Unmanned aerial vehicle self-adaptive inspection path planning system
By identifying the order of task numbers, processing boundary relationships, locating interruption points, and correcting abnormal sections, the system solves the problems of missing path logic and insufficient connectivity in traditional UAV inspection path planning systems, and achieves efficient path reconstruction and task dispatch.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-29
- Publication Date
- 2026-04-14
AI Technical Summary
Traditional UAV adaptive inspection path planning systems rely on prior maps and fixed inspection points, lacking global perception and real-time reconstruction capabilities for dynamic environments. This results in missing, overlapping, or gapped path logic, failure to participate in spatial judgment of task boundaries, incomplete identification of path connectivity, misjudgment of task status, and a lack of numbering mapping and node access mechanisms for path adjustment, thus limiting reconstruction capabilities and dispatch efficiency.
The task association extraction module identifies the order of task numbers, the layer path division module processes boundary relationships, the node connectivity filtering module locates the interruption location, the flight segment status identification module judges consistency, and the path reconstruction and integration module forms a complete path, ensuring logical consistency of the path, reducing spatial intersections and duplicate coverage, maintaining path continuity, identifying and correcting abnormal segments, and completing path completion.
It achieves internal logical consistency of task paths, reduces spatial overlap and redundant coverage, maintains path continuity, identifies and corrects abnormal segments, and forms a continuous path structure that can be directly used for task dispatch, thereby improving the efficiency and accuracy of UAV inspection tasks.
Smart Images
Figure CN121857772A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of unmanned aerial vehicle (UAV) path planning technology, and more particularly to an adaptive inspection path planning system for UAVs. Background Technology
[0002] The field of UAV path planning technology involves the intelligent generation and dynamic adjustment of paths for unmanned aerial vehicles (UAVs) during mission execution, mainly including core components such as path modeling, planning algorithms, obstacle avoidance, environmental perception, and multi-aircraft collaboration. By constructing a 3D flight model and combining real-time perception and multi-objective optimization, efficient navigation of UAVs in complex environments can be achieved.
[0003] Traditional UAV adaptive inspection path planning systems refer to systems that autonomously generate and adjust flight paths based on the characteristics of the inspection area and the distribution of targets, primarily addressing path optimization problems in complex inspection tasks. Traditional methods often rely on prior maps and fixed inspection points, employing algorithms such as A* or ant colony optimization for path generation, and adjusting local paths using simple rules in unforeseen circumstances. They lack the ability to achieve global perception and real-time reconstruction of dynamic environments.
[0004] Path generation relies on fixed points and preset layers, ignoring the sequential relationships in the task text, which easily leads to confused numbering and missing path logic. Task boundaries are not involved in spatial judgment, layer division is rough, and task overlap or gaps are common. Path connectivity identification depends on the results of a single mapping, lacks node relationship tracking, and breakpoint identification is incomplete. Task status determination is based on a single completion marker, and trajectory anomalies and offsets cannot be reflected, causing misjudgments. Path adjustment lacks numbering mapping and node access mechanisms, limiting reconstruction capabilities and dispatch efficiency. Summary of the Invention
[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing an adaptive inspection path planning system for unmanned aerial vehicles (UAVs).
[0006] To achieve the above objectives, the present invention adopts the following technical solution: an adaptive inspection path planning system for unmanned aerial vehicles (UAVs), the system comprising, The task association extraction module reads the number and location in the drone inspection task list, identifies number pairs with execution relationship in the description text, arranges the number pairs in sequence, extracts the task numbers involved in the arrangement as the processing target, and obtains the path logic task connection number group. The layer path division module calls the number in the path logic task connection number group, extracts the corresponding coordinates of the task and the boundary line of the coverage area, processes the boundary contact relationship, and groups the non-overlapping task numbers into the same type of set to obtain the layer task path grouping mapping list. The node connectivity filtering module reads the set of numbers in the task path grouping mapping list of the layer, retrieves the node data corresponding to the task path, identifies the connection relationship between nodes, finds the missing part, includes the path interruption position and task number in the result range, and obtains the set of breakpoint inspection path connection intervals. The flight segment status identification module calls the task number in the set of breakpoint inspection path connection intervals, retrieves the flight status, completion mark and trajectory offset content, judges the consistency of the task preset requirements, and merges the inconsistent task numbers and path nodes into the current segment result to obtain the path status difference segment number set. The path reconstruction and integration module reads the task number within the path state difference segment number set, locates the node start and end information in the original path, calls the relevant number node data in the layer task path grouping mapping list, and integrates it into the existing node sequence to form a complete path for task dispatch, thus obtaining the UAV adaptive inspection path structure sequence.
[0007] As a further embodiment of the present invention, the path logic task connection number group includes task sequence relationship number pairs, task sorting result set, and path connection logic structure; the layer task path grouping mapping list includes layer division identifier, set of similar task numbers, and non-overlapping task boundary relationships; the breakpoint inspection path connection interval set includes path interruption position number, missing connection task pairs, and abnormal continuation segment identifier; the path state difference segment number set includes execution state abnormal number, completion state conflict number, and trajectory offset abnormal number; and the UAV adaptive inspection path structure sequence includes node reconstruction results, layer path integration number, and task dispatch path structure unit.
[0008] As a further aspect of the present invention, the task association extraction module includes: The task number extraction submodule retrieves the text content from the drone inspection task list, calls the task number and corresponding location description in the text, maps the task number to the location one by one, indexes and compares the number location according to the order of the number appearance, deletes duplicate numbers, and marks the number order to form task number order mapping information. The sequence relationship identification submodule identifies sequence words and directional terms in the description text based on the task number and description content in the task number sequence mapping information, determines whether there is an execution sequence between tasks, extracts the task numbers involved in the sequence relationship by comparing the numbers, and checks the position between the numbers mentioned in the sequence text to obtain the task number sequence pointing relationship. The path number connection submodule calls the number information involved in the task number sequence pointer relationship, connects the task number pairs in sequence according to the order of the description text, links each group of task numbers according to the description order, and merges all task number pairs into a set of sequential number relationships to obtain the path logical task connection number group.
[0009] As a further aspect of the present invention, the layer path division module includes: The task coordinate extraction submodule calls all task numbers in the path logical task connection number group, obtains the center point coordinates and coverage area boundary line data of the task corresponding to each number, associates the coordinate data with the position number, performs continuous edge point splitting processing on the boundary line data and associates the number order, and obtains the task number boundary coordinate information. The boundary relationship discrimination submodule, based on the boundary line point set corresponding to each number in the task number boundary coordinate information, performs envelope region intersection judgment on any two boundary point sets, extracts boundary contact point segments, identifies and merges number pairs without intersection areas, and obtains non-intersecting task number set information. The task classification and filtering submodule performs duplicate comparison and sequential screening on the task number combinations in the non-handover task number set information, filters out task numbers with overlapping boundaries, and processes the screened numbers sequentially to obtain the layer task path grouping mapping list.
[0010] As a further aspect of the present invention, the node connectivity filtering module includes: The path node reading submodule reads the contents of each number set in the task path grouping mapping list of the layer, calls the node coordinate data and number order of the task corresponding to the number, organizes the node coordinates in each path according to the number order, detects whether there is continuous jump behavior between coordinate points, and obtains the continuous status information of the path node. The node relationship discrimination submodule calculates the spatial interval between adjacent coordinate points based on the coordinate jump segments in the continuous state information of the path nodes, locates the node number of the coordinate jump location, collects and compares the task numbers before and after the jump segment, and obtains the interrupted node number docking record. The breakpoint segment extraction submodule calls all numbered segments in the interrupted node number docking record, and extracts segments of the path number with breakpoints in combination with the original path number order. The corresponding task number is segmented into a set according to the order before and after consecutive jumps, and the breakpoint inspection path connection interval set is obtained.
[0011] As a further aspect of the present invention, the flight segment status identification module includes: The task status reading submodule calls the task number in the set of breakpoint inspection path connection intervals, collects the corresponding task flight execution status, completion mark and trajectory offset description, pairs the task number with the execution order of each status field and aligns the fields, performs position verification on missing status fields, and forms a continuous status correspondence relationship based on the number order to obtain the task status corresponding information. The status comparison and discrimination submodule, based on the flight execution status, completion identifier and trajectory deviation description in the task status corresponding information, compares the flight status description, completion node description and trajectory direction description in the task preset requirements, performs item-by-item comparison of each field under the same task number, identifies inconsistent positions and locks the corresponding path nodes, and obtains the status deviation node information. The abnormal flight segment extraction submodule calls the task number and path node in the state deviation node information, checks the position of the node in the path in a sequential manner, includes the path segments with state deviation between consecutive nodes in the same range, and uniformly classifies the involved task numbers and path segments into the result set to obtain the path state difference segment number set.
[0012] As a further aspect of the present invention, the path reconstruction and integration module includes: The number node positioning submodule reads the task number from the path state difference segment number set, retrieves the corresponding start and end node positions in the original path node sequence, performs one-to-one pairing of task number and node position, and performs continuity judgment on the distance between start and end nodes to obtain path breakpoint node sequence information. The node data extraction submodule retrieves all node coordinates related to the corresponding number in the layer task path grouping mapping list based on the task number in the path breakpoint node sequence information, performs number order alignment and position verification on the extracted nodes, and performs a removal operation on duplicate node positions to obtain supplementary path node information. The path structure update submodule calls the node content in the path supplement node information, performs access processing on the end of the original path node sequence, adds the new nodes to the end of the original path in numerical order, and performs node continuity verification on the entire path to obtain the UAV adaptive inspection path structure sequence.
[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the path connection sequence is constructed by parsing the numbering relationship in the task description text, so that the task connection has an inherent logical consistency. The task layer is divided by the contact relationship of the region boundary to reduce spatial intersection and repeated coverage. The path interruption section is located by analyzing the node connection relationship to maintain the path continuity. Abnormal sections are identified by comparing the flight status, completion mark and trajectory offset. The path is completed by node reconnection and number mapping to form a continuous path structure that can be directly used for task dispatch. Attached Figure Description
[0014] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart illustrating the task association extraction module of the present invention. Figure 3 This is a flowchart illustrating the acquisition process of the layer path partitioning module in this invention. Figure 4 This is a flowchart illustrating the node connectivity filtering module of the present invention. Figure 5 This is a flowchart illustrating the acquisition process of the flight segment status identification module of the present invention. Figure 6 This is a flowchart of the path reconstruction and integration module of the present invention. Detailed Implementation
[0015] The technical solution of the present invention will now be described with reference to the accompanying drawings.
[0016] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.
[0017] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.
[0018] Please see Figure 1 This invention provides a technical solution: an adaptive inspection path planning system for unmanned aerial vehicles (UAVs), the system comprising: The task association extraction module reads the task number and location description in the drone inspection task list, identifies the number pairs with execution sequence relationship in the description text, arranges all number pairs in sequence, extracts the task numbers involved in the arrangement and uses them as the target task number set, and obtains the path logic task connection number group. The layer path division module calls the number in the path logic task connection number group, extracts the coordinate information of the corresponding task and the boundary line of the coverage area, identifies the contact relationship between the area boundaries, and groups the task numbers that do not overlap between the boundaries into the same type of set, thus obtaining the layer task path grouping mapping list. The node connectivity filtering module reads the set of numbers in the layer task path grouping mapping list, retrieves the node data corresponding to the task path, identifies the continuity relationship between nodes, finds the missing connection parts, includes the path interruption position and related task number in the result range, and obtains the set of breakpoint inspection path connection intervals. The flight segment status identification module calls the task number in the set of breakpoint inspection path connection intervals, retrieves the flight execution status, completion mark and trajectory offset description of the task, judges the consistency with the task preset requirements, and imports the task number and path node involved in the inconsistent content into the results of this segment to obtain the set of path status difference segment numbers. The path reconstruction and integration module reads the task number from the path status difference segment number set, locates the start and end information of nodes in the original path, calls the node set with the relevant number in the layer task path grouping mapping list, and connects the node set with the relevant number to the end of the original path node sequence to form a complete path that can be used for task dispatch, thus obtaining the UAV adaptive inspection path structure sequence.
[0019] The path logic task connection number group includes task sequence relationship number pairs, task sorting result set, and path connection logic structure. The layer task path grouping mapping list includes layer division identifier, set of similar task numbers, and non-overlapping relationship of task boundaries. The breakpoint inspection path connection interval set includes path interruption position number, missing connection task pairs, and abnormal continuation segment identifier. The path status difference segment number set includes execution status abnormal number, completion status conflict number, and trajectory offset abnormal number. The UAV adaptive inspection path structure sequence includes node reconstruction result, layer path integration number, and task dispatch path structure unit.
[0020] Please see Figure 2 The task association extraction module includes: The task number extraction submodule retrieves the text content from the drone inspection task list, calls the task number and corresponding location description in the text, maps the task number to the location one by one, indexes and compares the number location according to the order of the number appearance, deletes duplicate numbers, and marks the number order to form task number order mapping information. The system retrieves a plain text dataset from the drone inspection task list, which records all work locations involved in power grid inspections. Task numbers such as "TASK-001" and "TASK-002" in the text are associated with their following location description fields, such as "110kV line tower No. 01". During execution, the system first performs a full-field scan of the text, extracting task strings containing specific prefixes and simultaneously capturing the geographical location descriptions within a 50-byte range following each task number. A one-to-one mapping relationship is established between task numbers and physical locations, forming a location association base library. Based on the physical order of the numbers appearing in the text, the starting byte position of each number is indexed. By comparison, duplicate indexes are found; for example, "TASK-002" is recorded in both byte 150 and byte 450 of the document. At this point, the program initiates deduplication logic, comparing the semantic consistency of the two positions and performing duplicate deletion, retaining only the index position of the first occurrence. Based on this, the system assigns incremental sequential scalars to the deduplicated numbers in their original order, such as labeling TASK-001 as sequence 1, TASK-003 as sequence 2, and TASK-002 as sequence 3. By calculating the byte index offset between adjacent numbers, comparing the physical spacing of different tasks in the document, and marking the numbering order, a task numbering order mapping information is formed. This process ensures the originality and accuracy of the task list when it is converted into a logical sequence through precise anchoring of the text's physical location, outputting complete and non-redundant task numbering order mapping information.
[0021] The sequence relationship identification submodule identifies the task numbers and descriptions in the task number sequence mapping information, identifies sequence words and directional terms in the description text, determines whether there is an execution sequence between tasks, extracts the task numbers involved in the sequence relationship by comparing the numbers, and checks the position between the numbers mentioned in the sequence text to obtain the task number sequence pointing relationship. The system reads the task number sequence mapping information and performs a deep semantic search on the descriptive text to identify time-related words such as "first," "next," and "after," as well as directional heading words such as "northward inspection," "turn left," and "fly along the route." During execution, the submodule determines whether there is an execution order between tasks based on the task numbers and descriptions in the task number sequence mapping information. For example, if the text description is "After completing TASK-001, fly 300 meters north to reach TASK-003," the logic determines that TASK-001 is the preceding task and TASK-003 is the following task. The system compares the extracted sequence words with the task numbers based on their proximity. The sequence word recognition window is set to 20 characters before and after the task number. If two task numbers and sequential conjunctions such as "after" appear simultaneously within this window, the system extracts the task number pairs involving a sequential relationship through number comparison. The positions between the numbers mentioned in the sequential text are checked, the geographic orientation vector between the two points is calculated, and the matching of the heading terms in the text with the actual latitude and longitude displacement direction is verified. If the displacement vector direction is due north and the text description is "northward," then the logical relationship is confirmed, and the sequential pointing relationship of the task numbers is obtained. This process eliminates any logical ambiguity in the text description, ensures the topological correctness of the inspection task chain, and generates a sequential pointing relationship of task numbers with physical basis.
[0022] The path number connection submodule calls the number information involved in the task number sequence pointer relationship, connects the task number pairs in sequence according to the order of the description text, links each group of task numbers according to the description order, and merges all task number pairs into a set of sequential number relationships to obtain the path logical task connection number group.
[0023] The system retrieves the numbering information from the task number sequence pointers in the relation, connecting the task number pairs sequentially according to the order in the description. During execution, the submodule first reads all binary relation groups, such as "TASK-001 to TASK-003" and "TASK-003 to TASK-002," using TASK-003 as the logical connection anchor point for cascading processing. The system links each group of task numbers according to the described order, checking whether the first and last identifiers of the number pairs can form a closed loop or a continuous chain, and uniformly aggregating the scattered task number pairs according to the logical sequence. For task points with branches or intersections, the system sorts the paths according to the priority weight in the task number sequence pointer relation, ensuring that all task number pairs are uniformly merged into a single sequential numbering relation. By performing a cyclic consistency check on the connected number stream, logical errors that may cause track backflow are eliminated, and all independent task points are linked into a logical long chain with a unique flow direction. After the connection is completed, the system evaluates the integrity of the entire chain, confirming that it covers all deduplicated numbers in the task list, and obtains the path logical task connection number group.
[0024] Please see Figure 3 The layer path division module includes: The task coordinate extraction submodule calls all task numbers in the path logic task connection number group, obtains the center point coordinates and coverage area boundary line data of the task corresponding to each number, associates the coordinate data with the position number, performs continuous edge point splitting processing on the boundary line data and associates the number order, and obtains the task number boundary coordinate information. The system invokes the path logic task connection group to access all task numbers, retrieving the center point coordinates and coverage area boundary data for each task. During execution, the submodule first accesses the power grid geographic information system (GIS) database, retrieving the center point geographic coordinates (e.g., 30.123456°N, 120.654321°E) for task number TASK-001, along with a set of 20 closed boundary vector points. The system associates the coordinate data with location numbers, tagging each latitude and longitude coordinate with a corresponding task ID. It then performs continuous edge point splitting on the boundary data, disassembling the originally closed polygonal boundary line clockwise into discrete edge coordinate points with sequential indexes, and associating them with the number sequence. This process transforms the static area into a dynamic sequence of edge points, allowing the system to obtain precise spatial occupancy information for the tasks. After extracting the coordinates of all task numbers, the submodule constructs a multi-dimensional data matrix, integrating and verifying the center point coordinates, edge point sets, and task sequence to ensure the spatial continuity and accuracy of the task number boundary coordinate information. This process maps the logical task number to the real three-dimensional geographic space by obtaining the task number boundary coordinate information.
[0025] The boundary relationship discrimination submodule, based on the boundary line point set corresponding to each number in the boundary coordinate information of the task number, performs envelope region intersection judgment on any two boundary point sets, extracts the boundary contact point segment, identifies and merges number pairs without intersection areas, and obtains information on the set of non-intersecting task numbers. Based on the boundary point sets corresponding to each task number in the boundary coordinate information, the submodule performs envelope region intersection judgment on any two boundary point sets. During execution, the submodule first establishes a minimum bounding rectangle (MBR) envelope for the edge point sets of TASK-001 and TASK-003, and calculates the overlap area of the two envelopes. If the overlap area is greater than 0, the minimum Euclidean distance between the two boundary point sets is further calculated. The system sets the contact judgment threshold to 0.05 meters. If the calculated shortest distance is less than or equal to this threshold, the boundary contact point segment is extracted, and the two tasks are marked as geographically "continuously adjacent". Conversely, if the shortest distance between TASK-001 and TASK-005 is 45.80 meters, which is much greater than the set spatial offset margin of 20 meters, it is determined to be a non-intersection area. The number pairs of non-intersection areas are identified and merged, classified as "isolated task blocks", and the non-intersection task number set information is obtained. This process uses high-precision geometric spatial discrimination to identify continuous and broken segments in the spatial distribution of the inspection path, avoiding the forced classification of physically unconnected tasks into the same layer, thereby obtaining a scientifically reasonable set of non-handover task numbers.
[0026] The task classification and filtering submodule performs duplicate comparison and sequential screening on the task number combinations in each group of number sets based on the non-handover task number set information. It filters out task numbers with overlapping boundaries between numbers, and processes the numbers that pass the screening in sequence to obtain the layer task path grouping mapping list.
[0027] Based on the task number combinations in the non-handover task number set information, the system performs duplicate comparison and sequence screening on the numbers in each number set. During execution, the submodule first performs internal consistency verification for each non-handover set to check if any task numbers have been incorrectly assigned to multiple unrelated sets. The system then performs sequence screening on each group of numbers, retrieving their position in the original path logical task connection number group, and filtering out task numbers with overlapping boundaries to prevent UAV flight conflicts caused by boundary overlap. The numbers that pass the screening are then reordered to ensure that the task points within each path group satisfy a single flow direction and do not interfere with each other. The system verifies the rationality of the grouping by calculating the centroid displacement vector of each group of tasks, ensuring that the task paths within the same layer have a high degree of clustering characteristics in physical distribution. After completing the iterative processing of all sets, the submodule generates a detailed list, recording the task number sequence and its spatial weight value corresponding to each layer, resulting in a layer task path grouping mapping list.
[0028] Please see Figure 4 The node connectivity filtering module includes: The path node reading submodule reads the contents of each number set in the layer task path grouping mapping list, calls the node coordinate data and number order of the task corresponding to the number, organizes the node coordinates in each path according to the number order, detects whether there is continuous jump behavior between coordinate points, and obtains the continuous status information of the path node. The system reads the contents of each numbered set in the layer task path grouping mapping list and retrieves the node coordinate data and number order of the corresponding task. During execution, the submodule sequentially accesses each path group and retrieves the full node coordinates of the task points contained in each path, including takeoff points, track turning points, equipment monitoring points, and landing points. The system organizes the node coordinates in each path according to their number order and establishes a bidirectional index based on timestamps and positional order. Jump behavior detection is performed, calculating the displacement vector length and angle change rate between adjacent coordinate points. If the Euclidean distance between two consecutive nodes exceeds the maximum step distance of 100 meters for a single UAV cruise, or the instantaneous change in track angle exceeds 90 degrees, the system detects it as a continuous jump behavior between coordinate points at the data level. By traversing thousands of node coordinates, the submodule can identify smooth transition segments and sudden jump segments in the space of the track and accurately mark the starting point coordinates of the jump segments to obtain continuous state information of the path nodes. This process enables refined analysis from "task blocks" to "trajectory points," and by quantifying the physical connections between nodes, it obtains continuous state information of path nodes that reflects the connectivity performance of the airway.
[0029] The node relationship discrimination submodule calculates the spatial interval between adjacent coordinate points based on the coordinate jump segments in the continuous status information of the path nodes, locates the node number of the coordinate jump location, collects and compares the task numbers before and after the jump segment, and obtains the interruption node number docking record. Based on the coordinate jump segments in the continuous status information of path nodes, the spatial interval between adjacent coordinate points is calculated. During execution, the submodule calls the three-dimensional Euclidean distance formula to accurately measure all segments marked as "jumps". For example, the calculated spatial interval between Node-24 and Node-25 is 120.5 meters, far exceeding the preset smooth connection threshold of 80 meters. The system further locates the coordinate jump locations by node number, identifying the jump point as corresponding to a specific tower crossing segment in the inspection path. The task numbers before and after the jump segment are collected and compared to analyze whether the jump behavior is caused by signal packet loss, obstruction, or original logical break. By comparing with the segment requirements in the original task specification, the system classifies and marks these jump points to determine whether they belong to a legal logical transition segment. After completing the comparison, the submodule associates and stores all abnormal jumps and their associated task numbers, coordinate deviation values, and node indexes, and obtains the interrupted node number docking record.
[0030] The breakpoint segment extraction submodule calls all numbered segments in the interrupted node number docking record, and extracts segments of the path number with breakpoints by combining the original path number order. The corresponding task number is segmented into a set according to the order before and after consecutive jumps, and the set of breakpoint inspection path connection intervals is obtained.
[0031] The system retrieves all numbered segments from the interruption node number docking record and extracts fragments from the path numbers containing breakpoints, based on the original path numbering order. During execution, the submodule first locates each interruption node index in the record, such as the physical breakpoint between TASK-002 and TASK-004. Based on the continuity requirements of the original path, the system segments the path numbers containing breakpoints into a set according to the order of consecutive jumps, breaking a complete task sequence into several physically continuous coordinate segments. A topology consistency check is performed on each extracted segment to ensure that the coordinates of the first and last nodes of each segment are within the corresponding task boundary. The system categorizes and summarizes the path segments before and after the breakpoint through logical mapping of the corresponding task numbers, marking the missing spacing and completion priority of each segment. During processing, the submodule excludes minor displacement deviations caused by normal hovering obstacle avoidance, retaining only break intervals that substantially affect track connectivity. By recombining and mapping these fragmented path segments, a set of structured data packets is generated, obtaining the set of path connection intervals for breakpoint inspection.
[0032] Please see Figure 5The flight segment status identification module includes: The task status reading submodule calls the task number in the breakpoint inspection path connection interval set, collects the corresponding task flight execution status, completion mark and trajectory offset description, matches and aligns the task number with the execution order of each status field, performs position verification for missing status fields, and forms a continuous status correspondence relationship based on the number order to obtain the task status corresponding information. The system calls upon the task number in the breakpoint inspection path connection interval set to collect the corresponding task's flight execution status, completion indicator, and trajectory offset description. During execution, the submodule retrieves data in real time from the flight control system and task management terminal via API interface, obtaining status fields such as "normal cruise," "hovering and shooting," and "mission interrupted." For each task number, the system performs field pairing and alignment operations, precisely binding the timestamp, node coordinates, and flight status. For task points with missing status fields, the submodule initiates a position verification program, filling in the missing fields by retrieving historical sensor logs or associated data from surrounding nodes using linear interpolation or nearest neighbor estimation. A continuous status correspondence is formed according to the number order, ensuring that each inspection node has a complete status description; for example, the completion indicator for TASK-003 is "0" (not completed), and the trajectory offset description value is "4.5 meters." Through the standardization processing of these heterogeneous data, the submodule constructs a data cube reflecting the actual mission execution status, obtaining the corresponding mission status information.
[0033] The status comparison and discrimination submodule compares the flight execution status, completion mark and trajectory deviation description in the mission status information with the flight status description, completion node description and trajectory direction description in the mission preset requirements. It performs item-by-item comparison of each field under the same mission number, identifies inconsistencies and locks the corresponding path nodes to obtain status deviation node information. Based on the flight execution status, completion marker, and trajectory deviation description in the mission status information, the system compares these with the baseline parameters in the mission's preset requirements. During execution, the system retrieves the preset inspection track description, which defines a normal trajectory deviation threshold of 2.0 meters and requires the flight state to remain "uniform straight line." The submodule performs a step-by-step comparison of each field under the same mission number, calculating the dispersion between the real-time value and the preset value. For example, if the calculated difference between the real-time trajectory deviation value of 4.5 meters and the threshold of 2.0 meters is 2.5 meters, and this increment exceeds the set safety redundancy, the system determines that the node is in a "deviation" state. Simultaneously, the submodule identifies inconsistent locations and locks the corresponding path nodes, for example, marking Node-25 as the abnormal starting point. The system further combines the flight status fields to analyze whether the deviation is accompanied by attitude abnormalities or signal interruptions, thereby eliminating sporadic measurement errors. After completing the comparison of all nodes, the submodule summarizes all node indices and their deviation values that do not meet the preset standards to obtain the status deviation node information.
[0034] The abnormal flight segment extraction submodule calls the task number and path node in the status deviation node information, checks the position of the node in the path in a sequential manner, includes the path segments with status deviation between consecutive nodes in the same scope, and uniformly classifies the involved task numbers and path segments into the result set to obtain the path status difference segment number set.
[0035] The system retrieves the task number and path node from the status deviation node information and checks the sequential position of each node within the path. During execution, the submodule initiates a sliding window scan to check the spatial and temporal continuity of nodes marked as "deviation". If Node-25, Node-26, and Node-27 are found to be in a deviation state, the system includes all path segments with status deviations between these consecutive nodes in the same scope and classifies them as "abnormal segments". The submodule uniformly merges the relevant task numbers and path segments and classifies the severity of the deviation in each segment. The system evaluates the impact weight of the abnormal segment on the overall inspection task by calculating the total length and deviation area of the abnormal segment. All path segments identified as abnormal are uniformly included in the result set and assigned corresponding abnormality type labels (such as "heading deviation" or "execution failure"). A record set with detailed coordinate ranges and task indices is generated, resulting in a set of path status difference segment numbers. The execution of this module aggregates scattered error points into logically continuous error segments, providing a clear area to be optimized for subsequent path reconstruction. The output is a set of path state difference segment numbers.
[0036] Please see Figure 6 The path reconstruction and integration module includes: The number node positioning submodule reads the task number from the path status difference segment number set, retrieves the corresponding start and end node positions in the original path node sequence, performs one-to-one pairing of task number and node position, and performs continuity judgment on the distance between start and end nodes to obtain the path breakpoint node sequence information. The system reads the task numbers from the path status difference segment number set and retrieves the corresponding start and end node positions in the original path node sequence. During execution, the submodule first locates the damaged or abnormal task number in the global track index table. For example, for TASK-003, the system finds its start node index to be 45 and its end node index to be 60 in the original sequence. The submodule performs a one-to-one pairing of task numbers and node positions to ensure that the reconstruction process does not negatively affect unrelated segments. It performs a continuity check on the distance between start and end nodes and calculates the spatial Euclidean distance between nodes. If the calculated distance changes abruptly from the normal 15 meters to 35 meters, the system accurately identifies this as a physical breakpoint. By comparing the length of the abnormal segment with the original design length, the submodule can pinpoint the spatial span required for reconstruction. After locating all abnormal tasks, the system constructs a structured list containing start and end indices, breakpoint positions, and spatial gap values, obtaining the path breakpoint node sequence information.
[0037] The node data extraction submodule retrieves all node coordinates related to the corresponding number in the layer task path grouping mapping list based on the task number in the path breakpoint node sequence information. It then aligns the extracted nodes by number order and verifies their positions, and removes duplicate node positions to obtain supplementary path node information. Based on the task number in the path breakpoint node sequence information, the system retrieves the coordinates of all nodes corresponding to that number from the layer task path grouping mapping list. During execution, the submodule performs a reverse lookup to extract the original design node sequence corresponding to the task from a high-precision geographic database, ensuring the completed data has the highest geometric accuracy. The system aligns the extracted nodes by number order and verifies their positions, checking the azimuth deviation between the completed points and the original sequence connection points. If any coordinates in the completed nodes overlap with those in the original path, the system calculates the distance between adjacent points. If the distance is less than 0.01 meters, a removal operation is performed to prevent the UAV from repeatedly navigating at the same location. The extracted longitude, latitude, and altitude data undergo smoothing preprocessing to ensure the node distribution density conforms to the UAV's dynamic constraints. After completing the full extraction, the system verifies the data quality to ensure that each completed node has complete 3D attributes and a task identifier, thus obtaining the path completion node information.
[0038] The path structure update submodule calls the node content in the path supplement node information, performs access processing on the end of the original path node sequence, adds the new nodes to the end of the original path in numerical order, and performs node continuity verification on the entire path to obtain the UAV adaptive inspection path structure sequence.
[0039] The system calls the node information in the path supplementation module to handle the connection points at the end or logical breaks of the original path node sequence. During execution, the submodule first identifies the connection point positions of the original path node sequence and adds the new nodes sequentially to the end of the original path or the corresponding break points according to their numbering. The system performs a topology stitching operation to smoothly correct the connection vectors of the connection points, ensuring that the UAV does not experience severe attitude jitter during flight. A node continuity check is performed on the entire path, traversing all nodes. Given a set of nodes, calculate any two adjacent nodes. and The spatial distance between points is considered. If the distance between all points is less than the preset smoothing constraint value of 20 meters, the reconstruction is considered successful. Global optimization is performed on the generated long chain path, redundant turnaround points are removed, and the node density is adaptively adjusted according to flight control performance parameters. After reconstruction, the system comprehensively scores the connectivity and coverage of the new path to confirm that it fully covers the inspection targets required by the task list. Through this series of integration and verification actions, the UAV adaptive inspection path structure sequence is output. This result shows that the system has achieved self-healing of route breaks and automatic correction of state deviations, obtaining the UAV adaptive inspection path structure sequence.
[0040] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. An adaptive inspection path planning system for unmanned aerial vehicles (UAVs), characterized in that, The system includes: The task association extraction module reads the number and location in the drone inspection task list, identifies number pairs with execution relationship in the description text, arranges the number pairs in sequence, extracts the task numbers involved in the arrangement as the processing target, and obtains the path logic task connection number group. The layer path division module calls the number in the path logic task connection number group, extracts the corresponding coordinates of the task and the boundary line of the coverage area, processes the boundary contact relationship, and groups the non-overlapping task numbers into the same type of set to obtain the layer task path grouping mapping list. The node connectivity filtering module reads the set of numbers in the task path grouping mapping list of the layer, retrieves the node data corresponding to the task path, identifies the connection relationship between nodes, finds the missing part, includes the path interruption position and task number in the result range, and obtains the set of breakpoint inspection path connection intervals. The flight segment status identification module calls the task number in the set of connection intervals of the breakpoint inspection path, retrieves the flight status, completion mark and trajectory offset content, judges the consistency of the task preset requirements, and merges the inconsistent task numbers and path nodes into the results of this segment to obtain the set of path status difference segment numbers.
2. The UAV adaptive inspection path planning system according to claim 1, characterized in that: The path logic task connection number group includes task sequence relationship number pairs, task sorting result set, and path connection logic structure. The layer task path grouping mapping list includes layer division identifier, set of similar task numbers, and non-overlapping task boundary relationship. The breakpoint inspection path connection interval set includes path interruption position number, missing connection task pairs, and abnormal continuation segment identifier. The path state difference segment number set includes execution state abnormal number, completion state conflict number, and trajectory offset abnormal number.
3. The UAV adaptive inspection path planning system according to claim 1, characterized in that, The task association extraction module includes: The task number extraction submodule retrieves the text content from the drone inspection task list, calls the task number and corresponding location description in the text, maps the task number to the location one by one, indexes and compares the number location according to the order of the number appearance, deletes duplicate numbers, and marks the number order to form task number order mapping information. The sequence relationship identification submodule identifies sequence words and directional terms in the description text based on the task number and description content in the task number sequence mapping information, determines whether there is an execution sequence between tasks, extracts the task numbers involved in the sequence relationship by comparing the numbers, and checks the position between the numbers mentioned in the sequence text to obtain the task number sequence pointing relationship. The path number connection submodule calls the number information involved in the task number sequence pointer relationship, connects the task number pairs in sequence according to the order of the description text, links each group of task numbers according to the description order, and merges all task number pairs into a set of sequential number relationships to obtain the path logical task connection number group.
4. The UAV adaptive inspection path planning system according to claim 1, characterized in that, The layer path division module includes: The task coordinate extraction submodule calls all task numbers in the path logical task connection number group, obtains the center point coordinates and coverage area boundary line data of the task corresponding to each number, associates the coordinate data with the position number, performs continuous edge point splitting processing on the boundary line data and associates the number order, and obtains the task number boundary coordinate information. The boundary relationship discrimination submodule, based on the boundary line point set corresponding to each number in the task number boundary coordinate information, performs envelope region intersection judgment on any two boundary point sets, extracts boundary contact point segments, identifies and merges number pairs without intersection areas, and obtains non-intersecting task number set information. The task classification and filtering submodule performs duplicate comparison and sequential screening on the task number combinations in the non-handover task number set information, filters out task numbers with overlapping boundaries, and processes the screened numbers sequentially to obtain the layer task path grouping mapping list.
5. The UAV adaptive inspection path planning system according to claim 1, characterized in that, The node connectivity filtering module includes: The path node reading submodule reads the contents of each number set in the task path grouping mapping list of the layer, calls the node coordinate data and number order of the task corresponding to the number, organizes the node coordinates in each path according to the number order, detects whether there is continuous jump behavior between coordinate points, and obtains the continuous status information of the path node. The node relationship discrimination submodule calculates the spatial interval between adjacent coordinate points based on the coordinate jump segments in the continuous state information of the path nodes, locates the node number of the coordinate jump location, collects and compares the task numbers before and after the jump segment, and obtains the interrupted node number docking record. The breakpoint segment extraction submodule calls all numbered segments in the interrupted node number docking record, and extracts segments of the path number with breakpoints in combination with the original path number order. The corresponding task number is segmented into a set according to the order before and after consecutive jumps, and the breakpoint inspection path connection interval set is obtained.
6. The UAV adaptive inspection path planning system according to claim 1, characterized in that, The flight segment status identification module includes: The task status reading submodule calls the task number in the set of breakpoint inspection path connection intervals, collects the corresponding task flight execution status, completion mark and trajectory offset description, pairs the task number with the execution order of each status field and aligns the fields, performs position verification on missing status fields, and forms a continuous status correspondence relationship based on the number order to obtain the task status corresponding information. The status comparison and discrimination submodule, based on the flight execution status, completion identifier and trajectory deviation description in the task status corresponding information, compares the flight status description, completion node description and trajectory direction description in the task preset requirements, performs item-by-item comparison of each field under the same task number, identifies inconsistent positions and locks the corresponding path nodes, and obtains the status deviation node information. The abnormal flight segment extraction submodule calls the task number and path node in the state deviation node information, checks the position of the node in the path in a sequential manner, includes the path segments with state deviation between consecutive nodes in the same range, and uniformly classifies the involved task numbers and path segments into the result set to obtain the path state difference segment number set.
7. The UAV adaptive inspection path planning system according to claim 1, characterized in that, The system also includes; The path reconstruction and integration module reads the task number in the path state difference segment number set, locates the node start and end information in the original path, calls the relevant number node data in the layer task path grouping mapping list, and integrates it into the existing node sequence to form a complete path for task dispatch, thus obtaining the UAV adaptive inspection path structure sequence. The UAV adaptive inspection path structure sequence includes node reconstruction results, layer path integration numbering, and task dispatch path structure unit.
8. The UAV adaptive inspection path planning system according to claim 7, characterized in that, The path reconstruction and integration module includes: The number node positioning submodule reads the task number from the path state difference segment number set, retrieves the corresponding start and end node positions in the original path node sequence, performs one-to-one pairing of task number and node position, and performs continuity judgment on the distance between start and end nodes to obtain path breakpoint node sequence information. The node data extraction submodule retrieves all node coordinates related to the corresponding number in the layer task path grouping mapping list based on the task number in the path breakpoint node sequence information, performs number order alignment and position verification on the extracted nodes, and performs a removal operation on duplicate node positions to obtain supplementary path node information. The path structure update submodule calls the node content in the path supplement node information, performs access processing on the end of the original path node sequence, adds the new nodes to the end of the original path in numerical order, and performs node continuity verification on the entire path to obtain the UAV adaptive inspection path structure sequence.