Sweeping robot cruising path planning method and system based on environment modeling

Abstract

The invention discloses a sweeping robot cruise path planning method and system based on environment modeling, and relates to the technical field of machine learning. The method comprises the following steps: acquiring raster data through a sensor, and constructing a dynamic environment map; performing local path planning from the starting point according to the map to generate a local path; after local coverage is completed, backtracking target points in an uncovered area to generate a global path; global and local paths are fused and mapped to a map for optimization, and a dynamic cruise real-time path is formulated; and controlling the robot to execute and perceive in real time, updating a map adjustment path, and constructing a dynamic path. The technical problems that the local area is not covered and the path planning is not accurate when the sweeping robot carries out the path planning due to the continuous change of obstacles and paths in a dynamic environment are solved, and the purpose of carrying out the joint path planning by sensing the environment change in real time and backtracking and selecting the uncovered area is achieved. And the full coverage efficiency and the path planning precision of the sweeping robot in the dynamic environment are improved.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and specifically to a method and system for navigation path planning of a sweeping robot based on environmental modeling. Background Technology

[0002] With the rapid development of smart home technology, robotic vacuum cleaners have become a common cleaning tool in modern homes. However, in practical applications, traditional robotic vacuum cleaners still face significant challenges in their path planning capabilities. Many existing robotic vacuum cleaners rely on preset static maps for navigation, making them unable to effectively adapt to dynamic changes in the home environment, such as furniture movement and the appearance of temporary obstacles, resulting in low cleaning coverage and frequent repetitive cleaning. Furthermore, due to the lack of intelligent backtracking mechanisms for uncleaned areas and global path connection strategies, robots often cannot efficiently transition to the next uncleaned area after completing local cleaning, leading to a waste of time and energy. Summary of the Invention

[0003] This application provides a method and system for navigation path planning of a robotic vacuum cleaner based on environmental modeling, which solves the technical problems that robotic vacuum cleaners are prone to incomplete coverage of local areas and inaccurate path planning when performing path planning due to the continuous changes of obstacles and paths in dynamic environments.

[0004] The first aspect of this application provides a method for planning the navigation path of a robotic vacuum cleaner based on environmental modeling, the method comprising: Sensors mounted on the robotic vacuum cleaner collect grid data of the environment to construct a dynamic environmental map. Based on the dynamic environmental map, a local coverage path is planned according to the starting point to generate a local coverage path. When the local coverage path is completed, the target point of the uncovered area is selected by backtracking to plan the area connection path and generate a global connection path. The global connection path and the local coverage path are merged to construct an initial full coverage path, which is then mapped onto the dynamic environmental map for optimization to formulate a cruise path. The robotic vacuum cleaner is controlled to execute the cruise path for real-time perception, and the dynamic environmental map is updated in reverse based on the perception results to adjust the cruise path in real-time, constructing a dynamic cruise real-time path.

[0005] A second aspect of this application provides a navigation path planning system for a robotic vacuum cleaner based on environmental modeling, the system comprising: Data Acquisition Module: Collects grid data of the environment using sensors mounted on the robot vacuum cleaner to construct a dynamic environment map; Local Path Planning Module: Plans a local area coverage path based on the dynamic environment map and starting point to generate a local coverage path; Global Path Planning Module: When local area coverage is completed according to the local coverage path, it plans a regional connection path based on backtracking and selecting target points in uncovered areas to generate a global connection path; Path Fusion and Optimization Module: Fuses the global connection path with the local coverage path to construct an initial full coverage path, maps it to the dynamic environment map for optimization, and formulates a cruise path; Real-time Path Adjustment Module: Controls the robot vacuum cleaner to execute the cruise path for real-time perception, and updates the dynamic environment map based on the perception results to adjust the cruise path in real-time, constructing a dynamic cruise real-time path.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, sensors on the robot vacuum collect grid data of the environment to build a dynamic environmental map. Then, based on this map, it plans a cleaning path for a localized area from the starting point, generating a corresponding localized coverage path. After cleaning a localized area, the robot backtracks and selects the uncleaned areas, planning connecting paths to ensure these uncovered areas are cleaned. Next, the localized coverage path is merged with the global connecting path to generate a preliminary full-coverage path, which is then mapped onto the dynamic environmental map for optimization, creating a complete navigation path. Finally, the robot executes cleaning according to this path, updating the dynamic environmental map in real time during execution and adjusting the navigation path accordingly to ensure efficient and accurate cleaning. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic diagram of the navigation path planning method for a sweeping robot based on environmental modeling, provided in an embodiment of this application.

[0009] Figure 2 A schematic diagram of the navigation path planning system for a sweeping robot based on environmental modeling, provided in an embodiment of this application.

[0010] Figure labeling: Data acquisition module 11, Local path planning module 12, Global path planning module 13, Path fusion and optimization module 14, Real-time path adjustment module 15. Detailed Implementation

[0011] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0012] Example 1, as Figure 1 As shown, this application provides a method for cruise path planning of a sweeping robot based on environmental modeling. The method includes: By collecting grid data of the environment through sensors mounted on the robot vacuum cleaner, a dynamic environmental map is constructed.

[0013] In this embodiment, the robotic vacuum cleaner is equipped with various sensors, such as lidar, infrared sensors, and ultrasonic sensors, to acquire information about the surrounding environment in real time. The raw data collected by the sensors includes distance, obstacle locations, ground type, and environmental features. This raw data is transmitted to the robot's processing unit for processing and transformation. In the processing unit, the system performs rasterization processing on the collected raw data, dividing the entire environment into small grid cells. Then, by performing multi-source data fusion calculations on each grid cell and combining it with environmental features, a dynamic activity value is calculated and assigned to each grid cell. These dynamic activity values ​​reflect the frequency of change, importance, or cleaning priority of the area. Finally, based on the dynamic activity values ​​of these grid cells, a dynamic environmental map is formed, reflecting the current state of the robot's environment in real time. This dynamic environmental map is not only a static obstacle distribution map but also reflects dynamic changes in the environment, such as changes in furniture position and object movement, providing real-time and effective data support for subsequent path planning and navigation.

[0014] Furthermore, by using sensors mounted on a robotic vacuum cleaner to collect grid data of the environment and construct a dynamic environmental map, the methods include: The process involves: activating sensors mounted on a robotic vacuum cleaner to sense the environment and obtain a raw environmental dataset; segmenting the environment into grids based on the robot's task requirements and setting a grid resolution; constructing a two-dimensional grid coordinate system using the robot's initial position as the origin and the grid resolution; uniformly dividing the environmental space into multiple grid cells based on this coordinate system; mapping the raw environmental dataset to these grid cells for multi-source fusion calculation to obtain multiple occupancy probabilities, which correspond to the grid cells; and assigning dynamic activity values ​​to the grid cells based on these occupancy probabilities and environmental features to construct a dynamic environmental map.

[0015] Preferably, the initial setup involves importing the desired working environment information. This information can come from user commands input via a mobile application or control panel, such as the cleaning area's range and cleaning mode, or from previously scanned environmental maps stored within the robot's own memory. Based on this information, the robot uses its onboard sensors to collect surrounding environmental data, forming a raw environmental dataset. This dataset includes, but is not limited to, obstacle distance information, terrain information, obstacle dimensions, ground type, and wall boundaries. Subsequently, based on the robot's task requirements, the area to be cleaned is divided into several uniform small areas, each corresponding to a grid cell. The grid segmentation process divides the entire environment into a regular grid structure. These grid cells reflect the size and structural characteristics of the environmental space. The grid resolution is set according to the actual cleaning needs. For example, if precise cleaning or navigating smaller obstacles is required, the grid resolution is set to the lowest level; conversely, if the cleaning area is large and there are few obstacles, the grid resolution can be set to the highest level. Next, using the robot vacuum's current position as the origin, and combining it with the set grid resolution, a two-dimensional grid coordinate system is established along the horizontal and vertical directions. This coordinate system has the horizontal x-axis and the vertical y-axis, with each grid cell having a unique coordinate value (x, y) within this system, allowing for precise positioning of the robot vacuum in the environment. By establishing this grid coordinate system, the robot vacuum can accurately pinpoint the position of each grid cell and calculate its relative position to each grid cell in the environment. After constructing the two-dimensional grid coordinate system, the robot vacuum uniformly divides the entire environmental space into multiple grid cells. Each grid cell is a regular square, its size determined by the grid resolution. For example, if the grid resolution is 10cm × 10cm, then each grid cell represents a 10cm × 10cm actual area. Through spatial division, a set of multiple grid cells is obtained, covering the entire working environment space, providing a foundation for subsequent environmental data mapping and processing. Next, the robotic vacuum cleaner maps the raw environmental datasets collected by its sensors into each grid cell. For example, the distance information of an obstacle measured by a LiDAR sensor can determine the grid cell where the obstacle is located based on its direction and distance. Since data from different sensors has different characteristics and advantages, to more accurately describe the environmental state of each grid cell, it is necessary to fuse the raw environmental datasets from multiple sensors. For instance, LiDAR sensors have high accuracy in distance measurement but may not be able to identify the type of object; vision sensors can identify object types but may have some errors in distance measurement. By fusing data from these two types of sensors, it is possible to more accurately determine whether an obstacle exists within the grid cell and what type of obstacle it is.Fusion computing can employ methods such as weighted averaging and Bayesian estimation, assigning appropriate weights based on the reliability and accuracy of different sensors to ultimately obtain the probability value of each grid cell being occupied. This probability value typically ranges from 0 to 1, where 0 indicates the grid cell is completely empty and not occupied by any object; 1 indicates the grid cell is completely occupied by an obstacle; and values ​​between 0 and 1 represent the likelihood of the grid cell being occupied. By fusing this multi-source data, the occupancy probability of each grid cell can be obtained, reflecting the probability that the grid cell is occupied by an obstacle. In addition to the occupancy probability, the environment also has other characteristics, such as the stability of obstacles (whether they are fixed or moving), the material of the floor (wooden flooring, carpet, etc.), and the cleaning priority of the area (e.g., the cleaning priority of the living room may be higher than that of the storage room). These environmental characteristics will affect the cleaning strategy and path planning of the robotic vacuum cleaner in that area. The system assigns a dynamic activity value to each grid cell based on its occupancy probability and the aforementioned environmental characteristics. This dynamic activity value is a comprehensive indicator reflecting the environmental state and importance of the grid cell. For example, a grid cell with a high occupancy probability and that is a fixed obstacle can be assigned a lower activity value, indicating that the robot needs to avoid that area. Conversely, a grid cell with a low occupancy probability, suitable floor material, and high cleaning priority can be assigned a higher activity value, indicating that the robot should prioritize cleaning that area. The specific value is obtained by weighting these characteristics. Finally, the dynamic activity values ​​of all grid cells are integrated to form a dynamic environmental map. This map not only includes the occupancy probability information of each grid cell but also reflects dynamic information such as environmental characteristics and cleaning priorities. The robotic vacuum cleaner can adjust its navigation path in real time based on this dynamic environmental map to adapt to the constantly changing environment, achieving efficient and intelligent cleaning.

[0016] Based on the dynamic environment map, a local area coverage path is planned according to the starting point to generate a local coverage path.

[0017] In one embodiment, after constructing the dynamic environment map, the robot vacuum cleaner performs a coverage search based on its current location information and the dynamic activity value of each grid cell in the dynamic environment map, generating multiple coverage search results. Based on these coverage search results, a local coverage path is generated by calculating the search cost. This local coverage path guides the robot vacuum cleaner to effectively cover the designated area while ensuring efficient cleaning of all uncleaned areas.

[0018] Furthermore, based on the dynamic environment map, a local area coverage path is planned according to the starting point to generate a local coverage path. The method includes: Based on the dynamic activity value of the dynamic environment map, a direction-first analysis is performed to construct a search strategy. Historical cruise logs are introduced, containing multiple historical cruise areas. Using the robot's location as the starting point, the environmental space is divided into multiple local areas based on these historical cruise areas. A coverage search is performed on these local areas according to the search strategy to generate multiple coverage search results. Search costs are calculated based on these multiple coverage search results to determine multiple cost items. The search strategy is then executed according to these cost items to generate the local coverage path.

[0019] Preferably, the robot vacuum first performs direction-priority analysis based on the dynamic activity value of each grid cell in the dynamic environment map. This direction-priority analysis is based on gradient incremental analysis and greedy calculation, which guides the robot to select the optimal search direction. For example, if the activity value of the grid cell in front is higher than that of the current grid cell, the front is set as the search direction; if the activity values ​​on the left and right sides differ significantly, the side with the higher activity value is selected as the search direction. By summarizing these search directions, a search strategy can be constructed. Subsequently, a historical navigation log is introduced. This historical navigation log records the robot's previous cleaning trajectory and coverage area under different environmental conditions, usually including information such as the boundaries of cleaned areas and the location of obstacles. The system uses the robot vacuum's current position as the origin and combines the boundaries of the historical navigation area to dynamically divide the environmental space into multiple local areas. For example, the living room is divided into sub-areas such as "sofa area," "coffee table area," and "TV cabinet area," each sub-area corresponding to a different activity value distribution and cleaning priority. Next, within the defined local area, a coverage search is performed according to the search strategy. During this process, the robot can start from its current position and explore along the direction of the search strategy, recording the changes in the activity values ​​of uncovered grids along the coverage path to form multiple coverage search results. For these coverage search results, the system calculates the search cost for each result. Specifically, it calculates the Euclidean distance from the current position to the next grid as the path length cost. This cost ensures that the path is as short as possible, reducing unnecessary movement. Based on the activity value density of the local area (the sum of activity values ​​per unit area) and the uncovered time (time since the last cleaning), the system calculates the coverage efficiency cost by weighting the inverse of the activity value density with the uncovered time. Generally, areas with high activity value density and long uncovered time have low coverage efficiency costs and are prioritized for coverage; conversely, areas with high costs are covered later. The system also considers the robot's kinematic constraints (such as the maximum turning angle) and turning energy consumption to calculate the turning cost. For example, if a 90° turn is required from the current direction, the turning cost is the energy consumption value corresponding to the turning angle. Frequent turns increase the total turning cost, so the system tends to choose paths with fewer turns. By summarizing these cost items, multiple cost items can be obtained for each coverage search result. Then, the path length, coverage efficiency, and turning cost are weighted and fused to form a comprehensive cost function. Then, according to the principle of minimizing the comprehensive cost, a coverage search is performed on each local area. Within the "sofa area", paths with short path lengths, high coverage efficiency, and few turning times are prioritized for cleaning.Finally, based on the search results, the search cost of each local area is recalculated, and the path is optimized for high-cost areas. For example, if the path length cost of a certain area is too high, the path direction is adjusted or intermediate nodes are added to reduce the path length. Then, the optimized local area coverage paths are spliced ​​together to form a complete local coverage path, ensuring that the robot can continuously and efficiently complete the cleaning task of the entire area, while avoiding repeated cleaning.

[0020] Furthermore, based on the dynamic activity value of the dynamic environment map, a direction-priority analysis is performed to construct a search strategy, the method of which includes: A grid activity value distribution map is constructed based on the dynamic activity value of the dynamic environment map, and the grid activity value distribution map contains activity value gradient data; an incremental analysis is performed according to the activity value gradient data to determine the initial search direction; a multi-neighborhood search is performed with the position of the sweeping robot as the starting point according to the initial search direction to obtain multiple neighborhood grid activity values; a greedy calculation is performed based on the multiple neighborhood grid activity values ​​to construct the search strategy.

[0021] Optionally, based on the dynamic activity values ​​of each grid cell in the dynamic environment map, a continuous activity value surface is generated using an interpolation algorithm. For example, Kriging interpolation is used to smooth the discrete grid activity values, forming a two-dimensional distribution map containing activity value gradient data. This two-dimensional distribution map is then used as the grid activity value distribution map. This map not only displays the activity value of each grid cell but also includes the activity value gradient data between different grid cells, reflecting the rate of change of activity values ​​between different grid cells. Subsequently, based on the activity value gradient data in the grid activity value distribution map, the activity value gradient in each direction (up, down, left, right, and diagonal) is obtained within the eight-neighborhood of the robot's current position. Incremental analysis is then performed based on these activity value gradients. During the analysis, the direction with the largest activity value gradient is preferentially selected as the initial search direction. For example, if the gradient in the northwest direction is +0.4, the northwest 45° direction is set as the initial search direction, ensuring that the robot preferentially moves towards the area with the fastest growth in activity value, achieving efficient coverage. Next, using the robot's current position as the starting point and the initially determined search direction, a multi-neighborhood search is initiated. For each neighboring grid cell, its activity value and Euclidean distance to the current position are recorded. For example, in a 3×3 neighborhood, the center point is the robot's current position, and the surrounding 8 grid cells correspond to activity values ​​in different directions, such as 0.7 for the grid directly in front, 0.6 for the grid to the left front, and 0.8 for the grid to the right front. Then, based on the activity values ​​of the neighboring grid cells, a greedy algorithm is used to select the neighboring grid cell with the highest activity value as the next moving target at each step, in order to optimize the robot's cleaning efficiency in local areas as much as possible. Specifically, the path selection cost is calculated based on the activity value of each neighboring grid cell and their relative positions. Neighboring grid cells with higher activity values ​​are given higher priority, meaning that the robot will prioritize cleaning these high-activity areas. The greedy algorithm will select the area with the highest current activity value and the most urgent coverage at each step, thereby ensuring that the robot can quickly cover key areas and avoid unnecessary repeated paths. Through greedy calculation, the cost of each neighborhood grid can be evaluated based on the activity value, and the search strategy can be determined. This allows the robot vacuum cleaner to flexibly and efficiently adjust its cleaning route in dynamic environments and prioritize important areas, ensuring overall cleaning effectiveness while saving time and energy.

[0022] When local area coverage is completed according to the local coverage path, the regional connection path is planned based on the target points of the uncovered area selected by backtracking, and a global connection path is generated.

[0023] In one embodiment, after the robotic vacuum cleaner completes cleaning of a local area according to a partial coverage path, the system analyzes the cleaned and uncleaned areas. During this process, a backtracking operation is used to identify which areas have not yet been cleaned, especially those areas missed due to path planning limitations or environmental complexity. This backtracking process is based on grid activity value distribution identification using a dynamic environmental map to determine the set of uncovered areas. Then, based on the location of the uncovered areas, one or more target points are selected from these areas. These target points are typically areas close to the robot's current location. By planning regional connection paths for these target points, the optimal path from the robot's current location to the target points can be calculated, forming a global connection path. This global connection path guides the robotic vacuum cleaner from the already cleaned areas to achieve the final goal of full coverage cleaning.

[0024] Furthermore, when local area coverage is completed according to the aforementioned local coverage path, a regional connection path is planned based on backtracking to select target points in uncovered areas, generating a global connection path. The method includes: When local area coverage is completed according to the local coverage path, the distribution of grid activity values ​​is identified based on the dynamic environment map to determine the set of uncovered areas; the location of the robot vacuum cleaner is used as the origin to traverse the set of uncovered areas and calculate distances to obtain multiple distance values; the multiple distance values ​​are arranged in ascending order, and the first distance value in the distance sequence is selected as the index to search the set of uncovered areas to determine the target uncovered area; the boundary grid point of the target uncovered area is used as the target point; and the area connection path is planned based on the target point to generate the global connection path.

[0025] Preferably, after the robotic vacuum cleaner completes cleaning of the current local area according to the local coverage path, the system re-identifies the current environment using the grid activity value distribution in the dynamic environment map. Specifically, it reads the activity value of each grid and identifies grid cells that still maintain high activity values. These grid cells typically have activity values ​​higher than a preset activity value threshold, representing grid cells not covered by the local path. After identifying these discrete uncovered grid cells, the system performs connectivity clustering on these uncovered grid cells to form a continuous set of uncovered regions. For each uncovered region in the set, boundary grid points are extracted using an edge detection algorithm, such as the Sobel operator, as candidate target points for subsequent path planning. Subsequently, using the current position of the robotic vacuum cleaner as the reference origin, the Euclidean distance between each candidate target point in the uncovered region and the robot's current position is calculated. All candidate target point distance values ​​are then sorted in ascending order, and the candidate target point with the smallest distance is selected as the index. This index is then used to retrieve the uncovered region containing that point from the set of uncovered regions, which is then designated as the target uncovered region. Next, the boundary grid points represented by the candidate target points in the uncovered areas are used as target points. The A* algorithm or Dijkstra's algorithm, combined with multi-cost constraints, is used to plan a path from the current position to the target point. For example, in path search, the path length cost term has a weight of 0.6, the turning cost term has a weight of 0.3, and the coverage efficiency term has a weight of 0.1, ensuring the path is short and minimizing unnecessary turns. During path planning, the dynamic environment map is updated in real time, and the path cost is recalculated to ensure the path adapts to environmental changes. For example, if a temporary obstacle appears near the target point, the path automatically detours and selects the next nearest boundary point. Then, the planned path is smoothed using a Bézier curve to reduce path inflection points and lower the robot's turning frequency. For example, a 90° right-angle path is optimized into a smooth 45° curve to reduce turning energy consumption. Finally, the generated connecting path is spatiotemporally aligned with the completed local coverage path to form a complete global navigation path. This global connecting path is used to connect cleaning tasks in different areas, achieving continuous cleaning coverage globally.

[0026] Furthermore, when local area coverage is completed according to the local coverage path, the distribution of raster activity values ​​is identified based on the dynamic environment map to determine the set of uncovered areas. The method includes: When local area coverage is completed according to the local coverage path, the raster activity values ​​of multiple local areas covered by the local coverage path are extracted; a raster activity value threshold is set, and the raster activity values ​​of the multiple local areas are compared with the raster activity value threshold; the raster activity values ​​of local areas with values ​​higher than the raster activity value threshold are identified, and multiple uncovered rasters are delineated based on the identification results; the multiple uncovered rasters are connected and clustered to construct multiple clusters; and region evaluation is performed according to the multiple clusters to construct the set of uncovered regions.

[0027] Optionally, after the robot vacuum completes cleaning a local area, the system uses the mapping relationship between path point coordinates and grid coordinates to filter all grid cells traversed by the path and extract their corresponding grid activity values. Then, a pre-set grid activity value threshold is obtained. This threshold is set based on environmental characteristics and task requirements; for example, in deep cleaning mode, the threshold can be set to 0.5, and in quick cleaning mode, the threshold can be adjusted to 0.3 to expand the coverage area. The extracted grid activity values ​​of the local area are then compared with the grid activity value threshold. If the grid activity value of a certain grid cell is higher than the threshold, it indicates that the area is not effectively covered and requires further cleaning; in this case, the area is marked. Afterwards, by summarizing the marked areas, multiple uncovered grids are obtained. After marking the uncovered grids, the system performs connectivity analysis on these uncovered grids. Specifically, the system first arbitrarily selects an unvisited grid from the set of uncovered grids as a starting point and adds it to the current cluster. Then, using this grid as the center, it checks whether there are other uncovered grids in its four- or eight-neighborhood. If there are unvisited neighboring grids, the grid is added to the same cluster, and the search operation is recursively performed with these neighboring grids as new search nodes. Through continuous recursion, all uncovered grids connected to the starting point can be traversed and classified into the same cluster. After completing the search of this cluster, a new starting point is selected from the remaining unvisited uncovered grids, and the above process is repeated until all uncovered grids have been visited, thereby constructing multiple spatially independent clusters, each cluster representing a connected uncovered region. Then, the multiple clusters are evaluated to quantify indicators such as region area, activity density, and distance from the robot's current position. The region area can be quantified by calculating the number of grids in each cluster; the distance from the robot's current position can be quantified by calculating the distance of each cluster from the robot's current position. Based on these evaluation metrics, the system will construct a set of uncovered areas and prioritize each cluster according to the evaluation results, providing a basis for subsequent path planning and cleaning tasks.

[0028] The global connection path and the local coverage path are merged to construct an initial full coverage path, which is then mapped onto a dynamic environment map for optimization, and a cruise path is determined.

[0029] In one embodiment, after determining the global connection path and the local coverage path, the global connection path and the local coverage path are spatiotemporally aligned, and then a path topology network is constructed through path stitching. Subsequently, this path topology network is adjusted for conflict identification to generate an initial full-coverage path. This initial full-coverage path includes all cleaned and uncleaned areas, ensuring that no area is missed during the cleaning task. Then, this initial full-coverage path is mapped onto a dynamic environment map and optimized in multiple dimensions to formulate the final navigation path. This navigation path not only ensures the robot's full-coverage cleaning task but also fully considers the real-time nature and flexibility of the path, avoiding repeated cleaning or missed areas, and reducing energy consumption and time waste.

[0030] Furthermore, the global connection path and the local coverage path are merged to construct an initial full coverage path, which is then mapped onto the dynamic environment map for optimization, and a cruise path is determined, including: The global connection path and the local coverage path are spatiotemporally aligned to generate path spatiotemporal registration information; the global connection path and the local coverage path are stitched together according to the spatiotemporal registration information to construct a path topology network; conflict detection is performed based on the path topology network, and the path topology network is identified based on the path conflict results to determine multiple conflicting paths; the multiple conflicting paths are locally adjusted, the path continuity of the adjusted paths is verified, and an initial full coverage path is constructed; the initial full coverage path is mapped to the dynamic environment map for multi-dimensional layered optimization to formulate the cruise path.

[0031] Preferably, after obtaining the local coverage path and the global connection path, since the two paths were generated at different times, have inconsistent starting points, and may have different directions of movement, the system performs spatiotemporal alignment on the two paths. That is, it matches the starting position of the global connection path with the ending position of the local coverage path through coordinate mapping to ensure that the two are continuously connected in the same coordinate system. Then, based on the movement sequence of the robot vacuum cleaner on the local coverage path, the access time of the global connection path is synchronized to make the path connection on the time axis reasonable. After the spatiotemporal alignment is completed, path spatiotemporal registration information is generated, including position mapping matrix, time synchronization parameters, direction correction information, etc., for subsequent path stitching. Subsequently, based on the spatiotemporal registration information, the starting point of the global connection path and the ending point of the local coverage path are connected by straight lines to form a continuous path topology network. For example, after cleaning the living room, the robot can directly enter the bedroom for cleaning through the connection path without having to go through the corridor again. This path topology network is stored in a graph structure, where nodes are key points of the path, such as inflection points and target points, and edges are path segments with attributes including path length, turning angle, etc. Next, conflict detection is performed on the path topology network, including spatial conflicts, temporal conflicts, and logical conflicts. Spatial conflicts occur when path segments overlap with obstacle areas, or multiple path segments may overlap or intersect, causing repeated cleaning or path discontinuity. Temporal conflicts occur when multiple path segments occupy the same grid at the same time, or when the connection point is too early or too late. Logical conflicts occur when the path direction changes too drastically, causing the robot to turn sharply beyond kinematic constraints. The system uses graph analysis technology to perform conflict detection in these three aspects of the topology network, marking problematic path segments as conflict paths for subsequent correction. Then, local adjustments are made to the identified conflict paths, such as smoothing conflict path segments using B-spline curves or reducing path intersections by adjusting the turning angle. For example, if a path segment is detected to conflict with an obstacle, the obstacle is bypassed and the path length and turning cost are recalculated to ensure that the adjusted path still meets the multi-cost constraints. After adjustment, path continuity verification algorithms are used, such as checking whether path segments are continuous without breaks and whether the turning angle is smooth, to ensure path continuity. After continuous verification, optimization is completed, forming an initial full-coverage path. Then, to ensure the path fully adapts to the real-time changing environment, the system remaps the initial full-coverage path back to the dynamic environment map and performs multi-dimensional hierarchical optimization.Specifically, the system optimizes the path geometry using Reeds-Shepp curves to reduce inflection points and lower the robot's turning frequency. It also combines the robot's kinematics model to calculate the energy cost of path segments (e.g., straight segments have lower energy consumption, curved segments have higher energy consumption), optimizing the path to minimize total energy consumption. This robot kinematics model can be a dual-wheel differential drive kinematics model, constructing a linear velocity-angular velocity relationship based on the robot's left and right drive wheel speeds and the distance between the drive wheels. This is then combined with the energy consumption-velocity function and path curvature characteristics to construct a path segment energy consumption estimation model, used to calculate the motion energy consumption of different path segments, serving as the basis for energy consumption optimization. Finally, the path obtained after multi-dimensional hierarchical optimization is the robot's final navigation path for performing the cleaning task. This navigation path is characterized by optimality, continuity, and strong adaptability, enabling efficient and intelligent cleaning of the robot vacuum in complex environments.

[0032] The robot vacuum cleaner is controlled to perform real-time perception along the navigation path. Based on the perception results, the dynamic environment map is updated in reverse to plan and adjust the navigation path in real time, thus constructing a dynamic real-time navigation path.

[0033] In one embodiment, when controlling the robotic vacuum cleaner to execute a navigation path, the robot uses its onboard sensors to perceive the environment in real time, continuously collecting environmental data and transmitting this data to the system for real-time analysis, forming real-time perception data of the current environment. Subsequently, this perception data is mapped back onto a dynamic environmental map, and a difference analysis is performed with existing data on the map to identify changes in the environment. Simultaneously, based on this real-time perception data, real-time positioning is performed, and the position is updated in a two-dimensional grid coordinate system to ensure that path planning is synchronized with the robot's current position. Then, based on the differences between the current position and environmental changes, the navigation path is adjusted in real time. By adjusting the path, a more dynamic real-time navigation path that conforms to the current environment can be generated, enabling the robot to bypass new obstacles or avoid impassable areas, ensuring that the cleaning task is performed efficiently and accurately in a dynamic environment.

[0034] Furthermore, controlling the robotic vacuum cleaner to execute the navigation path for real-time perception, and updating the dynamic environment map based on the perception results to plan and adjust the navigation path in real time, thereby constructing a dynamic real-time navigation path, includes the following methods: The system controls the robotic vacuum cleaner to execute the cruise path, activating sensors mounted on the robot to perceive the cruise environment and obtain real-time environmental perception data. This real-time environmental perception data is then mapped back to the dynamic environment map for difference analysis to obtain the degree of environmental change difference. Based on the cruise path, the robot vacuum cleaner is positioned in real-time, and the real-time positioning data is synchronized to the two-dimensional grid coordinate system to determine the real-time grid position. The real-time grid position is used as the first priority, and neighboring grids are sequentially assigned a descending priority order to construct a map update priority sequence. The cruise path is then planned and adjusted in real-time according to the degree of environmental change difference and the map update priority sequence to construct the dynamic real-time cruise path.

[0035] Optionally, while the robotic vacuum cleaner is performing its cleaning task along its navigation path, the system automatically activates various sensors on the robot to continuously perceive its surrounding environment in real time. The same calculations are then performed to generate real-time environmental perception data, reflecting the dynamic changes in the environment during the robot's operation. Subsequently, the robot's real-time pose is used to map this environmental perception data to the grid coordinate system of a dynamic environment map through coordinate transformation, and then back to the corresponding grid cell position on the dynamic environment map. This is compared with previously stored environmental states to perform difference analysis. For example, if the real-time occupancy probability of a grid cell is 0.9, while the occupancy probability of that grid cell on the map is 0.3, the calculated environmental change difference is 0.6. A higher environmental change difference indicates a greater environmental change, requiring more significant adjustments to the path. Afterward, SLAM (Simultaneous Localization and Mapping) technology is used to accurately locate the robotic vacuum cleaner's position in real time. The real-time positioning data is then transformed as described above and simultaneously mapped to a two-dimensional grid coordinate system to determine the robot's current real-time grid position. To effectively update the dynamic environment map, the system prioritizes the real-time grid position as the highest priority grid. Centered on this grid, it distributes priority to surrounding grids in a layer-by-layer diffusion pattern, forming a map update priority sequence. The real-time grid position has the highest priority, the first ring of neighboring grids has the second highest priority, the second ring has the third highest priority, and so on, decreasing until the entire local area is covered. Then, the determined environmental change difference is combined with the map update priority sequence, and real-time planning adjustments are made through steps such as local path replanning, path quality assessment, and smooth path transitions to generate the final dynamic real-time navigation path. The final constructed dynamic real-time navigation path can adaptively respond to the constantly changing environment during cleaning, enabling the robot vacuum to avoid obstacles while maintaining optimal cleaning efficiency.

[0036] Furthermore, the dynamic real-time cruise path is constructed by real-time planning and adjustment of the cruise path based on the environmental change difference and the map update priority sequence, and the method includes: Based on the environmental change differences, the cruise path is analyzed for path adjustment, and a path adjustment level is set. Following the map update priority sequence, the first K grids are selected, centered on the real-time grid position, to form an initial region to be optimized, where K is an integer greater than 1. Environmental change propagation characteristics are introduced and combined with the initial region to be optimized to construct a rolling optimization window, which is used to delineate the region to be optimized. Based on the path adjustment level, local path replanning is performed on the region to be optimized, generating a candidate path set. The candidate path set is traversed to perform path quality evaluation, and the cruise optimization path is determined based on the quality evaluation results. The cruise optimization path is smoothly connected to the cruise path to construct the dynamic real-time cruise path.

[0037] Optionally, the adjustment requirements for the cruise path are first analyzed based on the magnitude of environmental change differences. This is done by mapping the degree of environmental change differences to a pre-defined path adjustment grading table to determine the current path adjustment level. The grading table includes Level 1, Level 2, and Level 3 adjustments. Level 1 is a large-scale adjustment, requiring replanning of a large area; Level 2 is a moderate adjustment, requiring only partial path replanning; and Level 3 is a minor adjustment, involving path smoothing or fine-tuning. Next, based on the previously constructed map update priority sequence, the robot's current real-time grid position is used as the highest priority point. Neighboring grids are then arranged from highest to lowest priority. The top K highest-priority grids are selected from the priority sequence as the initial optimization area, where K is an integer greater than 1, and its value can be dynamically set according to the path adjustment level. Finally, an environmental change propagation model is introduced to analyze the characteristics of environmental change propagation. This model uses a pre-trained Markov chain to predict obstacle movement trends. By combining the predicted movement trend with the initial region to be optimized, a dynamic scrolling window is constructed. For example, predicting the trajectory of an obstacle ahead within 3 seconds expands the window boundary to two grid squares beyond the predicted path, forming a safety buffer. Through this spatiotemporal expansion mechanism based on the scrolling window, the system overlays the initial region to be optimized with propagation features to delineate the final region to be optimized. For instance, when an obstacle is predicted to move to the right, the window expands to the right, ensuring sufficient avoidance space is reserved for the route planning and allowing for the replanning of key areas. After delineating the region to be optimized, the system performs dynamic route replanning of corresponding intensity based on the route adjustment level. For Level 1 adjustments, the aforementioned route planning process is re-executed for global replanning to ensure route safety. For Level 2 adjustments, local area coverage route planning is performed on some affected grid squares. For Level 3 adjustments, only the A algorithm is used to fine-tune conflict nodes or sharp turn nodes to quickly generate candidate paths. After replanning, the system obtains multiple route schemes with different performance characteristics, forming a candidate path set. Next, each path in the candidate path set is evaluated for quality, with evaluation metrics including but not limited to path length, number of turns, safe distance from dynamic obstacles, and coverage efficiency. The system validates these paths using Monte Carlo simulations and selects the path with the highest weighted score as the optimized cruise path. Finally, to avoid path jumps, sharp turns, or breakpoints, the system smoothly connects the optimized cruise path with the original cruise path. This smooth connection can be achieved using cubic Bézier curve fitting to ensure a smooth path transition and reduce abrupt stops and starts by the robot. The final result is a dynamic real-time cruise path that satisfies dynamic environmental changes while seamlessly connecting with the original path. This dynamic real-time cruise path is immediately sent to the robot for execution, enabling efficient path adjustments in dynamic environments and ensuring the robot maintains optimal cleaning performance in complex environments.

[0038] In summary, the embodiments of this application have at least the following technical effects: First, sensors mounted on the robotic vacuum cleaner collect grid data of the environment to construct a dynamic environment map. Next, based on the dynamic environment map, a local coverage path is planned from the starting point, generating a local coverage path. Then, when local coverage is completed according to the local coverage path, a global connection path is generated by backtracking and selecting target points in uncovered areas. Then, the global connection path and the local coverage path are merged to construct an initial full coverage path, which is then mapped onto the dynamic environment map for optimization, thus formulating a cruise path. Finally, the robotic vacuum cleaner executes the cruise path for real-time perception, and the dynamic environment map is updated based on the perception results to adjust the cruise path in real-time, constructing a dynamic real-time cruise path. This solves the technical problem of missing local areas and inaccurate path planning caused by constantly changing obstacles and paths in a dynamic environment. It achieves the technical effect of improving the full coverage efficiency and path planning accuracy of the robotic vacuum cleaner in dynamic environments by sensing environmental changes in real time and backtracking to select uncovered areas for connection path planning.

[0039] Example 2, based on the same inventive concept as the environmental modeling-based navigation path planning method for sweeping robots in the previous examples, such as... Figure 2 As shown, this application provides a navigation path planning system for a robotic vacuum cleaner based on environmental modeling. The system includes: Data acquisition module 11: Collects grid data of the environment using sensors mounted on the robot vacuum cleaner to construct a dynamic environment map; Local path planning module 12: Plans a local area coverage path based on the dynamic environment map and starting point to generate a local coverage path; Global path planning module 13: When local area coverage is completed according to the local coverage path, it plans a regional connection path based on backtracking and selecting target points in uncovered areas to generate a global connection path; Path fusion optimization module 14: Merges the global connection path with the local coverage path to construct an initial full coverage path, maps it to the dynamic environment map for optimization, and formulates a cruise path; Real-time path adjustment module 15: Controls the robot vacuum cleaner to execute the cruise path for real-time perception, and updates the dynamic environment map in reverse based on the perception results to adjust the cruise path in real-time and construct a dynamic cruise real-time path.

[0040] Furthermore, the data acquisition module 11 is used to perform the following methods: The process involves: activating sensors mounted on a robotic vacuum cleaner to sense the environment and obtain a raw environmental dataset; segmenting the environment into grids based on the robot's task requirements and setting a grid resolution; constructing a two-dimensional grid coordinate system using the robot's initial position as the origin and the grid resolution; uniformly dividing the environmental space into multiple grid cells based on this coordinate system; mapping the raw environmental dataset to these grid cells for multi-source fusion calculation to obtain multiple occupancy probabilities, which correspond to the grid cells; and assigning dynamic activity values ​​to the grid cells based on these occupancy probabilities and environmental features to construct a dynamic environmental map.

[0041] Furthermore, the local path planning module 12 is used to perform the following method: Based on the dynamic activity value of the dynamic environment map, a direction-first analysis is performed to construct a search strategy. Historical cruise logs are introduced, containing multiple historical cruise areas. Using the robot's location as the starting point, the environmental space is divided into multiple local areas based on these historical cruise areas. A coverage search is performed on these local areas according to the search strategy to generate multiple coverage search results. Search costs are calculated based on these multiple coverage search results to determine multiple cost items. The search strategy is then executed according to these cost items to generate the local coverage path.

[0042] Furthermore, the local path planning module 12 is used to perform the following method: A grid activity value distribution map is constructed based on the dynamic activity value of the dynamic environment map, and the grid activity value distribution map contains activity value gradient data; an incremental analysis is performed according to the activity value gradient data to determine the initial search direction; a multi-neighborhood search is performed with the position of the sweeping robot as the starting point according to the initial search direction to obtain multiple neighborhood grid activity values; a greedy calculation is performed based on the multiple neighborhood grid activity values ​​to construct the search strategy.

[0043] Furthermore, the global path planning module 13 is used to perform the following methods: When local area coverage is completed according to the local coverage path, the distribution of grid activity values ​​is identified based on the dynamic environment map to determine the set of uncovered areas; the location of the robot vacuum cleaner is used as the origin to traverse the set of uncovered areas and calculate distances to obtain multiple distance values; the multiple distance values ​​are arranged in ascending order, and the first distance value in the distance sequence is selected as the index to search the set of uncovered areas to determine the target uncovered area; the boundary grid point of the target uncovered area is used as the target point; and the area connection path is planned based on the target point to generate the global connection path.

[0044] Furthermore, the global path planning module 13 is used to perform the following methods: When local area coverage is completed according to the local coverage path, the raster activity values ​​of multiple local areas covered by the local coverage path are extracted; a raster activity value threshold is set, and the raster activity values ​​of the multiple local areas are compared with the raster activity value threshold; the raster activity values ​​of local areas with values ​​higher than the raster activity value threshold are identified, and multiple uncovered rasters are delineated based on the identification results; the multiple uncovered rasters are connected and clustered to construct multiple clusters; and region evaluation is performed according to the multiple clusters to construct the set of uncovered regions.

[0045] Furthermore, the path fusion optimization module 14 is used to perform the following method: The global connection path and the local coverage path are spatiotemporally aligned to generate path spatiotemporal registration information; the global connection path and the local coverage path are stitched together according to the spatiotemporal registration information to construct a path topology network; conflict detection is performed based on the path topology network, and the path topology network is identified based on the path conflict results to determine multiple conflicting paths; the multiple conflicting paths are locally adjusted, the path continuity of the adjusted paths is verified, and an initial full coverage path is constructed; the initial full coverage path is mapped to the dynamic environment map for multi-dimensional layered optimization to formulate the cruise path.

[0046] Furthermore, the real-time path adjustment module 15 is used to perform the following method: The system controls the robotic vacuum cleaner to execute the cruise path, activating sensors mounted on the robot to perceive the cruise environment and obtain real-time environmental perception data. This real-time environmental perception data is then mapped back to the dynamic environment map for difference analysis to obtain the degree of environmental change difference. Based on the cruise path, the robot vacuum cleaner is positioned in real-time, and the real-time positioning data is synchronized to the two-dimensional grid coordinate system to determine the real-time grid position. The real-time grid position is used as the first priority, and neighboring grids are sequentially assigned a descending priority order to construct a map update priority sequence. The cruise path is then planned and adjusted in real-time according to the degree of environmental change difference and the map update priority sequence to construct the dynamic real-time cruise path.

[0047] Furthermore, the real-time path adjustment module 15 is used to perform the following method: Based on the environmental change differences, the cruise path is analyzed for path adjustment, and a path adjustment level is set. Following the map update priority sequence, the first K grids are selected, centered on the real-time grid position, to form an initial region to be optimized, where K is an integer greater than 1. Environmental change propagation characteristics are introduced and combined with the initial region to be optimized to construct a rolling optimization window, which is used to delineate the region to be optimized. Based on the path adjustment level, local path replanning is performed on the region to be optimized, generating a candidate path set. The candidate path set is traversed to perform path quality evaluation, and the cruise optimization path is determined based on the quality evaluation results. The cruise optimization path is smoothly connected to the cruise path to construct the dynamic real-time cruise path.

[0048] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for planning the navigation path of a sweeping robot based on environmental modeling, characterized in that, The method includes: By collecting grid data of the environment through sensors mounted on the robot vacuum cleaner, a dynamic environmental map is constructed. Based on the dynamic environment map, a local area coverage path is planned according to the starting point to generate a local coverage path; When local area coverage is completed according to the local coverage path, the regional connection path is planned based on the target points of the uncovered area selected by backtracking, and a global connection path is generated. The global connection path and the local coverage path are merged to construct an initial full coverage path, which is then mapped onto a dynamic environment map for optimization, and a cruise path is determined. The robot vacuum cleaner is controlled to perform real-time perception along the navigation path. Based on the perception results, the dynamic environment map is updated in reverse to plan and adjust the navigation path in real time, thus constructing a dynamic real-time navigation path.

2. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 1, characterized in that, The method involves using sensors mounted on a robotic vacuum cleaner to collect grid data of the environment and construct a dynamic environmental map, including: By introducing working environment information, the sensors mounted on the robot vacuum cleaner are activated to sense the environment and obtain the raw environmental dataset; Grid segmentation is performed based on the task requirements information of the robotic vacuum cleaner, and the grid resolution is set. A two-dimensional grid coordinate system is constructed using the initial position of the robot vacuum cleaner as the origin and the grid resolution. The environmental space is uniformly divided based on the two-dimensional grid coordinate system to obtain multiple grid units; The original environmental dataset is mapped to the multiple raster cells for multi-source fusion calculation to obtain multiple occupancy probabilities, and the multiple occupancy probabilities correspond to the multiple raster cells; Based on the multiple occupancy probabilities and environmental features, dynamic activity values ​​are assigned to the multiple grid cells to construct the dynamic environment map.

3. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 1, characterized in that, Based on the dynamic environment map, a local area coverage path is planned according to the starting point to generate a local coverage path. The method includes: Based on the dynamic activity value of the dynamic environment map, a direction-first analysis is performed to construct a search strategy; A historical cruise log is introduced, which contains multiple historical cruise areas. The environmental space is divided into multiple local areas by taking the location of the robot vacuum cleaner as the starting point and combining the multiple historical cruise areas. The search strategy is used to perform a coverage search on the multiple local regions, generating multiple coverage search results. Based on the multiple covered search results, the search cost is calculated to determine multiple cost items; The search strategy is executed according to the multiple cost terms to generate the local coverage path.

4. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 3, characterized in that, Based on the dynamic activity value of the dynamic environment map, a direction-first analysis is performed to construct a search strategy, the method of which includes: A raster activity value distribution map is constructed based on the dynamic activity values ​​of the dynamic environment map, and the raster activity value distribution map contains activity value gradient data; An incremental analysis is performed based on the activity value gradient data to determine the initial search direction; Starting from the position of the robot vacuum cleaner, perform a multi-neighborhood search according to the initial search direction to obtain multiple neighborhood grid activity values. The search strategy is constructed by performing greedy calculations based on the activity values ​​of the multiple neighboring grids.

5. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 1, characterized in that, When local area coverage is completed according to the local coverage path, a global connection path is generated by backtracking and selecting target points in uncovered areas. The method includes: When local area coverage is completed according to the local coverage path, the distribution of grid activity values ​​is identified based on the dynamic environment map to determine the set of uncovered areas. Using the location of the robotic vacuum cleaner as the origin, the set of uncovered areas is traversed to calculate the distance and obtain multiple distance values. The multiple distance values ​​are sorted in ascending order, and the first distance value in the distance sequence is selected as the index to search the set of uncovered areas and determine the target uncovered area. Use the boundary grid points of the area not covered by the target as the target points; Based on the target point, regional connection path planning is performed to generate the global connection path.

6. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 5, characterized in that, When local area coverage is completed according to the local coverage path, the distribution of raster activity values ​​is identified based on the dynamic environment map to determine the set of uncovered areas. The method includes: When local area coverage is completed according to the local coverage path, the raster activity values ​​of multiple local areas covered by the local coverage path are extracted. Set a grid activity value threshold, and compare the grid activity value of the multiple local regions with the grid activity value threshold; The grid activity values ​​of local areas that are higher than the grid activity value threshold are identified, and multiple uncovered grids are defined based on the identification results; The multiple uncovered grid cells are connected and clustered to construct multiple clusters; Regional evaluation is performed based on the multiple clusters to construct the set of uncovered regions.

7. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 1, characterized in that, The global connection path and the local coverage path are merged to construct an initial full coverage path, which is then mapped onto the dynamic environment map for optimization to determine the cruise path. The method includes: The global connection path and the local coverage path are spatiotemporally aligned to generate path spatiotemporal registration information; Based on the spatiotemporal registration information, the global connection path and the local coverage path are stitched together to construct a path topology network; Conflict detection is performed based on the path topology network, and the path topology network is identified based on the path conflict results to determine multiple conflicting paths. The multiple conflicting paths are locally adjusted to verify the path continuity of the adjusted paths and construct an initial full-coverage path. The initial full-coverage path is mapped to the dynamic environment map for multi-dimensional layered optimization to formulate the cruise path.

8. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 2, characterized in that, The method involves controlling a robotic vacuum cleaner to perform real-time perception along its navigation path, updating the dynamic environment map based on the perception results, and adjusting the navigation path in real-time to construct a dynamic real-time navigation path. Controlling the sweeping robot to execute the cruise path activates the sensors mounted on the sweeping robot to perceive the cruise environment and obtain real-time environmental perception data. The real-time environmental perception data of the cruise is back-mapped to the dynamic environmental map for difference analysis to obtain the degree of difference in environmental changes. Based on the cruise path, the robot vacuum cleaner is located in real time, and the real-time location data is synchronized to the two-dimensional grid coordinate system to determine the real-time grid position. The real-time grid position is used as the first priority to sequentially identify the neighboring grids in descending priority order, thus constructing a map update priority sequence; The cruise path is planned and adjusted in real time according to the environmental change difference and the map update priority sequence to construct the dynamic cruise real-time path.

9. The method for planning the navigation path of a sweeping robot based on environmental modeling as described in claim 8, characterized in that, The method for constructing the dynamic real-time cruise path involves real-time planning and adjustment of the cruise path based on the environmental change difference and the map update priority sequence, including: Based on the degree of environmental change, a path adjustment analysis is performed on the cruise route, and a path adjustment level is set. According to the map update priority sequence, with the real-time grid position as the center, select the first K grids, and form an initial area to be optimized based on the first K grids, where K is an integer greater than 1; By incorporating environmental change propagation characteristics and combining them with the initial region to be optimized, a rolling optimization window is constructed, and the region to be optimized is delineated through the rolling optimization window. Based on the path adjustment level, local path replanning is performed on the path region to be optimized to generate a candidate path set. The candidate path set is traversed to perform path quality evaluation, and the cruise optimization path is determined based on the quality evaluation results. The optimized cruise path is smoothly connected to the cruise path to construct the dynamic cruise real-time path.

10. A navigation path planning system for a sweeping robot based on environmental modeling, characterized in that, The system is used to implement the environmental modeling-based navigation path planning method for a sweeping robot according to any one of claims 1-9, the system comprising: Data acquisition module: Collects grid data of the environment through sensors mounted on the robot vacuum cleaner to build a dynamic environmental map; Local path planning module: Based on the dynamic environment map, it plans local area coverage paths according to the starting point and generates local coverage paths; Global path planning module: When local area coverage is completed according to the local coverage path, the module performs regional connection path planning based on the target points of the uncovered areas selected by backtracking, and generates a global connection path; Path fusion and optimization module: merges the global connection path with the local coverage path, constructs an initial full coverage path, maps it to the dynamic environment map for optimization, and formulates the cruise path; Real-time path adjustment module: Controls the robot vacuum cleaner to perform real-time perception of the cruise path, updates the dynamic environment map based on the perception results, and adjusts the cruise path in real time to construct a dynamic real-time cruise path.

Images