Dynamic path planning methods for large-scale raster maps

Abstract

This invention relates to the field of path planning technology, and provides a dynamic path planning method for large-scale raster maps. It constructs a three-layer policy mask comprising a permitted normal layer, an execution-state temporary extension layer, and a service alarm layer to clearly delineate path regions with different functions. Initial path planning is performed at the permitted normal layer based on a region of interest-thick line view-through direct pull search strategy, ensuring the normal security and stability of the initial path. Detour path planning is performed at both the permitted normal layer and the execution-state temporary extension layer based on the same strategy, enabling real-time, efficient, and secure orderly planning when detour requirements are detected. Regression convergence processing is performed on candidate detour path segments based on the baseline path and the initial rendered path, further improving path security and smoothness. Corner alignment resampling processing is performed on intermediate rendered paths to effectively avoid visual artifacts.

Description

Technical Field

[0001] This invention relates to the field of path planning technology, and in particular to a dynamic path planning method for large-scale raster maps. Background Technology

[0002] Currently, for large-scale grid maps of areas such as warehouses, industrial parks, factories, campuses, and commercial complexes, robots typically need to rely on accurate path planning when performing tasks.

[0003] However, existing path planning methods still have the following drawbacks:

[0004] (1) Real-time performance and security are difficult to reconcile;

[0005] (2) Business boundaries are difficult to solidify: a single mask makes it difficult to distinguish between "initial planning" and "execution-state detour", and it is impossible to carry out fine-grained management of temporary permits and sensitive areas;

[0006] (3) Disconnect between display and execution: linear rendering of the front end often creates a visual illusion that "it looks like it is crossing a barrier";

[0007] (4) Lack of alarm management: There is a lack of hierarchical response and record keeping for areas that are business-sensitive but can be passed through;

[0008] (5) Dynamic changes and reachability: Limited computing power on the edge and frequent dynamic changes in the environment make it difficult to achieve replanning. Summary of the Invention

[0009] In view of the above, it is necessary to provide a dynamic path planning method, apparatus, device and medium for large-scale raster maps, aiming to solve the problem of poor path planning performance in large-scale raster maps.

[0010] A dynamic path planning method for large-scale raster maps, the method comprising:

[0011] In response to robot path planning instructions based on an initial grid map, environmental change data is collected, and the initial grid map is processed according to the environmental change data to obtain a three-layer policy mask, an integral map, and an obstacle departure cost map; wherein, the three-layer policy mask includes a permitted normal layer, an execution-state temporary extension layer, and a service alarm layer;

[0012] Based on the region of interest-thick line view direct pull search strategy, the initial path planning is performed on the permitted normal layer according to the integral map and the obstacle removal cost map to obtain the baseline path and the initial rendering path.

[0013] The robot's execution status based on the baseline path and real-time environmental changes are monitored in real time, and whether a detour event or alarm event is triggered based on the execution status and real-time environmental changes.

[0014] When the detour event is detected, detour path planning is performed in the permitted normal layer and the execution temporary extension layer based on the region of interest-thick line view direct pull search strategy to obtain candidate detour path segments;

[0015] Based on the baseline path and the initial rendering path, the candidate detour path segments are subjected to regression convergence processing to obtain the target execution path and intermediate rendering path.

[0016] The intermediate rendering path is resampled with corner alignment to obtain the target rendering path;

[0017] The robot is controlled to perform tasks based on the target execution path, and the target rendering path is output to the designated display terminal.

[0018] A dynamic path planning device for a large-scale raster map, the dynamic path planning device for a large-scale raster map comprising:

[0019] The processing unit is used to respond to robot path planning instructions based on an initial grid map, collect environmental change data, and process the initial grid map according to the environmental change data to obtain a three-layer policy mask, an integral map, and an obstacle departure cost map; wherein, the three-layer policy mask includes a permitted normal layer, an execution-state temporary extension layer, and a service alarm layer;

[0020] The planning unit is used to perform initial path planning on the permitted normal layer based on the region of interest-thick line view direct pull search strategy, according to the integral map and the obstacle removal cost map, to obtain the baseline path and the initial rendering path.

[0021] The detection unit is used to monitor the robot's execution status and real-time environmental changes based on the baseline path in real time, and to detect whether a detour event or an alarm event is triggered based on the execution status and the real-time environmental changes.

[0022] The planning unit is also used to, when the detour event is detected, perform detour path planning in the permitted normal layer and the execution temporary extension layer based on the region of interest-thick line view direct pull search strategy to obtain candidate detour path segments.

[0023] A convergence unit is used to perform regression convergence processing on the candidate detour path segments based on the baseline path and the initial rendering path to obtain the target execution path and intermediate rendering path.

[0024] A sampling unit is used to perform corner-aligned resampling processing on the intermediate rendering path to obtain the target rendering path;

[0025] The control unit is used to control the robot to perform tasks based on the target execution path and output the target rendering path to a designated display terminal.

[0026] A computer device, the computer device comprising:

[0027] A memory for storing at least one instruction; and a processor for executing the instructions stored in the memory to implement the dynamic path planning method for the large-scale raster map.

[0028] A computer-readable storage medium storing at least one instruction, which is executed by a processor in a computer device to implement the dynamic path planning method for the large-scale raster map.

[0029] As can be seen from the above technical solutions, this invention can construct a three-layer strategy mask including a permission normal layer, an execution-state temporary extension layer, and a service alarm layer to clearly delineate path regions for different functions; the initial path planning is performed in the permission normal layer based on the region of interest-thick line view direct pull search strategy, ensuring the normal security and stability of the initial path; the detour path planning is performed in the permission normal layer and the execution-state temporary extension layer based on the region of interest-thick line view direct pull search strategy, enabling real-time, efficient, and safe orderly planning when detour requirements are detected; regression convergence processing is performed on candidate detour path segments based on the baseline path and the initial rendering path, further improving the security and smoothness of the path; corner point alignment resampling processing is performed on the intermediate rendering path, which can also effectively avoid visual artifacts. Attached Figure Description

[0030] Figure 1 This is a flowchart of a preferred embodiment of the dynamic path planning method for large-scale grid maps of the present invention;

[0031] Figure 2 This is a functional block diagram of a preferred embodiment of the dynamic path planning device for large-scale grid maps of the present invention;

[0032] Figure 3 This is a schematic diagram of the structure of a computer device that implements a dynamic path planning method for large-scale grid maps according to a preferred embodiment of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0034] like Figure 1The diagram shown is a flowchart of a preferred embodiment of the dynamic path planning method for large-scale grid maps according to the present invention. The order of the steps in this flowchart can be changed, and some steps can be omitted, depending on different requirements.

[0035] The dynamic path planning method for large-scale grid maps is applied to one or more computer devices. The computer device is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions. Its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0036] The computer device can be any electronic product that can interact with the user, such as a personal computer, tablet computer, smartphone, personal digital assistant (PDA), game console, interactive network television (IPTV), smart wearable device, etc.

[0037] The computer equipment may also include network equipment and / or user equipment. The network equipment includes, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of hosts or network servers.

[0038] The server can be a standalone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0039] Artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0040] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0041] The network in which the computer device is located includes, but is not limited to, the Internet, wide area network, metropolitan area network, local area network, and virtual private network (VPN).

[0042] S10, in response to the robot path planning instruction based on the initial grid map, collect environmental change data, and process the initial grid map according to the environmental change data to obtain a three-layer policy mask, an integral map (SummedArea Table, SAT), and an obstacle departure cost map; wherein, the three-layer policy mask includes a permitted normal layer, an execution state temporary extension layer, and a service alarm layer.

[0043] In this embodiment, the initial raster map can be a large-scale raster map of areas such as warehouses, industrial parks, factories, campuses, and commercial complexes.

[0044] In this embodiment, the robot may include a mobile robot such as a handling robot.

[0045] In this embodiment, the environmental change data may include, but is not limited to, obstacle location data, wall and other obstruction data, etc.

[0046] In this embodiment, the permitted normal layer forms an obstacle avoidance buffer zone based on the shrinking bandwidth, which is used to perform shrinking morphological processing on the passable area; wherein, the shrinking bandwidth is equal to the sum of the robot's half-width and the positioning deviation range, and the shrinking bandwidth is dynamically adjusted with the positioning accuracy; the cost model of the permitted normal layer adopts a center preference-obstacle avoidance penalty strategy for path selection guidance, and does not contain strategy penalty;

[0047] The execution-state temporary extension layer is used to perform outward morphological processing on the temporary permitted area; wherein, the outward bandwidth is less than the configured threshold; the activation constraints of the execution-state temporary extension layer include time window constraints, budget constraints, no prohibited conflict constraints, and concurrent capacity constraints; the policy cost of the execution-state temporary extension layer includes: additional penalties to guide the robot to prioritize the permitted normal layer; the revocation conditions of the execution-state temporary extension layer include: reaching the earliest return point, budget and time window exhaustion, manual revocation, and policy withdrawal; the execution-state temporary extension layer includes a cumulative cost plus time growth term to encourage the robot to return to the permitted normal layer as soon as possible;

[0048] The business alarm layer includes physically accessible but business-sensitive or authorized areas; wherein, when the robot enters the business alarm layer, a tiered alarm is triggered.

[0049] The obstacle departure cost map is constructed based on a distance field to guide the robot to prioritize paths away from obstacles; wherein, the distance field is calculated based on the Euclidean Distance Transform (EDT) algorithm.

[0050] The positioning deviation range can be 3σ positioning, that is, the inner bandwidth = robot half width + 3σ positioning, and σ can be configured according to actual needs.

[0051] The obstacle avoidance buffer zone can reduce the occurrence of contact with walls and corners.

[0052] The center preference-obstacle avoidance penalty strategy is used in a passable area to encourage the robot's path to be as close as possible to the center of the area. In order to ensure that the robot maintains a safe distance from obstacles during movement and avoids collisions, an additional penalty is imposed on the path that is close to the obstacle.

[0053] The reduced bandwidth can be dynamically adjusted to accommodate changes in accuracy caused by switching between day and night scenarios.

[0054] The execution-state temporary extension layer is only activated during execution-state local detours and does not participate in the initial planning under normal circumstances.

[0055] The fact that the external bandwidth is less than the configured threshold ensures that the expansion does not weaken the security boundary.

[0056] The time window constraint is used to limit the time window; the budget constraint is used to limit the maximum values ​​of indicators such as distance, time, and number of times; the no-no-traffic-conflict constraint is used to ensure that the execution-state temporary extension layer does not conflict with the business alarm layer, i.e., other no-traffic-conflict rules; and the concurrency capacity constraint is used to limit the maximum number of robots allowed to pass through narrow passages and temporary detour areas at the same time.

[0057] The additional penalty is greater than 0, and the additional penalty can be adaptively adjusted according to type, priority, and time to guide the priority selection of the permission normalization layer.

[0058] The business alarm layer is not included in the planning by default.

[0059] Within the aforementioned business alarm layer, unauthorized access is considered prohibited, while authorized or emergency permission is considered controlled access and triggers an alarm upon entry.

[0060] The integral image can achieve O(1) query of sub-window pixels, which facilitates the rapid calculation of the proportion of safe pixels within the region.

[0061] Of course, a distance field can also be used instead of the integral graph for subsequent calculations, which will not be elaborated here.

[0062] Through the above embodiments, path areas with different functions can be clearly defined. The normal permission layer ensures regular safe passage, the execution-state temporary extension layer meets temporary detour needs, and the business alarm layer realizes sensitive area control, providing a clear strategic basis for subsequent path planning.

[0063] Furthermore, this embodiment supports dynamic adjustment of parameters such as the bandwidth within the permitted normal layer, enabling the map preprocessing results to adapt to different environments and changes in positioning accuracy, thereby improving the flexibility and adaptability of planning.

[0064] S11, based on the Region of Interest (ROI) - thick line view direct pull search strategy, perform initial path planning on the permitted normal layer according to the integral map and the obstacle removal cost map to obtain the baseline path and the initial rendering path.

[0065] In this embodiment, the initial path planning based on the region of interest-thick line view direct pull search strategy, according to the integral map and the obstacle departure cost map, in the permitted normal layer, to obtain the baseline path and the initial rendering path includes:

[0066] Using the permitted normal layer as the basic range constraint, obstacle blocking data and positioning error are generated based on the obstacle departure cost map;

[0067] Obtain the queue status within the three-layer policy mask;

[0068] A first region of interest is generated based on the obstacle blocking data, the positioning error, and the queue situation, and multiple potential paths are searched within the first region of interest.

[0069] Sampling is performed along the multiple potential paths according to a preset step size to obtain multiple sampling points;

[0070] For the neighborhood window of each sampling point, the thick line view verification algorithm is used to calculate the safe pixel ratio based on the integral image, and the sampling point is determined to be safe when the safe pixel ratio is greater than a preset threshold.

[0071] Obtain potential paths that are safe at all sampling points as candidate paths;

[0072] The candidate path is subjected to path point safety simplification to obtain the baseline path and the initial rendering path.

[0073] The queue status is used to reflect the queuing situation of the robots, etc.

[0074] The preset step size can be customized, such as 0.5-1.0px.

[0075] The preset threshold can also be customized, such as 0.80 or 0.95.

[0076] In cases where the neighboring window extends beyond the raster map's boundaries (e.g., sampling points at the map edge where half the window is inside and half outside the map), directly counting the "out-of-bounds raster" would lead to incorrect safety decisions due to "no data." Therefore, specialized raster boundary handling rules are needed. Specifically, normalization can be performed using the effective pixel count.

[0077] For sub-pixel or jagged edges, a 2x oversampling can be performed to suppress false jagged edge detection.

[0078] If a distance field is used instead of the integral map for initial path planning, when performing safety checks on sampling points, it is necessary to directly query whether the corresponding EDT value is greater than the sum of the preset radius and the check value after sampling. When the EDT values ​​of all sampling points are greater than the sum of the preset radius and the check value, it is considered safe.

[0079] Accordingly, for cases where the neighboring window extends beyond the raster map, if a distance field is used, the out-of-bounds distance is considered to be 0; for sub-pixel or jagged edges, bilinear interpolation is enabled.

[0080] If an unsafe segment is detected, the process either reverts to the previous safe sampling point or performs a segmented binary search until the longest safe segment is found.

[0081] Among them, the search algorithm for potential paths can be flexibly selected to improve applicability.

[0082] In this embodiment, the process of performing path point security simplification on the candidate path includes:

[0083] From the candidate paths, obtain the corners that are less than the included angle threshold as true corners, and retain the sampling points corresponding to the true corners as true corner points;

[0084] Further simplification is prohibited within the preset angle after the corner;

[0085] Merge the sample points that are collinear in the candidate paths;

[0086] Identify spike segments in the candidate paths and merge the spike segments into adjacent steady-state segments;

[0087] Eliminate the closest points to the start and end points in the candidate paths;

[0088] Among them, all sampling points must remain safe when performing path point safety simplification processing;

[0089] In particular, after performing path point safety simplification, the turning angle of adjacent line segments is matched with the rotatable radius of the underlying controller.

[0090] The included angle threshold can be configured to be between 25° and 35°. Retaining the sampling points corresponding to the true corners can force the retention of the actual turning direction, ensuring the robot's turning safety.

[0091] The rule prohibits further simplification within a preset angle after a corner, which prevents repeated backtracking and ensures a smooth path transition.

[0092] Merging collinear sampling points in the candidate paths can effectively simplify the path.

[0093] Merging the spiked line segments into adjacent steady-state line segments can remove line segments with characteristics such as large folding amplitude and short arms, thus avoiding unnecessary turning and shaking of the robot.

[0094] Removing the closest points to the start and end points in the candidate paths is beneficial for robot start-up and path convergence.

[0095] In order to keep the rotation angle of adjacent line segments matched with the rotatable radius of the underlying controller, short arcs or transition points can be inserted when necessary.

[0096] Through the above embodiments, planning is only carried out at the permitted normal level, which ensures the normal safety and stability of the initial path and meets the basic requirements of daily patrols; the thick line vision verification algorithm ensures the safety of the path segment level and avoids safety hazards such as corner cutting, edge rubbing, and passing through narrow gaps; the path point safety simplification simplifies the path without sacrificing safety, and improves the robot's execution efficiency and stability.

[0097] S12, monitor the robot's execution status and real-time environmental changes based on the baseline path in real time, and detect whether a detour event or alarm event is triggered based on the execution status and the real-time environmental changes.

[0098] In this embodiment, detecting whether a detour event or alarm event is triggered based on the execution state and the real-time environmental changes includes:

[0099] The robot's real-time positioning data is obtained based on the execution status, and based on the real-time positioning data, the real-time environmental changes, and the baseline path, it is detected whether there are obstacles blocking the current execution path, and the blocking detection result is obtained.

[0100] Based on the real-time location data and the real-time environmental changes, the queue-yield detection result is obtained to determine whether there are queuing events and yielding events.

[0101] The deviation between the robot's actual position and the baseline path is calculated based on the real-time positioning data and the baseline path to obtain the positioning deviation detection result;

[0102] Real-time detection of task change information;

[0103] Based on the execution status and the real-time environmental changes, predict whether the alarm event will occur, and obtain the alarm event prediction result;

[0104] Whether to trigger the detour event is determined based on the blocking detection result, the queue-giveaway detection result, the positioning deviation detection result, and the task change information; and whether to trigger the alarm event is determined based on the alarm event prediction result.

[0105] Specifically, when an increase in occupancy probability is detected, a new obstacle is found through static graph differential analysis, or a vehicle or pedestrian is predicted to cut in, it can be determined that an obstacle is blocking the current execution path.

[0106] Specifically, when a concurrent capacity limit is hit or a queuing delay exceeds the threshold, it is determined that a queuing event and a yielding event exist.

[0107] The positioning deviation detection results may include tracking error, angle with the baseline path, lateral drift, etc.

[0108] The task change information may include task priority switching, emergency notices, and time windows that are about to expire.

[0109] The alarm event prediction results can be used to reflect whether the robot will enter the business alarm layer area during subsequent execution, so as to make preparations in advance.

[0110] Through the above embodiments, real-time monitoring can ensure the timely detection of various abnormal situations during the execution process, avoiding safety accidents or task failures caused by changes in the environment or abnormalities in the robot's own state.

[0111] S13, when the detour event is detected, detour path planning is performed on the permitted normal layer and the execution temporary extension layer based on the region of interest-thick line view direct pull search strategy to obtain candidate detour path segments.

[0112] In this embodiment, the detour path planning based on the region of interest-thick line view direct pull search strategy is performed in the permitted normal layer and the execution state temporary extension layer, and the candidate detour path segments obtained include:

[0113] Obtain the robot's current pose, obstacle envelope, and time step;

[0114] Predict the robot's position after the time step;

[0115] Determine the minimum bounded bounding box based on the current pose, the obstacle envelope, and the predicted position;

[0116] The minimum bounded bounding box is expanded by a preset multiple to obtain the second region of interest;

[0117] Detect whether the execution-state temporary extension layer satisfies the enable constraint;

[0118] When the execution-state temporary extension layer satisfies the enable constraint, the union of the permitted normal layer and the execution-state temporary extension layer in the second region of interest is determined as a passable region.

[0119] Based on the strategy cost, a path search is performed within the passable area to obtain multiple candidate sub-segments;

[0120] The thick line visibility verification algorithm is used to perform line segment-level security detection on the multiple candidate sub-segments;

[0121] Candidate sub-segments that pass the line segment-level safety detection are determined as candidate detour path sub-segments.

[0122] The preset multiplier can be configured according to actual needs, such as 1.2-1.5 times.

[0123] Specifically, when detecting whether the execution-state temporary extension layer meets the activation constraints, if the number of occupied areas in the current execution-state temporary extension layer region is greater than the upper limit of the concurrent capacity of the execution-state temporary extension layer, an application can be made to add the detouring robot to the queue and schedule it according to the principle of "priority first → first in first out of the same level → anti-starvation for timeout". Shortest jobs are prioritized to break up parallel queues, and the estimated entry time is estimated periodically. If the estimated entry time is greater than the service commitment threshold or the budget will be exceeded, the robot is guided to change course or abandon the use of the execution-state temporary extension layer.

[0124] When performing path search within the passable area, a search algorithm compatible with the initial planning can be used. During the search process, the additional penalty of the execution-state temporary extension layer must be considered, and the guided path should minimize its use within the execution-state temporary extension layer.

[0125] Through the above embodiments, local region of interest replanning avoids global recalculation. Combined with efficient search algorithms, it ensures the real-time performance of detour planning and meets the needs of scenarios with limited computing power at the lower end of large-scale raster maps. The multi-dimensional constraint verification enabled by the execution-state temporary extension layer ensures the compliance and security of detours, avoiding chaos and risks caused by disordered detours. The concurrent capacity arbitration mechanism balances efficiency and fairness, reducing congestion and deadlocks between vehicles. The union range search and thick line view verification, while expanding the possibilities of detour paths using the execution-state temporary extension layer, also ensure the security of detour paths and avoid introducing new security risks due to detours.

[0126] S14, perform regression convergence processing on the candidate detour path segments based on the baseline path and the initial rendering path to obtain the target execution path and intermediate rendering path.

[0127] In this embodiment, the step of performing regression convergence processing on the candidate detour path segments based on the baseline path and the initial rendering path to obtain the target execution path and intermediate rendering path includes:

[0128] Multiple candidate regression points are selected one by one along the potential splicing area between the candidate bypass path segment and the baseline path;

[0129] The thick line vision verification algorithm is used to perform security checks on the line segments from the candidate detour path sub-segments to each candidate regression point and the line segments from each candidate regression point to the subsequent segments of the baseline path, to obtain the thick line vision security verification results.

[0130] Calculate the total path cost after concatenating the multiple candidate regression points, and calculate the initial path cost of the corresponding path in the baseline path; compare the total path cost with the initial path cost to obtain the cost superiority verification result;

[0131] Calculate the alignment error between the candidate detour path segment and the baseline path at each candidate regression point to obtain the geometric alignment verification result;

[0132] When the thick line visibility security verification result corresponding to any candidate regression point is that the line segments from the candidate detour path sub-segment to each candidate regression point and the line segments from each candidate regression point to the subsequent segments of the baseline path all pass the security detection, the corresponding cost superiority verification result is that the total path cost is less than the difference between the initial path cost and the jitter value, and the corresponding geometric alignment verification result is that the alignment error is less than the error threshold, it is determined that the arbitrary candidate regression point meets the security conditions, and the detection continues to check whether a consecutive preset number of candidate regression points all meet the security conditions;

[0133] When the predetermined number of consecutive candidate regression points all meet the safety conditions, any candidate regression point is determined as a valid regression point.

[0134] Select the point closest to the starting point of the detour from all valid regression points as the earliest regression point;

[0135] The candidate detour path segment from the corresponding starting point to the earliest regression point is concatenated with the segment from the earliest regression point to the end point of the baseline path to obtain the target execution path;

[0136] Close the execution-state temporary extension layer and clear the execution data of the execution-state temporary extension layer;

[0137] The target execution path is rendered to obtain the intermediate rendering path.

[0138] When selecting multiple candidate regression points one by one along the potential splicing area between the candidate detour path segment and the baseline path, the candidate regression points must be located in the intersection or adjacent area of ​​the candidate detour path segment and the baseline path, and the path continuity from the candidate detour path segment to the candidate regression point and from the candidate regression point to the subsequent segment of the baseline path must be guaranteed.

[0139] The thick line visibility verification algorithm ensures that both line segments meet safety requirements and are free from issues such as cut corners or edge rubbing.

[0140] The jitter value is used for anti-shake purposes to avoid misjudgment due to minor fluctuations.

[0141] Among them, cost superiority verification can ensure that the regressed path is more cost-effective or not inferior to the original path.

[0142] The alignment error includes positional deviation, angular deviation, etc.

[0143] Among them, geometric alignment verification can ensure a smooth transition of the spliced ​​path and avoid large-scale turning or shaking of the robot.

[0144] The preset number can be 2-3. A valid regression point is determined only after 2-3 consecutive candidate regression points are deemed safe, which avoids misjudgments caused by instantaneous data fluctuations and ensures the stability of the regression.

[0145] If a candidate regression point is still in the execution-state temporary extension layer area and the budget or time window is about to run out, a regression point that is earlier and meets the budget or time window requirements should be selected first to ensure successful regression to the license normal layer before resources are exhausted.

[0146] Among them, determining the earliest regression point enables the fastest possible return to the baseline path, reducing the dwell time in the execution-state temporary extension layer.

[0147] The execution data may include time increments, budget accumulation, etc., and includes policy penalties and arbitration related to the temporary extension layer of the execution state. After completing path splicing and closing the temporary extension layer of the execution state, the robot resumes execution according to the baseline path and returns to the execution monitoring phase.

[0148] In the above embodiments, the multi-criteria fusion verification of the earliest regression point ensures the safety, cost advantage, and smoothness of the regression path, avoiding safety hazards or path abrupt changes after regression that could lead to robot instability. Closing the execution-state temporary extension layer as soon as possible can reduce the robot's dwell time in the execution-state temporary extension layer, reduce the increased cost caused by the additional penalties of the execution-state temporary extension layer, avoid long-term occupation of the execution-state temporary extension layer resources, improve the utilization efficiency of the execution-state temporary extension layer, and provide resources for other robots that need to detour.

[0149] S15, perform corner alignment resampling on the intermediate rendering path to obtain the target rendering path.

[0150] In this embodiment, the step of performing corner alignment resampling on the intermediate rendering path to obtain the target rendering path includes:

[0151] The intermediate rendering path is phase-reset segment by segment to obtain multiple sampling segments; wherein, when performing phase-reset segment by segment, all original polyline vertices are retained;

[0152] For each sampling segment, the previous sampling point is designated as the first sampling point, and the next sampling point is calculated along the sampling segment, starting from the first sampling point, according to the sampling step size. It is then checked whether the next sampling point enters the buffer at the end of the sampling segment. When the next sampling point enters the buffer, the end point of the sampling segment is designated as the target point. When the next sampling point does not enter the buffer, the thick line visibility check algorithm is used to perform a security check on the line segment formed by the first sampling point and the next sampling point. If the security check is passed, the next sampling point is designated as the target point, and the calculation of the next sampling point continues until all sampling segments are processed. Wherein, when the line segment formed by the first sampling point and the next sampling point fails the security check, the end point of the sampling segment is inserted as a bridging point into the sampling segment, the first sampling point is updated as the bridging point, and the processing of the next sampling segment formed by the bridging point continues.

[0153] The target rendering path is generated based on the original polyline vertices, the target point, and the bridging point.

[0154] In this process, all original polyline vertices are retained, and excessive smoothing is not applied on the rendering side to ensure that the corner features of the path are accurately presented, making it easier for users to identify semantic information such as path inflection points, deceleration points, and prompt points through a visual interface.

[0155] The target rendering path is used for monitoring.

[0156] Through the above embodiments, segment-by-segment phase reset and forced conformal mapping ensure that the displayed path accurately reflects the geometric features and semantic information of the original path, avoiding path distortion caused by excessive smoothing, and making it easier for users to intuitively understand the robot's actual driving trajectory and key nodes. The thick-line vision verification fallback mechanism completely eliminates the visual illusion of "display lines passing through obstacles," ensuring that "what you see is what you do," making the path observed by the user through the visualization interface strictly consistent with the path actually executed by the robot in terms of safety, and improving the user's trust in path planning. The segment-end buffer setting ensures that the sampling point accurately falls at the vertex at the end of the line segment, avoiding the offset of the displayed path at the vertex due to sampling errors, and further improving the accuracy and aesthetics of the displayed path.

[0157] S16, control the robot to perform tasks based on the target execution path, and output the target rendering path to the designated display terminal.

[0158] The designated display terminal can be a robot control terminal, etc.

[0159] In this embodiment, after the detour path planning is performed in the permitted normal layer and the execution-state temporary extension layer based on the region of interest-thick line view direct pull search strategy, the method further includes:

[0160] When detour path planning fails, the preset multiplier is incrementally increased to expand the second region of interest until the preset multiplier reaches its upper limit. Within the expanded second region of interest, detour path planning continues based on the region of interest-thick line view direct pull search strategy; and / or

[0161] Adjust the parameters of the strategy cost; and / or

[0162] Adjust the neighborhood window; and / or

[0163] Obtain the detour path planning record, report the detour path planning record to the designated operation and maintenance terminal, and guide the robot to a safe waiting position.

[0164] Among these, it can be anisotropically amplified along the permitted normal layer framework, prioritizing expansion towards passable areas and avoiding meaningless expansion of obstacle areas.

[0165] Specifically, parameters such as obstacle departure cost weight and smoothness weight can be reduced by a certain percentage (e.g., 15%) to lower the requirements for obstacle departure distance and smoothness, making it easier for the search algorithm to find feasible paths. Additionally, the size of the penalty for the execution-state temporary extension layer can be adjusted appropriately.

[0166] This allows for switching from the original domain restrictions to more lenient neighborhood restrictions, increasing the robot's choice of movement direction and expanding the path search space.

[0167] The detour route planning record may include detailed information on replanning failures, such as relaxed constraints at each stage, search results, map data snapshots, and strategy parameters.

[0168] The robot waits for human instructions or strategy updates at a safe waiting position; if it receives manually adjusted map data or strategy parameters (such as adding a temporary extended layer area in the execution state, adjusting the budget, etc.), it re-enters the detour triggering phase to attempt replanning; if it receives a mission termination instruction, it ends the current patrol mission.

[0169] In the above embodiments, the constraints are relaxed in stages and in an orderly manner (ROI → cost → neighborhood), which avoids a significant decrease in path safety caused by indiscriminate relaxation. Under the premise of ensuring safety, the path reachability is maximized and the task failure caused by local obstacles is reduced. After each relaxation, a thick line of sight verification is performed to ensure that the found path still meets the basic safety requirements even when the constraints are relaxed, and to avoid introducing new safety hazards. The reporting and safety waiting mechanism after failure ensures that the robot is in a safe state when it cannot find a path autonomously, while providing complete data support for human intervention, which facilitates the rapid location and resolution of problems and ensures the overall continuity of the patrol mission.

[0170] In this embodiment, after detecting whether a detour event or alarm event is triggered based on the execution state and the real-time environmental changes, the method further includes:

[0171] When the alarm event is detected, a graded response is performed according to the execution status and the real-time environmental changes, and the robot's execution data in the business alarm layer is monitored and recorded in real time.

[0172] When the robot is detected to have left the business alarm layer, it is checked whether the robot meets the regression conditions;

[0173] When the robot meets the regression condition, it stops performing hierarchical responses and controls the robot to return to the normal state within the baseline path.

[0174] For example, when the robot's current location reaches the warning threshold between itself and the business alarm layer area, or when the task planning determines that the robot is about to enter the business alarm layer, an alarm event warning can be triggered. Alternatively, the robot's authorization information can be queried to determine whether it has been authorized to enter the business alarm layer and the authorization level; if it has not been authorized or the authorization has expired, responses such as forced stop or rerouting are executed, and the event is immediately reported; if it has been authorized, the corresponding graded response measures are determined based on the authorization level and the sensitivity level of the business alarm layer.

[0175] The graded response may include:

[0176] Level L1 response (prompt): Applicable to situations with low authorization levels and low sensitivity of business alarm layers. It alerts operators that the robot is about to enter or has already entered the business alarm layer area through user interface display of prompt information, voice broadcast prompts, flashing lights (such as yellow lights), etc., and records the response start time.

[0177] Level 2 response (speed limit / beep): Suitable for situations with medium authorization level and high sensitivity of business alarm layer. In addition to implementing Level 1 prompt measures, the robot speed limit should be lowered to a preset value (e.g., 0.6m / s, which can be configured according to the requirements of business alarm layer), and the beeping device (e.g., buzzer) should be activated to warn surrounding personnel; at the same time, the derived cost of the business alarm layer area should be increased in the cost function to guide the robot to leave the business alarm layer as soon as possible; the speed limit value, beeping start time and other information should be recorded.

[0178] Level 3 Response (Forced Stop / Rerouting): Applicable to situations where the robot has entered the business alarm layer without authorization, the authorization has expired, or the business alarm layer is highly sensitive (such as involving core areas or dangerous areas). It immediately forces the robot to stop (if it has already entered the business alarm layer), or guides the robot to deviate from its original path and reroute around the business alarm layer area (if it is about to enter the business alarm layer); at the same time, it generates emergency alarm information, reports it to the management system, and notifies relevant personnel to handle it; and records information such as the forced stop location, rerouting path, and alarm time.

[0179] The execution data may include the robot's position, speed, and changes in the surrounding environment within the business alarm layer to ensure the effective execution of response measures. If an abnormal situation occurs (such as speed exceeding the limit or deviation from the designated path), the response measures will be adjusted in a timely manner (such as forcing the robot to stop again or enhancing the prompts).

[0180] The execution data may also include key information of alarm events recorded in real time, such as the time of entering the business alarm layer, the time of leaving the business alarm layer, the business alarm layer identifier, the business alarm layer penalty coefficient, the speed limit value, authorization information, and the execution status of response measures. At the same time, policy signatures are generated based on parameters such as budget snapshots and time windows to ensure that the information is auditable and traceable.

[0181] Specifically, it can detect whether the robot's position after leaving the business alarm layer is within a reasonable range of the baseline path, whether its speed has returned to normal, and whether the surrounding environment meets the conditions for continuing to execute the original task. If the regression conditions are met, the tiered response measures are stopped, and the robot's normal speed and execution status are restored. At the same time, all recorded information of the alarm event can be archived and stored, the policy signature can be updated, and the robot can return to the execution monitoring phase to continue to perform patrol tasks according to the baseline path.

[0182] In the above embodiments, the tiered response mechanism differentiates its handling based on the sensitivity and authorization status of the business alarm layer. While ensuring the security and compliance of the business alarm layer area, it avoids excessive control that could affect task execution efficiency, achieving flexible yet rigorous control over sensitive areas. Full-process information recording and policy signature archiving provide complete and traceable data support for subsequent audits of alarm events, meeting business compliance and security audit requirements and facilitating responsibility identification and problem investigation. Real-time status monitoring and anomaly adjustments ensure the effective implementation of tiered response measures, preventing safety incidents or violations by robots in the business alarm layer area, and protecting both business security and the robot's own safety.

[0183] In this embodiment, relevant parameters and execution records of the entire path planning and execution process can also be recorded, and signatures and comprehensive audit reports can be generated to form a governance closed loop, facilitating subsequent traceability. Comprehensive budget, time window, and concurrent capacity control ensure that the path planning and execution process complies with business resource constraints and time requirements, avoiding resource waste, inefficiency caused by disordered use, or conflicts, and improving the precision of business management; the alarm event audit and policy signature mechanism ensures the compliance and data security of sensitive area control, realizes event traceability and responsibility identification, and meets the needs of industry supervision and internal enterprise security management; the comprehensive audit report provides managers with intuitive and comprehensive business operation data, facilitating the analysis of business bottlenecks, optimization of policy parameters (such as adjusting budget, time window, and concurrent capacity), and improvement of path planning algorithms, promoting the continuous optimization and upgrading of the entire patrol system.

[0184] As can be seen from the above technical solutions, this invention can construct a three-layer strategy mask including a permission normal layer, an execution-state temporary extension layer, and a service alarm layer to clearly delineate path regions for different functions; the initial path planning is performed in the permission normal layer based on the region of interest-thick line view direct pull search strategy, ensuring the normal security and stability of the initial path; the detour path planning is performed in the permission normal layer and the execution-state temporary extension layer based on the region of interest-thick line view direct pull search strategy, enabling real-time, efficient, and safe orderly planning when detour requirements are detected; regression convergence processing is performed on candidate detour path segments based on the baseline path and the initial rendering path, further improving the security and smoothness of the path; corner point alignment resampling processing is performed on the intermediate rendering path, which can also effectively avoid visual artifacts.

[0185] like Figure 2 The diagram shown is a functional block diagram of a preferred embodiment of the dynamic path planning device for large-scale raster maps of the present invention. The dynamic path planning device 11 for large-scale raster maps includes a processing unit 110, a planning unit 111, a detection unit 112, a convergence unit 113, a sampling unit 114, and a control unit 115. The module / unit referred to in this invention is a series of computer program segments that can be executed by a processor and perform a fixed function, and which are stored in memory. In this embodiment, the functions of each module / unit will be described in detail in subsequent embodiments.

[0186] The processing unit 110 is used to respond to robot path planning instructions based on an initial grid map, collect environmental change data, and process the initial grid map according to the environmental change data to obtain a three-layer policy mask, an integral map, and an obstacle departure cost map; wherein, the three-layer policy mask includes a permitted normal layer, an execution state temporary extension layer, and a service alarm layer.

[0187] The planning unit 111 is used to perform initial path planning on the permitted normal layer based on the region of interest-thick line view direct pull search strategy, according to the integral map and the obstacle removal cost map, to obtain the baseline path and the initial rendering path.

[0188] The detection unit 112 is used to monitor the robot's execution status and real-time environmental changes based on the baseline path in real time, and to detect whether a detour event or an alarm event is triggered based on the execution status and the real-time environmental changes.

[0189] The planning unit 111 is also used to, when the detour event is detected, perform detour path planning in the permitted normal layer and the execution temporary extension layer based on the region of interest-thick line view direct pull search strategy to obtain candidate detour path segments.

[0190] The convergence unit 113 is used to perform regression convergence processing on the candidate detour path segments according to the baseline path and the initial rendering path to obtain the target execution path and intermediate rendering path.

[0191] The sampling unit 114 is used to perform corner alignment resampling processing on the intermediate rendering path to obtain the target rendering path;

[0192] The control unit 115 is used to control the robot to perform tasks based on the target execution path and output the target rendering path to a designated display terminal.

[0193] As can be seen from the above technical solutions, this invention can construct a three-layer strategy mask including a permission normal layer, an execution-state temporary extension layer, and a service alarm layer to clearly delineate path regions for different functions; the initial path planning is performed in the permission normal layer based on the region of interest-thick line view direct pull search strategy, ensuring the normal security and stability of the initial path; the detour path planning is performed in the permission normal layer and the execution-state temporary extension layer based on the region of interest-thick line view direct pull search strategy, enabling real-time, efficient, and safe orderly planning when detour requirements are detected; regression convergence processing is performed on candidate detour path segments based on the baseline path and the initial rendering path, further improving the security and smoothness of the path; corner point alignment resampling processing is performed on the intermediate rendering path, which can also effectively avoid visual artifacts.

[0194] like Figure 3 The diagram shown is a schematic representation of the structure of a computer device for implementing a dynamic path planning method for large-scale grid maps according to a preferred embodiment of the present invention.

[0195] The computer device 1 may include a memory 12, a processor 13, and a bus (the arrow in the figure represents the bus), and may also include a computer program stored in the memory 12 and executable on the processor 13, such as a dynamic path planning program for a large-scale grid map.

[0196] Those skilled in the art will understand that the schematic diagram is merely an example of computer device 1 and does not constitute a limitation on computer device 1. Computer device 1 can be either a bus topology or a star topology. Computer device 1 may also include more or fewer other hardware or software than shown in the diagram, or different component arrangements. For example, computer device 1 may also include input / output devices, network access devices, etc.

[0197] It should be noted that the computer device 1 described is merely an example. Other existing or future electronic products that are adaptable to this invention should also be included within the scope of protection of this invention and are incorporated herein by reference.

[0198] The memory 12 includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the computer device 1, such as a portable hard drive of the computer device 1. In other embodiments, the memory 12 can be an external storage device of the computer device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the computer device 1. Furthermore, the memory 12 can include both internal and external storage units of the computer device 1. The memory 12 can be used not only to store application software and various types of data installed on the computer device 1, such as the code of a dynamic path planning program for a large-scale raster map, but also to temporarily store data that has been output or will be output.

[0199] In some embodiments, the processor 13 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits packaged with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 13 is the control unit of the computer device 1, connecting various components of the computer device 1 via various interfaces and lines. It executes programs or modules stored in the memory 12 (e.g., dynamic path planning programs for large-scale grid maps) and calls data stored in the memory 12 to perform various functions of the computer device 1 and process data.

[0200] The processor 13 executes the operating system of the computer device 1 and various installed applications. The processor 13 executes these applications to implement the steps in the above embodiments of the dynamic path planning method for large-scale raster maps, for example... Figure 1 The steps are shown.

[0201] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 12 and executed by the processor 13 to complete the present invention. The one or more modules / units may be a series of computer-readable instruction segments capable of performing specific functions, which describe the execution process of the computer program in the computer device 1. For example, the computer program may be divided into a processing unit 110, a planning unit 111, a detection unit 112, a convergence unit 113, a sampling unit 114, and a control unit 115.

[0202] The integrated unit implemented as a software functional module described above can be stored in a computer-readable storage medium. This software functional module, stored in a storage medium, includes several instructions to cause a computer device (which may be a personal computer, computer equipment, or network device, etc.) or processor to execute the dynamic path planning method for large-scale raster maps described in the various embodiments of this invention.

[0203] If the modules / units integrated in the computer device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware devices. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above.

[0204] The computer program includes computer program code, which may be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory, etc.

[0205] Furthermore, the computer-readable storage medium may primarily include a stored program area and a stored data area, wherein the stored program area may store the operating system, an application program required for at least one function, etc.; and the stored data area may store data created based on the use of blockchain nodes, etc.

[0206] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0207] The bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, in... Figure 3 The bus is represented by only one straight line, but this does not mean that there is only one bus or one type of bus. The bus is configured to enable communication between the memory 12 and at least one processor 13, etc.

[0208] Although not shown, the computer device 1 may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to the at least one processor 13 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The computer device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0209] Furthermore, the computer device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish communication connections between the computer device 1 and other computer devices.

[0210] Optionally, the computer device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the computer device 1 and to display a visual user interface.

[0211] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0212] It will be understood by those skilled in the art that Figure 3 The structure shown does not constitute a limitation on the computer device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0213] Combination Figure 1 The memory 12 in the computer device 1 stores multiple instructions to implement a dynamic path planning method for a large-scale raster map, and the processor 13 can execute the multiple instructions to achieve:

[0214] In response to robot path planning instructions based on an initial grid map, environmental change data is collected, and the initial grid map is processed according to the environmental change data to obtain a three-layer policy mask, an integral map, and an obstacle departure cost map; wherein, the three-layer policy mask includes a permitted normal layer, an execution-state temporary extension layer, and a service alarm layer;

[0215] Based on the region of interest-thick line view direct pull search strategy, the initial path planning is performed on the permitted normal layer according to the integral map and the obstacle removal cost map to obtain the baseline path and the initial rendering path.

[0216] The robot's execution status based on the baseline path and real-time environmental changes are monitored in real time, and whether a detour event or alarm event is triggered based on the execution status and real-time environmental changes.

[0217] When the detour event is detected, detour path planning is performed in the permitted normal layer and the execution temporary extension layer based on the region of interest-thick line view direct pull search strategy to obtain candidate detour path segments;

[0218] Based on the baseline path and the initial rendering path, the candidate detour path segments are subjected to regression convergence processing to obtain the target execution path and intermediate rendering path.

[0219] The intermediate rendering path is resampled with corner alignment to obtain the target rendering path;

[0220] The robot is controlled to perform tasks based on the target execution path, and the target rendering path is output to the designated display terminal.

[0221] Specifically, the processor 13's implementation method for the above instructions can be found in [reference needed]. Figure 1 The descriptions of the relevant steps in the corresponding embodiments are not repeated here.

[0222] It should be noted that all data involved in this case was legally obtained. Software tools or components not belonging to this company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.

[0223] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0224] This invention can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0225] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0226] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0227] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0228] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0229] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices described in this invention can also be implemented by a single unit or device through software or hardware. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0230] Finally, it should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A dynamic path planning method for large-scale raster maps, characterized in that, The dynamic path planning method for large-scale raster maps includes: In response to robot path planning instructions based on an initial grid map, environmental change data is collected, and the initial grid map is processed according to the environmental change data to obtain a three-layer policy mask, an integral map, and an obstacle departure cost map; wherein, the three-layer policy mask includes a permitted normal layer, an execution-state temporary extension layer, and a service alarm layer; Based on the Region of Interest (ROI)-thick line viewthrough direct search strategy, initial path planning is performed on the permitted normal layer according to the integral map and the obstacle departure cost map to obtain the baseline path and the initial rendering path. This includes: using the permitted normal layer as the basic range constraint, generating obstacle blocking data and positioning errors based on the obstacle departure cost map; obtaining the queue situation within the three-layer strategy mask; generating a first ROI based on the obstacle blocking data, the positioning error, and the queue situation, and searching for multiple potential paths within the first ROI; sampling along the multiple potential paths at a preset step size to obtain multiple sampling points; for the neighborhood window of each sampling point, using the thick line viewthrough verification algorithm, calculating the safe pixel ratio based on the integral map, and determining the sampling point to be safe when the safe pixel ratio is greater than a preset threshold; obtaining potential paths where all sampling points are safe as candidate paths; and performing path point safety simplification processing on the candidate paths to obtain the baseline path and the initial rendering path. The robot's execution status based on the baseline path and real-time environmental changes are monitored in real time, and whether a detour event or alarm event is triggered based on the execution status and real-time environmental changes. When the detour event is detected, detour path planning is performed in the permitted normal layer and the execution temporary extension layer based on the region of interest-thick line view direct pull search strategy to obtain candidate detour path segments; Based on the baseline path and the initial rendering path, the candidate detour path segments are subjected to regression convergence processing to obtain the target execution path and intermediate rendering path. The intermediate rendering path is resampled with corner alignment to obtain the target rendering path; The robot is controlled to perform tasks based on the target execution path, and the target rendering path is output to the designated display terminal.

2. The dynamic path planning method for large-scale raster maps as described in claim 1, characterized in that: The permitted normal layer forms an obstacle avoidance buffer zone based on the shrinking bandwidth, which is used to perform shrinking morphological processing on the passable area; wherein, the shrinking bandwidth is equal to the sum of the robot's half-width and the positioning deviation range, and the shrinking bandwidth is dynamically adjusted with the positioning accuracy; the cost model of the permitted normal layer adopts a center preference-obstacle avoidance penalty strategy for path selection guidance, and does not contain strategy penalty. The execution-state temporary extension layer is used to perform outward morphological processing on the temporary permitted area; wherein, the outward bandwidth is less than the configured threshold; the activation constraints of the execution-state temporary extension layer include time window constraints, budget constraints, no prohibited conflict constraints, and concurrent capacity constraints; the policy cost of the execution-state temporary extension layer includes: additional penalties to guide the robot to prioritize the permitted normal layer; the revocation conditions of the execution-state temporary extension layer include: reaching the earliest return point, budget and time window exhaustion, manual revocation, and policy withdrawal; the execution-state temporary extension layer includes a cumulative cost plus time growth term to encourage the robot to return to the permitted normal layer as soon as possible; The business alarm layer includes physically accessible but business-sensitive or authorized areas; wherein, when the robot enters the business alarm layer, a tiered alarm is triggered. The obstacle departure cost map is constructed based on a distance field to guide the robot to prioritize paths away from obstacles; wherein, the distance field is calculated based on the Euclidean distance transformation algorithm.

3. The dynamic path planning method for large-scale raster maps as described in claim 1, characterized in that, The process of performing path point safety simplification on the candidate path includes: From the candidate paths, obtain the corners that are less than the included angle threshold as true corners, and retain the sampling points corresponding to the true corners as true corner points; Further simplification is prohibited within the preset angle after the corner; Merge the sample points that are collinear in the candidate paths; Identify spike segments in the candidate paths and merge the spike segments into adjacent steady-state segments; Eliminate the closest points to the start and end points in the candidate paths; Among them, all sampling points must remain safe when performing path point safety simplification processing; In particular, after performing path point safety simplification, the turning angle of adjacent line segments is matched with the rotatable radius of the underlying controller.

4. The dynamic path planning method for large-scale raster maps as described in claim 1, characterized in that, The step of detecting whether a detour event or alarm event is triggered based on the execution state and the real-time environmental changes includes: The robot's real-time positioning data is obtained based on the execution status, and based on the real-time positioning data, the real-time environmental changes, and the baseline path, it is detected whether there are obstacles blocking the current execution path, and the blocking detection result is obtained. Based on the real-time location data and the real-time environmental changes, the queue-yield detection result is obtained to determine whether there are queuing events and yielding events. The deviation between the robot's actual position and the baseline path is calculated based on the real-time positioning data and the baseline path to obtain the positioning deviation detection result; Real-time detection of task change information; Based on the execution status and the real-time environmental changes, predict whether the alarm event will occur, and obtain the alarm event prediction result; Whether to trigger the detour event is determined based on the blocking detection result, the queue-giveaway detection result, the positioning deviation detection result, and the task change information; and whether to trigger the alarm event is determined based on the alarm event prediction result.

5. The dynamic path planning method for large-scale raster maps as described in claim 2, characterized in that, The region-of-interest-thick-line-view-direct-pull search strategy performs detour path planning in the permitted normal layer and the execution-state temporary extension layer, resulting in candidate detour path segments including: Obtain the robot's current pose, obstacle envelope, and time step; Predict the robot's position after the time step; Determine the minimum bounded bounding box based on the current pose, the obstacle envelope, and the predicted position; The minimum bounded bounding box is expanded by a preset multiple to obtain the second region of interest; Detect whether the execution-state temporary extension layer satisfies the enable constraint; When the execution-state temporary extension layer satisfies the enable constraint, the union of the permitted normal layer and the execution-state temporary extension layer in the second region of interest is determined as a passable region. Based on the strategy cost, a path search is performed within the passable area to obtain multiple candidate sub-segments; The thick line visibility verification algorithm is used to perform line segment-level security detection on the multiple candidate sub-segments; Candidate sub-segments that pass the line segment-level safety detection are determined as candidate detour path sub-segments.

6. The dynamic path planning method for large-scale raster maps as described in claim 1, characterized in that, The step of performing regression convergence processing on the candidate detour path segments based on the baseline path and the initial rendering path to obtain the target execution path and intermediate rendering path includes: Multiple candidate regression points are selected one by one along the potential splicing area between the candidate bypass path segment and the baseline path; The thick line vision verification algorithm is used to perform security checks on the line segments from the candidate detour path sub-segments to each candidate regression point and the line segments from each candidate regression point to the subsequent segments of the baseline path, to obtain the thick line vision security verification results. Calculate the total path cost after concatenating the multiple candidate regression points, and calculate the initial path cost of the corresponding path in the baseline path; compare the total path cost with the initial path cost to obtain the cost superiority verification result; Calculate the alignment error between the candidate detour path segment and the baseline path at each candidate regression point to obtain the geometric alignment verification result; When the thick line visibility security verification result corresponding to any candidate regression point is that the line segments from the candidate detour path sub-segment to each candidate regression point and the line segments from each candidate regression point to the subsequent segments of the baseline path all pass the security detection, the corresponding cost superiority verification result is that the total path cost is less than the difference between the initial path cost and the jitter value, and the corresponding geometric alignment verification result is that the alignment error is less than the error threshold, it is determined that the arbitrary candidate regression point meets the security conditions, and the detection continues to check whether a consecutive preset number of candidate regression points all meet the security conditions; When the predetermined number of consecutive candidate regression points all meet the safety conditions, any candidate regression point is determined as a valid regression point. Select the point closest to the starting point of the detour from all valid regression points as the earliest regression point; The candidate detour path segment from the corresponding starting point to the earliest regression point is concatenated with the segment from the earliest regression point to the end point of the baseline path to obtain the target execution path; Close the execution-state temporary extension layer and clear the execution data of the execution-state temporary extension layer; The target execution path is rendered to obtain the intermediate rendering path.

7. The dynamic path planning method for large-scale raster maps as described in claim 1, characterized in that, The step of performing corner alignment resampling on the intermediate rendering path to obtain the target rendering path includes: The intermediate rendering path is phase-reset segment by segment to obtain multiple sampling segments; wherein, when performing phase-reset segment by segment, all original polyline vertices are retained; For each sampling segment, the previous sampling point is designated as the first sampling point, and the next sampling point is calculated along the sampling segment, starting from the first sampling point, according to the sampling step size. It is then checked whether the next sampling point enters the buffer at the end of the sampling segment. When the next sampling point enters the buffer, the end point of the sampling segment is designated as the target point. When the next sampling point does not enter the buffer, the thick line visibility check algorithm is used to perform a security check on the line segment formed by the first sampling point and the next sampling point. If the security check is passed, the next sampling point is designated as the target point, and the calculation of the next sampling point continues until all sampling segments are processed. Wherein, when the line segment formed by the first sampling point and the next sampling point fails the security check, the end point of the sampling segment is inserted as a bridging point into the sampling segment, the first sampling point is updated as the bridging point, and the processing of the next sampling segment formed by the bridging point continues. The target rendering path is generated based on the original polyline vertices, the target point, and the bridging point.

8. The dynamic path planning method for large-scale raster maps as described in claim 5, characterized in that, After the region of interest-thick line view direct pull search strategy performs detour path planning in the permitted normal layer and the execution state temporary extension layer, the method further includes: When detour path planning fails, the preset multiplier is incrementally increased to expand the second region of interest until the preset multiplier reaches its upper limit. Within the expanded second region of interest, detour path planning continues based on the region of interest-thick line view direct pull search strategy; and / or Adjust the parameters of the strategy cost; and / or Adjust the neighborhood window; and / or Obtain the detour path planning record, report the detour path planning record to the designated operation and maintenance terminal, and guide the robot to a safe waiting position.

9. The dynamic path planning method for large-scale raster maps as described in claim 1, characterized in that, After detecting whether a detour event or alarm event is triggered based on the execution state and the real-time environmental changes, the method further includes: When the alarm event is detected, a graded response is performed according to the execution status and the real-time environmental changes, and the robot's execution data in the business alarm layer is monitored and recorded in real time. When the robot is detected to have left the business alarm layer, it is checked whether the robot meets the regression conditions; When the robot meets the regression condition, it stops performing hierarchical responses and controls the robot to return to the normal state within the baseline path.

Images