A mobile robot large scene operation map real-time updating method

Abstract

This invention discloses a method for real-time updating of a large-scale mobile robot operation map, relating to the field of robot operations, including the following steps: S101: Based on robot pose and laser point cloud data, a new two-dimensional grid map of the environment is created; S102: The grid map is saved, and all sub-maps are marked as frozen sub-maps; S103: The grid map is loaded, and initial positioning is performed; S104: Active sub-maps are generated based on keyframes during robot operations; S105: Every few active sub-maps, the grids of all active sub-maps are thinned to generate a temporary global map; S106: Based on the overlap rate, non-latest active sub-maps are deleted. This invention ensures long-term consistency of positioning and reusability of historical maps, and can automatically identify and delete non-latest active sub-maps, thereby effectively suppressing the unlimited growth of map data without destroying the core map structure, so that the map size can be kept within a controllable range during long-term operation.

Description

Technical Field

[0001] This invention relates to the field of robot operation technology, specifically a method for real-time updating of large-scale operation maps for mobile robots. Background Technology

[0002] When mobile robots operate in large-scale environments, a two-dimensional grid map of the environment is typically created first to support subsequent localization and navigation. In existing technologies, the map is usually composed of multiple sub-maps, which are continuously generated during operation. However, retaining all sub-maps would lead to an explosive increase in storage and computing resources, impacting system real-time performance and stability. Therefore, how to efficiently manage and update sub-maps has become a key issue in improving robot performance.

[0003] The patent "A Method, System, Device and Medium for Dynamic Updating Robot Maps" (CN117288179B) directly traverses all grids on the submap without thinning, which is inefficient in large scenes. In addition, if a grid is observed by multiple submaps, all submaps before the grid's creation time are deleted, which will cause submaps in the mapping process to be deleted as well. Submaps in the mapping process should be fixed; otherwise, the map will be offset and misaligned after deletion. Furthermore, it is not enough to only judge one grid; the overlap rate between active submaps should be judged, and active submaps with earlier creation times should only be deleted when the overlap rate is high. Summary of the Invention

[0004] (a) Technical problems to be solved In view of the above-mentioned shortcomings of the existing technology, the present invention provides a method for real-time updating of large-scale operation maps for mobile robots, which can effectively solve the problems of the existing technology.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: This invention discloses a method for real-time updating of a large-scale operation map for mobile robots, comprising the following steps: Step S101: Based on the robot pose and laser point cloud data, create a new two-dimensional grid map of the environment. This map is composed of multiple sub-maps. The sub-maps generated during the mapping process will be marked as frozen sub-maps and cannot be deleted in subsequent operations. Step S102: Save the raster map and mark all submaps as frozen submaps; Step S103: Load the saved raster map and perform initial positioning; Step S104: During robot operation, an activity sub-graph is generated based on keyframes. Keyframes are generated by changes in robot movement distance, angle, or time interval. Step S105: At intervals of several active subgraphs (e.g., M), the raster of all active subgraphs is thinned, and a temporary global map is quickly generated using the thinned raster to meet the computational needs of large scenes. Step S106: Delete non-latest active subgraphs based on overlap rate. Based on the temporary global map, calculate the overlap rate between each active subgraph and other subgraphs, and delete non-latest active subgraphs with an overlap rate higher than a set threshold. Step S107: After the task is completed, save the frozen submap and the remaining active submap for use in the next operation. When the next task starts, the map can be loaded directly for initial positioning and continue to be updated in real time according to the above method.

[0006] Furthermore, the frozen submap in step S102 and the active submap in step S104 are allowed to overlap in physical space; in terms of logical storage and management, the active submap and the frozen submap are independent of each other, wherein the active submap is used to record environmental changes or newly added areas during the operation process, and the frozen submap serves as a stable historical map reference.

[0007] Furthermore, the triggering conditions for generating keyframes in step S104 are at least one of the following: the straight-line distance of the robot's movement exceeds a preset distance threshold; the angle change of the robot's rotation exceeds a preset angle threshold; the time interval since the last keyframe generation exceeds a preset time threshold; each keyframe contains distorted two-dimensional laser point cloud data and is associated with the robot's pose at that moment; N consecutive keyframes constitute a sub-graph, and adjacent sub-graphs share N / 2 keyframes to ensure smooth transition between sub-graphs.

[0008] Furthermore, the working logic of the thinning process in step S105 is as follows: When traversing each active subgraph, grid cells are extracted in both the width and height directions at a preset thinning interval; For each thinned raster, let its coordinates in the subgraph be (cx, cy), and the coordinates of the subgraph origin in the world coordinate system be (ox, oy). Then its world coordinates... for: ; Initialize the origin of the temporary global map Typically, this can be set to the bottom left corner of the area near the current robot's position or the area covered by all submaps, and then the coordinates of this grid in the temporary global map are calculated. : ; in The parentheses in the code indicate a function that rounds to zero, and the result may be negative. The thinned raster information is filled into the temporary global map, and the data of each raster being accessed by several active submaps is recorded.

[0009] Furthermore, the value of the thinning interval is dynamically adjusted according to the total number of active subgraphs: when the total number of active subgraphs exceeds a preset first threshold, the value of the thinning interval is increased by a preset scale to speed up the generation of temporary global maps and the speed of overlap detection; when the total number of active subgraphs is lower than a preset second threshold, the value of the thinning interval is decreased by a preset scale to improve the accuracy of overlap detection.

[0010] Furthermore, when the temporary global map receives all the thinned raster information, if there is only one active subgraph contributing thinned raster data at a certain global coordinate position, then the data is directly filled into the corresponding position of the temporary global map; if there are multiple active subgraphs contributing thinned raster data at a certain global coordinate position, then these data are fused and calculated, and the result is filled into the corresponding position of the temporary global map.

[0011] Furthermore, the working process of step S106 includes: Step S1061: Based on the temporary global map, establish a boundary region for each active subgraph that covers all its thinned raster cells; Step S1062: For any two active subgraphs A and B, if the boundary regions of A and B intersect, then calculate the overlap rate between A and B. Step S1063: Traverse all active subgraph pairs. If the overlap rate between active subgraph X and another active subgraph Y is higher than the preset overlap threshold, and X is not one of the newly generated M active subgraphs, then mark subgraph X as to be deleted. Step S1064: Delete all active subgraphs marked as to be deleted and their corresponding original non-sparsed data.

[0012] Furthermore, in step S1063, when calculating the overlap rate of active subgraphs, only the overlap between active subgraphs is considered, and the overlap between active subgraphs and frozen subgraphs is not calculated; the value range of the overlap threshold is 60% to 90%.

[0013] Furthermore, in step S1064, when deleting an active subgraph, a processing cycle is set. After generating a new active subgraph corresponding to the number of cycles, an operation of thinning, temporary global map construction, overlap rate calculation, and deletion of all existing active subgraphs is automatically triggered.

[0014] Furthermore, in step S107, the saved map data adopts a hierarchical storage structure: the first layer is the complete data of the frozen submap set; the second layer is the complete data of all active submaps retained after the end of this operation; and the global pose index of all frozen submaps and active submaps is saved for quick restoration of the global map view during the next load.

[0015] (III) Beneficial Effects Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects: By setting up a frozen submap and active submap architecture, the map data is managed throughout its lifecycle. The stable environment structure established during initialization is marked as a frozen submap, serving as an unchangeable and reliable benchmark, ensuring long-term consistency of positioning and reusability of historical maps. Newly sensed data during operation generates active submaps, representing changes in the environment or newly added areas. Through overlap rate detection based on a thinned global map, the system can automatically identify and delete non-updated active submaps. This effectively suppresses the unlimited growth of map data without disrupting the core map structure, ensuring that the map size remains within a controllable range during long-term operation.

[0016] By introducing thinning processing, interval sampling significantly reduces the amount of data involved in global computation, quickly generating low-resolution temporary global maps for overlap analysis. This enables the robot to perform map tidying more frequently, promptly remove redundant information, and always maintain lightweight system operation, freeing up sufficient computing resources for complex path planning and task decision-making. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0018] Figure 1 This is a schematic diagram of the overall method flow for real-time updating of large-scale operation maps in this invention. Figure 2 This is a schematic diagram illustrating the relationship between subgraph generation and keyframes in an embodiment of the present invention; Figure 3 This is a schematic diagram of the active subgraph raster thinning process in this invention; Figure 4 This is a schematic diagram illustrating the reduction of subgraphs based on the overlap rate of active subgraphs in an embodiment of the present invention; Figure 5 This is a schematic diagram showing the effect of the map before the update according to the present invention; Figure 6 This is a schematic diagram illustrating the effect of the updated map according to the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] The present invention will be further described below with reference to embodiments.

[0021] This embodiment provides a method for real-time updating of a mobile robot's large-scale operational map, such as... Figure 1 As shown, it includes the following steps: Step S101: Based on the robot pose and laser point cloud data, create a new two-dimensional grid map of the environment. This map is composed of multiple sub-maps. The sub-maps generated during the mapping process will be marked as frozen sub-maps and cannot be deleted in subsequent operations.

[0022] Step S102: Save the raster map and mark all submaps as frozen submaps; the frozen submaps in step S102 and the active submaps in step S104 are allowed to overlap in physical space; in terms of logical storage and management, the active submaps and frozen submaps are independent of each other, wherein the active submaps are used to record environmental changes or newly added areas during the operation process, and the frozen submaps serve as a stable historical map reference.

[0023] Step S103: Load the saved raster map and perform initial positioning.

[0024] Step S104: During robot operation, an active subgraph is generated based on keyframes. Keyframes are triggered by changes in robot movement distance, angle, or time intervals; for example... Figure 2 As shown, during the robot's movement, a key frame is generated for each preset distance, rotation angle, or elapsed time. The triggering conditions for generating a key frame in step S104 are at least one of the following: the straight-line distance moved by the robot exceeds a preset distance threshold; the angle change of the robot's rotation exceeds a preset angle threshold; the time interval since the last key frame generation exceeds a preset time threshold. In this embodiment, the preset distance threshold is 0.2 meters, the preset angle threshold is 5 degrees, and the preset time threshold is 2 seconds. Each key frame contains distortion-free two-dimensional laser point cloud data and is associated with the robot's pose at that moment. N consecutive key frames constitute a sub-image, and adjacent sub-images share N / 2 key frames to ensure smooth transition between sub-images. In this embodiment, N=90, resulting in 45 keyframes overlapping adjacent sub-maps. The point cloud data of each keyframe is transformed to the world coordinate system through its pose, and then a raster map is generated based on the sub-map origin (usually the pose of the first keyframe) and a preset resolution res (e.g., 0.05 meters / pixel). Each active sub-map is a two-dimensional raster array with width W and height H. The raster size is determined by the actual point cloud coverage area, so the size of different sub-maps may be different.

[0025] Step S105: At intervals of several active subgraphs (e.g., M), thin the raster of all active subgraphs, and use the thinned raster to quickly generate a temporary global map to meet the computational needs of large scenes; for example... Figure 3 As shown, to avoid excessive computation caused by directly traversing every cell of all sub-graphs, a thinning strategy is adopted. The working logic of the thinning process in step S105 is as follows: When traversing each active subgraph, grids are extracted in both the width and height directions at a preset thinning interval dN (in this embodiment, the preset interval dN=3), that is, every other grid is sampled once, thereby reducing the number of grids involved in subsequent calculations to 1 / dN² (1 / 9 in this case), greatly improving the processing speed. The value of the thinning interval is dynamically adjusted according to the total number of active subgraphs: when the total number of active subgraphs exceeds a preset first threshold, the value of the thinning interval is increased by a preset scale to speed up the generation of temporary global maps and the speed of overlap detection; when the total number of active subgraphs is lower than a preset second threshold, the value of the thinning interval is decreased by a preset scale to improve the accuracy of overlap detection. For each thinned raster, let its coordinates in the subgraph be (cx, cy), and the coordinates of the subgraph origin in the world coordinate system be (ox, oy). Then its world coordinates... for: ; Initialize the origin of the temporary global map Typically, this can be set to the bottom left corner of the area near the current robot's position or the area covered by all submaps, and then the coordinates of this grid in the temporary global map are calculated. : ; in The parentheses in the code indicate a function that rounds to zero, and the result may be negative. All the thinned raster information is filled into the temporary global map, and the data accessed by each raster by several active subgraphs is recorded. When the temporary global map receives all the thinned raster information, if only one active subgraph contributes thinned raster data to a certain global coordinate position, then that data is directly filled into the corresponding position of the temporary global map. If multiple active subgraphs contribute thinned raster data to a certain global coordinate position, then these data are fused and the result is filled into the corresponding position of the temporary global map. The fusion calculation prioritizes the principle of "occupied" state over "idle" state; that is, when at least one of the multiple data is in an "occupied" state, that position is marked as "occupied" in the temporary global map.

[0026] like Figure 4 As shown, the deletion logic is illustrated using three subgraphs as an example. Among them, the subgraphs... Figure 1 Frozen submaps (not deletable) generated during the mapping process. Figure 2 Kazuko Figure 3 It is an active subgraph, and the subgraph is... Figure 3 For the newly generated active subgraph, utilize the sub Figure 2 Kazuko Figure 3 After generating a temporary global map from the thinned raster, iterate through each raster on the temporary global map: For a raster covered by multiple subgraphs simultaneously (as shown by the dashed line area in the figure), first exclude the newest subgraph (sub-subgraph). Figure 3 Access records for ) . If other subgraphs (such as sub-graphs) still exist after excluding them. Figure 2 Accessing this raster indicates that the raster has duplicate coverage, and the sub-raster will be accessed. Figure 2 Mark it as pending deletion and increment its overlapping raster counter by 1.

[0027] After the traversal is complete, the number of overlapping rasters (overlap_n) for each submap to be deleted is obtained, along with the total number of rasters (submap_n) after thinning (i.e., the number of rasters involved in the calculation). Overlap rate. The calculation formula is: ; like If the overlap rate exceeds the preset threshold (0.7 in this embodiment), the subgraph is determined to be a redundant subgraph and is deleted. It is important to emphasize that this method only deletes non-latest active submaps with high overlap rates, and does not delete the latest submaps even if they are repeatedly covered, to ensure that the map can reflect the latest environmental changes. At the same time, frozen submaps are always retained to avoid map baseline shifts.

[0028] Step S106: Delete non-latest active subgraphs based on overlap rate. Based on the temporary global map, calculate the overlap rate between each active subgraph and other subgraphs, and delete non-latest active subgraphs with overlap rates exceeding a set threshold; for example... Figure 5 The image shown is the map before the update, as follows: Figure 6 The image shows the updated map. Through real-time updates, newly added obstacles, such as temporarily placed items, can be added to the map, while obstacles that existed during map creation but have disappeared during operation, such as moved boxes, are removed, thus ensuring the timeliness and accuracy of the map. The working process of step S106 includes: Step S1061: Based on the temporary global map, establish a boundary region for each active subgraph that covers all its thinned raster cells; Step S1062: For any two active subgraphs A and B, if the boundary regions of A and B intersect, calculate the overlap rate between A and B; where, overlap rate = (number of thinned grid cells in subgraph A that are both occupied in the same global coordinates as subgraph B) / (total number of thinned grid cells in subgraph A). Step S1063: Traverse all active subgraph pairs. If the overlap rate between active subgraph X and another active subgraph Y is higher than a preset overlap threshold, and X is not one of the most recently generated M active subgraphs, then mark subgraph X as to be deleted. When calculating the active subgraph overlap rate, only the overlap between active subgraphs is considered, and the overlap between active subgraphs and frozen subgraphs is not calculated. The overlap threshold ranges from 60% to 90%. Step S1064: Delete all active subgraphs marked as to be deleted and their corresponding original non-thinned data; when deleting active subgraphs, set a processing cycle, and after each generation of a new active subgraph corresponding to the number of cycles, automatically trigger a thinning, temporary global map construction, overlap rate calculation and deletion operation for all existing active subgraphs; wherein, the protection scope of the latest active subgraph is dynamically updated to several subgraphs generated in the current cycle during each batch processing.

[0029] Step S107: After the task is completed, save the frozen submap and the remaining active submap for use in the next job. When the next task starts, the map can be directly loaded for initial positioning and continue to be updated in real time according to the above method. The saved map data adopts a hierarchical storage structure: the first layer is the complete data of the frozen submap set; the second layer is the complete data of all active submaps retained after the end of this job; save the global pose index of all frozen submaps and active submaps for quick restoration of the global map view when loaded next time.

[0030] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A mobile robot large scene operation map real-time updating method, characterized in that, Includes the following steps: Step S101: Based on the robot pose and laser point cloud data, create a new two-dimensional grid map of the environment; Step S102: Save the raster map and mark all submaps as frozen submaps; Step S103: Load the raster map and perform initial positioning; Step S104: Generate an activity subgraph based on keyframes during robot operation; Step S105: At intervals of several active subgraphs, thin the raster of all active subgraphs to generate a temporary global map; Step S106: Delete non-latest active subgraphs based on the overlap rate; Step S107: Save the frozen subgraph and the remaining active subgraph for use in the next job.

2. The method of claim 1, wherein, The frozen submap in step S102 and the active submap in step S104 are allowed to overlap in physical space; in terms of logical storage and management, the active submap and the frozen submap are independent of each other, wherein the active submap is used to record environmental changes or new areas during the operation process, and the frozen submap serves as a stable historical map reference.

3. The method of claim 1, wherein, The triggering conditions for generating keyframes in step S104 are at least one of the following: the straight-line distance of the robot's movement exceeds a preset distance threshold; the angle of the robot's rotation changes beyond a preset angle threshold; the time interval since the last keyframe generation exceeds a preset time threshold; each keyframe contains distortion-free two-dimensional laser point cloud data and is associated with the robot's pose at that moment; N consecutive keyframes constitute a sub-graph, and adjacent sub-graphs share N / 2 keyframes.

4. The method for real-time updating of a large-scale mobile robot operation map according to claim 1, characterized in that, The working logic of the thinning process in step S105 is as follows: When traversing each active subgraph, grid cells are extracted in both the width and height directions at a preset thinning interval; For each thinned raster, let its coordinates in the subgraph be (cx, cy), and the coordinates of the subgraph origin in the world coordinate system be (ox, oy). Then its world coordinates... for: ； Initialize the origin of the temporary global map Set the grid to the bottom left corner of the area near the current robot position or the area covered by all submaps, and then calculate the coordinates of this grid in the temporary global map. : ； in The brackets in the text represent the zero-rounding function, which fills all the thinned raster information into the temporary global map and records the data of each raster being accessed by several active submaps.

5. The method for real-time updating of a large-scale mobile robot operation map according to claim 4, characterized in that, The thinning interval is dynamically adjusted based on the total number of active subgraphs: when the total number of active subgraphs exceeds a preset first threshold, the thinning interval is increased by a preset scale; when the total number of active subgraphs is lower than a preset second threshold, the thinning interval is decreased by a preset scale.

6. The method for real-time updating of a large-scale mobile robot operation map according to claim 4, characterized in that, When the temporary global map receives all the thinned raster information, if there is only one active subgraph contributing thinned raster data at a certain global coordinate position, then the data is directly filled into the corresponding position of the temporary global map; if there are multiple active subgraphs contributing thinned raster data at a certain global coordinate position, then these data are fused and calculated, and the result is filled into the corresponding position of the temporary global map.

7. The method for real-time updating of a large-scale mobile robot operation map according to claim 1, characterized in that, The working process of step S106 includes: Step S1061: Based on the temporary global map, establish a boundary region for each active subgraph that covers all its thinned raster cells; Step S1062: For any two active subgraphs A and B, if the boundary regions of A and B intersect, then calculate the overlap rate between A and B. Step S1063: Traverse all active subgraph pairs. If the overlap rate between active subgraph X and another active subgraph Y is higher than the preset overlap threshold, and X is not one of the newly generated M active subgraphs, then mark subgraph X as to be deleted. Step S1064: Delete all active subgraphs marked as to be deleted and their corresponding original non-sparsed data.

8. A method for real-time updating of a large-scale mobile robot operation map according to claim 7, characterized in that, In step S1063, when calculating the overlap rate of active subgraphs, only the overlap between active subgraphs is considered, and the overlap between active subgraphs and frozen subgraphs is not calculated; the value range of the overlap threshold is 60% to 90%.

9. A method for real-time updating of a large-scale mobile robot operation map according to claim 7, characterized in that, In step S1064, when deleting an active subgraph, a processing cycle is set. After generating a new active subgraph corresponding to the number of cycles, an operation of thinning, temporary global map construction, overlap rate calculation and deletion of all existing active subgraphs is automatically triggered.

10. A method for real-time updating of a large-scale mobile robot operation map according to claim 1, characterized in that, In step S107, the saved map data adopts a hierarchical storage structure: the first layer is the complete data of the frozen submap set; the second layer is the complete data of all active submaps retained after the end of this operation; and the global pose index of all frozen submaps and active submaps is saved for quick restoration of the global map view during the next load.

Images