; connect the K dynamic obstacle perception models in parallel, and after completing the construction of the dynamic obstacle perception architecture, load the dynamic obstacle perception architecture into the multiple perception robots, and construct communication links between the K sensor entities and the K dynamic obstacle perception models in the multiple perception robots.

[0061] Specifically, the recognition characteristics of multiple sample dynamic obstacles in the target parking lot are interactively obtained, including the shapes, sizes, moving speeds, etc. of different obstacles, to obtain multiple sample recognition characteristics. Test the recognition accuracy of different types of sensors (such as lidar, cameras, ultrasonic sensors, etc.) for specific dynamic obstacles to determine which sensors are more suitable for identifying which obstacles. Aggregate the collected multiple sample recognition characteristics and classify them into K types of sample sensors and K groups of sample dynamic obstacles. In other words, after combining multiple sample recognition characteristics, select the most suitable sensor for each obstacle. For example, lidar is used to identify obstacles at a longer distance, and cameras are used to identify objects with clear morphological features.

[0062] Based on K sample sensors, multiple obstacle avoidance sensing arrays are configured on multiple perception robots. Each obstacle avoidance sensing array includes K sensor entities, and each sensor is responsible for sensing a specific obstacle type or environmental feature. The obstacle avoidance sensing array refers to combining multiple sensors into an array that work together to sense dynamic obstacles in the environment. Each array contains multiple sensor entities that can provide perception information of different angles and types.

[0063] The K perception data types of K sample sensors refer to the data categories that each sensor can obtain. For example, the laser radar collects distance information and the camera collects morphological features. The perception data (such as position, speed, size, etc.) and spatial features (such as distance, direction, relative speed, etc.) of K groups of sample dynamic obstacles are obtained through network connection. The spatial features are the recognition and analysis results of the perception data, and K groups of sample perception data sets and K groups of sample perception spatial feature sets are obtained. The sample perception spatial feature set is a further analysis of the perception data to obtain the spatial relationship information of the obstacle, such as the distance, direction, relative speed, etc. between the robot and the obstacle.

[0064] K groups of sample perception data sets and K groups of sample perception spatial feature sets are used as training data to build K dynamic obstacle perception models. Each model outputs obstacle status information, such as position, speed, and direction, based on the input perception data. Independent dynamic obstacle perception branches are designed for each sensor, and each branch corresponds to a sensor and the type of obstacle it identifies. K groups of dynamic obstacle perception branches are integrated through a parallel architecture to form a complete dynamic obstacle perception capability. Each branch independently processes its corresponding sensor data and passes the perception results to the central module for fusion.

[0065] K dynamic obstacle perception models are connected in parallel to build a complete dynamic obstacle perception architecture. Each dynamic obstacle perception model is responsible for processing data from a specific sensor, such as the lidar branch processing position and contour data, the camera branch processing obstacle category data, and the ultrasonic branch processing close-range detection data. The integration of parallel models uses middleware to perform data fusion and uniformly pass branch outputs to the decision layer. The dynamic obstacle perception architecture is loaded onto multiple perception robots. For example, the dynamic obstacle perception architecture is deployed into the main control module of the perception robot through a software interface to ensure that the architecture is compatible with the hardware in the robot and can receive sensor data in real time and perform dynamic processing.

[0066] Each sensor entity (such as lidar, camera, ultrasonic sensor) communicates with the corresponding perception model through a high-speed communication interface. A low-latency data transmission channel is built to ensure that the data collected by the sensor can be passed to the perception model in milliseconds. The prediction results processed by the dynamic obstacle perception model are returned to the robot decision system through the communication link to guide real-time obstacle avoidance actions. Through the combination of multiple sensors, the recognition accuracy of different types of dynamic obstacles is improved, avoiding misjudgments caused by the limitations of a single sensor.

[0067] Furthermore, the present application also includes the following steps:

[0068] In the process of controlling the first perception robot to perform the cleaning task of the first cleaning sub-area with the first driving speed identifier and the first standard obstacle avoidance cleaning trajectory, the local environment is perceived in real time through the dynamic obstacle perception architecture installed on the first perception robot to obtain K real-time perception data sets; the K real-time perception data sets are mapped and sent to the K dynamic obstacle perception models of the dynamic obstacle perception architecture configured by the first perception robot through a communication link to perform spatial feature recognition of dynamic obstacles to obtain first dynamic obstacle coordinates; the first dynamic obstacle coordinates are fitted to the first standard obstacle avoidance cleaning trajectory to perform dynamic obstacle avoidance demand judgment to obtain a first dynamic obstacle avoidance area; the first standard obstacle avoidance cleaning trajectory is optimized according to the first dynamic obstacle avoidance area; and so on, during the execution of the first standard obstacle avoidance cleaning trajectory, dynamic obstacle avoidance optimization of the cleaning trajectory is performed based on the dynamic obstacle avoidance demand judgment.

[0069] Specifically, the first perception robot is controlled to clean the first cleaning sub-area through the first driving speed mark and the first standard obstacle avoidance cleaning trajectory, and a dynamic obstacle perception architecture is installed inside the first perception robot to perceive the surrounding dynamic obstacles in real time through sensors (such as lidar, cameras, etc.). For example, if there are pedestrians or other moving objects in the parking lot, the perception architecture can detect and generate a real-time perception data set in real time, including the current coordinates, speed, direction and other information of the obstacles. The dynamic obstacle perception architecture refers to the perception system installed on the robot, which is used to identify and monitor dynamic obstacles (such as other vehicles, pedestrians, etc.) in the environment in real time. Data is collected through sensors (such as lidar, cameras, ultrasonic sensors, etc.), and these data are fed back and analyzed in real time with the robot. The real-time perception data set refers to the data set collected in real time by the perception architecture during the execution of the cleaning robot's task. It includes information such as the position, speed, and direction of dynamic obstacles.

[0070] The K real-time perception data sets are mapped and sent to the dynamic obstacle perception architecture of the first perception robot through the communication link, and the K dynamic obstacle perception models are used to analyze the K real-time perception data sets to identify the spatial characteristics of dynamic obstacles, including the relative speed, direction, distance, etc. of the obstacles. The first dynamic obstacle coordinates are the position coordinates of the obstacles in space obtained after the dynamic obstacles are identified. These coordinates help determine the position of the obstacles relative to the robot.

[0071] Based on the spatial characteristic coordinates of dynamic obstacles, the distance and relative speed between the obstacle and the current path of the robot are calculated to determine whether there is a need for dynamic obstacle avoidance. If the distance between the obstacle and the path is less than a certain threshold, or the speed and direction of the obstacle may cause a collision, it will be judged as a dynamic obstacle avoidance need. The first dynamic obstacle avoidance area is determined based on the result of the obstacle avoidance requirement judgment. An area that needs to be avoided is determined. It is the spatial range that the robot needs to avoid or replan the path. The obstacle avoidance area is compared with the first standard obstacle avoidance cleaning trajectory, the obstacle avoidance path is calculated, and the original trajectory is optimized to ensure that the robot avoids dynamic obstacles. During the execution of the cleaning task, the environmental changes are monitored in real time, and the cleaning trajectory is continuously optimized based on the new dynamic obstacle information and obstacle avoidance requirements. The optimized trajectory will be dynamically adjusted to ensure that the robot can avoid obstacles and complete the cleaning task safely.

[0072] Similarly, when multiple perception robots perform cleaning tasks in multiple cleaning sub-areas, the above steps are performed to identify the dynamic obstacles in the cleaning sub-areas, and the corresponding standard obstacle avoidance cleaning trajectory is updated according to the obstacles, and dynamic obstacle avoidance optimization of the cleaning trajectory is performed. By calculating the spatial characteristics of obstacles and their relative relationship with the path, the robot can intelligently avoid obstacles to ensure the smooth execution of cleaning tasks. The path optimization method based on dynamic obstacle avoidance improves the flexibility and safety of the robot in complex environments.

[0073] In summary, the dynamic obstacle avoidance method of a perception robot provided in this application has the following technical effects:

[0074] Obtain historical cleaning records of a target parking lot through interaction; perform cleaning demand analysis based on the historical cleaning records, and divide the target parking lot into multiple cleaning sub-areas based on the analysis results, wherein the multiple cleaning sub-areas have multiple cleaning cycle configurations; configure multiple perception robots in the multiple cleaning sub-areas; obtain real-time parking information interactively, and construct multiple static obstacle maps for the multiple cleaning sub-areas based on the real-time parking information; perform cleaning route fitting on the multiple static obstacle maps based on the multiple cleaning cycle configurations and multiple equipment performance information of the multiple perception robots, and obtain multiple standard obstacle avoidance cleaning trajectories, wherein the multiple standard obstacle avoidance cleaning trajectories have multiple driving speed identifiers; in the process of controlling the multiple perception robots to perform cleaning tasks in the multiple cleaning sub-areas with the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, identify dynamic obstacles in real time based on local environment perception, and perform dynamic obstacle avoidance optimization on the multiple standard obstacle avoidance cleaning trajectories based on the identification results. In other words, a static obstacle map of the parking lot (such as fixed vehicles and pillar positions) is constructed through real-time parking information, the standard obstacle avoidance cleaning trajectory is dynamically optimized, and the robot path is adjusted in real time to bypass dynamic obstacles. This improves the real-time nature of obstacle recognition and the flexibility of path planning, improves the automation and management efficiency of the parking lot, and optimizes the parking lot environment.

[0075] Embodiment 2: Based on the same inventive concept as the dynamic obstacle avoidance method of a perception robot in the aforementioned embodiment 1, the present application also provides a dynamic obstacle avoidance system for a perception robot, please refer to the attached Figure 2 , the dynamic obstacle avoidance system of the perception robot comprises:

[0076] A cleaning record acquisition module 11, the cleaning record acquisition module 11 is used to interactively obtain the historical cleaning records of the target parking lot; a cleaning area division module 12, the cleaning area division module 12 is used to perform cleaning demand analysis based on the historical cleaning records, and divide the target parking lot into multiple cleaning sub-areas according to the analysis results, wherein the multiple cleaning sub-areas have multiple cleaning cycle configurations; a robot configuration module 13, the robot configuration module 13 is used to configure multiple perception robots in the multiple cleaning sub-areas; an obstacle map construction module 14, the obstacle map construction module 14 is used to interactively obtain real-time parking information, and construct multiple cleaning sub-areas according to the real-time parking information. A static obstacle map; a route fitting module 15, wherein the route fitting module 15 is used to perform cleaning route fitting in the multiple static obstacle maps according to the multiple cleaning cycle configurations and the multiple equipment performance information of the multiple perception robots, and obtain multiple standard obstacle avoidance cleaning trajectories, wherein the multiple standard obstacle avoidance cleaning trajectories have multiple driving speed identifiers; an obstacle recognition module 16, wherein the obstacle recognition module 16 is used to control the multiple perception robots to perform the cleaning tasks of the multiple cleaning sub-areas with the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, to identify dynamic obstacles in real time based on local environment perception, and to perform dynamic obstacle avoidance optimization on the multiple standard obstacle avoidance cleaning trajectories according to the recognition results.

[0077] Furthermore, the cleaning area division module 12 in the dynamic obstacle avoidance system of the perception robot is also used for:

[0078] A pre-set area division grid is provided; the target parking lot is divided into H sample cleaning areas by using the area division grid; historical traffic flow information of the target parking lot is interactively obtained; cleaning demand analysis is performed based on the historical traffic flow information and historical cleaning records to obtain H cleaning cycle constraints; adjacent areas of the H sample cleaning areas are merged based on the H cleaning cycle constraints to obtain the multiple cleaning sub-areas.

[0079] Furthermore, the cleaning area division module 12 in the dynamic obstacle avoidance system of the perception robot is also used for:

[0080] A usage frequency distribution analysis is performed based on the historical traffic flow information to obtain H sample usage frequencies of the H sample cleaning areas; the H sample cleaning frequencies of the H sample cleaning areas are obtained by calling the historical cleaning records; a historical usage frequency set, a historical cleaning frequency set and a historical cleaning cycle set are interactively obtained; a multivariate regression analysis is performed on the historical usage frequency set, the historical cleaning frequency set and the historical cleaning cycle set, and a cleaning cycle calculation model is output; the H sample usage frequencies and the H sample cleaning frequencies are loaded into the cleaning cycle calculation model to solve the cleaning interval cycle, and the H cleaning cycle constraints are obtained.

[0081] Furthermore, the robot configuration module 13 in the dynamic obstacle avoidance system of the perception robot is also used for:

[0082] Interactively obtain multiple sample recognition characteristics of multiple sample dynamic obstacles; obtain K types of sample sensors and K groups of sample dynamic obstacles by aggregating the multiple sample recognition characteristics; configure multiple obstacle avoidance sensing arrays in the multiple perception robots according to the K types of sample sensors, wherein each obstacle avoidance sensing array includes K sensor entities; with the K types of perception data of the K types of sample sensors as constraints, obtain K groups of sample perception data sets and K groups of sample perception space feature sets of the K groups of sample dynamic obstacles through network calls; use the K groups of sample perception data sets and the K groups of sample perception space feature sets as training data to construct K dynamic obstacle perception models, wherein the K dynamic obstacle perception models include K groups of dynamic obstacle perception branches connected in parallel; connect the K dynamic obstacle perception models in parallel, and after completing the construction of the dynamic obstacle perception architecture, load the dynamic obstacle perception architecture into the multiple perception robots, and construct communication links between the K sensor entities and the K dynamic obstacle perception models in the multiple perception robots.

[0083] Furthermore, the obstacle map construction module 14 in the dynamic obstacle avoidance system of the perception robot is also used for:

[0084] Interactively obtain the regional layout data and architectural design information of the target parking lot; construct a regional space model according to the architectural design information; load the regional layout data into the regional space model to obtain a regional layout model; preset a cleaning cross-scale; expand the multiple cleaning sub-areas according to the cleaning cross-scale to obtain multiple actual cleaning ranges; and divide the regional layout model into the multiple static obstacle maps according to the multiple actual cleaning ranges.

[0085] Furthermore, the route fitting module 15 in the dynamic obstacle avoidance system of the perception robot is also used for:

[0086] A second cleaning width is obtained by extracting the second device performance information of the second perceptual robot; a second grid scale is constructed based on the second cleaning width; after selecting the static obstacle distribution in the second static obstacle map, the second static obstacle map is divided into a second cleaning grid array using the second grid scale; grid connection of the second cleaning grid array is performed based on a U-shaped path to obtain a second initial path and multiple discrete cleaning grids; grid connection of the multiple discrete cleaning grids is performed based on a serpentine path to obtain multiple supplementary paths; the multiple supplementary paths are fitted to the second initial path to generate a second standard obstacle avoidance cleaning trajectory; a second driving speed is obtained according to the second cleaning cycle configuration and the trajectory degree of the second standard obstacle avoidance cleaning trajectory, and the second driving speed is used to identify the second standard obstacle avoidance cleaning trajectory.

[0087] Furthermore, the obstacle recognition module 16 in the dynamic obstacle avoidance system of the perception robot is also used for:

[0088] In the process of controlling the first perception robot to perform the cleaning task of the first cleaning sub-area with the first driving speed identifier and the first standard obstacle avoidance cleaning trajectory, the local environment is perceived in real time through the dynamic obstacle perception architecture installed on the first perception robot to obtain K real-time perception data sets; the K real-time perception data sets are mapped and sent to the K dynamic obstacle perception models of the dynamic obstacle perception architecture configured by the first perception robot through a communication link to perform spatial feature recognition of dynamic obstacles to obtain first dynamic obstacle coordinates; the first dynamic obstacle coordinates are fitted to the first standard obstacle avoidance cleaning trajectory to perform dynamic obstacle avoidance demand judgment to obtain a first dynamic obstacle avoidance area; the first standard obstacle avoidance cleaning trajectory is optimized according to the first dynamic obstacle avoidance area; and so on, during the execution of the first standard obstacle avoidance cleaning trajectory, dynamic obstacle avoidance optimization of the cleaning trajectory is performed based on the dynamic obstacle avoidance demand judgment.

[0089] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. Figure 1 The dynamic obstacle avoidance method and specific examples of a perception robot in the first embodiment are also applicable to the dynamic obstacle avoidance system of a perception robot in the present embodiment. Through the above detailed description of the dynamic obstacle avoidance method of a perception robot, those skilled in the art can clearly understand the dynamic obstacle avoidance system of a perception robot in the present embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0090] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0091] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A dynamic obstacle avoidance method for a perception robot, characterized in that: include: Interactively obtain the historical cleaning records of the target parking lot; Performing a cleaning demand analysis according to the historical cleaning records, and dividing the target parking lot into a plurality of cleaning sub-areas according to the analysis results, wherein the plurality of cleaning sub-areas have a plurality of cleaning cycle configurations; Arrange a plurality of sensing robots in the plurality of cleaning sub-areas; interactively obtaining real-time parking information, and constructing a plurality of static obstacle maps of the plurality of cleaning sub-areas according to the real-time parking information; According to the multiple cleaning cycle configurations and multiple equipment performance information of the multiple perception robots, cleaning route fitting is performed on the multiple static obstacle maps to obtain multiple standard obstacle avoidance cleaning trajectories, wherein the multiple standard obstacle avoidance cleaning trajectories have multiple driving speed identifiers; In the process of controlling the multiple perception robots to perform cleaning tasks in the multiple cleaning sub-areas using the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, dynamic obstacles are identified in real time based on local environment perception, and dynamic obstacle avoidance optimization is performed on the multiple standard obstacle avoidance cleaning trajectories based on the identification results.

2. A dynamic obstacle avoidance method for a perception robot as claimed in claim 1, characterized in that: The dynamic obstacle avoidance method of the perception robot also includes: Interactively obtain multiple sample recognition characteristics of multiple sample dynamic obstacles; By aggregating the multiple sample identification characteristics, K types of sample sensors and K groups of sample dynamic obstacles are obtained; According to the K types of sample sensors, a plurality of obstacle avoidance sensing arrays are configured on the plurality of sensing robots, wherein each obstacle avoidance sensing array includes K sensor entities; Taking the K perception data types of the K types of sample sensors as constraints, obtaining K groups of sample perception data sets and K groups of sample perception space feature sets of the K groups of sample dynamic obstacles through online calling; Using the K groups of sample perception data sets and the K groups of sample perception space feature sets as training data, constructing K dynamic obstacle perception models, wherein the K dynamic obstacle perception models include K groups of dynamic obstacle perception branches connected in parallel; The K dynamic obstacle perception models are connected in parallel to complete the construction of the dynamic obstacle perception architecture, then the dynamic obstacle perception architecture is loaded into the multiple perception robots, and communication links between the K sensor entities and the K dynamic obstacle perception models are constructed in the multiple perception robots.

3. A dynamic obstacle avoidance method for a perception robot as claimed in claim 2, characterized in that: In the process of controlling the multiple perception robots to perform the cleaning tasks of the multiple cleaning sub-areas using the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, dynamic obstacles are identified in real time based on local environment perception, and dynamic obstacle avoidance optimization is performed on the multiple standard obstacle avoidance cleaning trajectories according to the identification results, further comprising: In the process of controlling the first perception robot to perform the cleaning task of the first cleaning sub-area with the first driving speed identifier and the first standard obstacle avoidance cleaning trajectory, the local environment is perceived in real time by using the dynamic obstacle perception architecture installed on the first perception robot to obtain K real-time perception data sets; Mapping and sending the K real-time perception data sets to the K dynamic obstacle perception models of the dynamic obstacle perception architecture configured by the first perception robot through a communication link to perform spatial feature recognition of dynamic obstacles, and obtain first dynamic obstacle coordinates; Fitting the first dynamic obstacle coordinates to the first standard obstacle avoidance cleaning trajectory to determine the dynamic obstacle avoidance requirement, and obtaining a first dynamic obstacle avoidance area; Optimizing the first standard obstacle avoidance cleaning trajectory according to the first dynamic obstacle avoidance area; By analogy, during the execution of the first standard obstacle avoidance cleaning trajectory, dynamic obstacle avoidance optimization of the cleaning trajectory is performed based on dynamic obstacle avoidance demand judgment.

4. A dynamic obstacle avoidance method for a perception robot as claimed in claim 1, characterized in that: Constructing a plurality of static obstacle maps of the plurality of cleaning sub-areas further comprises: Interactively obtaining regional layout data and architectural design information of the target parking lot; Constructing a regional space model according to the architectural design information; Loading the regional layout data into the regional spatial model to obtain a regional layout model; Preset cleaning cross scale; Expand the plurality of cleaning sub-areas according to the cleaning cross scale to obtain a plurality of actual cleaning ranges; The area layout model is segmented into the plurality of static obstacle maps according to the plurality of actual cleaning ranges.

5. A dynamic obstacle avoidance method for a perception robot as claimed in claim 4, characterized in that: According to the multiple cleaning cycle configurations and multiple equipment performance information of the multiple perception robots, cleaning route fitting is performed on the multiple static obstacle maps to obtain multiple standard obstacle avoidance cleaning trajectories, further comprising: extracting a second cleaning width from second equipment performance information of the second perceptual robot; constructing a second grid scale based on the second cleaning width; After selecting the static obstacle distribution in the second static obstacle map, dividing the second static obstacle map into a second clean grid array using the second grid scale; Connecting the second cleaning grid array based on the U-shaped path to obtain a second initial path and a plurality of discrete cleaning grids; Performing grid connection on the plurality of discrete cleaning grids based on a serpentine path to obtain a plurality of supplementary paths; Fitting the multiple supplementary paths to the second initial path to generate a second standard obstacle avoidance cleaning trajectory; A second driving speed is calculated based on the second cleaning cycle configuration and the trajectory degree of the second standard obstacle avoidance cleaning trajectory, and the second standard obstacle avoidance cleaning trajectory is identified by the second driving speed.

6. A dynamic obstacle avoidance method for a perception robot as claimed in claim 1, characterized in that: Performing a cleaning demand analysis according to the historical cleaning records, and dividing the target parking lot into a plurality of cleaning sub-areas according to the analysis results, further comprising: Preset area division grid; Using the area division grid to divide the target parking lot into H sample cleaning areas; Interactively obtain historical traffic flow information of the target parking lot; Perform cleaning demand analysis based on the historical traffic information and historical cleaning records to obtain H cleaning cycle constraints; Adjacent regions of the H sample cleaning regions are merged according to the H cleaning cycle constraints to obtain the multiple cleaning sub-regions.

7. A dynamic obstacle avoidance method for a perception robot as claimed in claim 6, characterized in that: Perform cleaning demand analysis based on the historical traffic information and historical cleaning records to obtain H cleaning cycle constraints, and also include: Performing a usage frequency distribution analysis based on the historical traffic flow information to obtain H sample usage frequencies of the H sample cleaning areas; Obtaining H sample cleaning frequencies of the H sample cleaning areas from the historical cleaning records; Interactively obtain a historical usage frequency set, a historical cleaning frequency set, and a historical cleaning cycle set; Performing a multiple regression analysis on the historical usage frequency set, the historical cleaning frequency set, and the historical cleaning cycle set, and outputting a cleaning cycle calculation model; The H sample usage frequencies and the H sample cleaning frequencies are loaded into the cleaning cycle calculation model to solve the cleaning interval period, and the H cleaning cycle constraints are obtained.

8. A dynamic obstacle avoidance system for a perception robot, characterized in that: The steps for implementing the dynamic obstacle avoidance method of a perception robot as described in any one of claims 1 to 7, wherein the dynamic obstacle avoidance system of the perception robot comprises: A cleaning record acquisition module, wherein the cleaning record acquisition module is used to interactively acquire historical cleaning records of a target parking lot; A cleaning area division module, the cleaning area division module is used to perform cleaning demand analysis according to the historical cleaning records, and divide the target parking lot into a plurality of cleaning sub-areas according to the analysis results, wherein the plurality of cleaning sub-areas have a plurality of cleaning cycle configurations; A robot configuration module, the robot configuration module is used to configure a plurality of perception robots in the plurality of cleaning sub-areas; An obstacle map construction module, the obstacle map construction module is used to interactively obtain real-time parking information and construct a plurality of static obstacle maps of the plurality of cleaning sub-areas according to the real-time parking information; A route fitting module, the route fitting module is used to perform cleaning route fitting on the multiple static obstacle maps according to the multiple cleaning cycle configurations and multiple equipment performance information of the multiple perception robots, and obtain multiple standard obstacle avoidance cleaning trajectories, wherein the multiple standard obstacle avoidance cleaning trajectories have multiple driving speed identifiers; An obstacle recognition module is used to identify dynamic obstacles in real time based on local environment perception during the process of controlling the multiple perception robots to perform cleaning tasks in the multiple cleaning sub-areas using the multiple driving speed identifiers and the multiple standard obstacle avoidance cleaning trajectories, and to dynamically optimize the multiple standard obstacle avoidance cleaning trajectories based on the recognition results.

Images