Autonomous navigation and obstacle avoidance control method and system for complex terrain optical cable inspection robot

Abstract

This invention relates to the field of navigation and control technology, and in particular to an autonomous navigation and obstacle avoidance control method and system for a fiber optic cable inspection robot in complex terrain. By acquiring the robot's current position information, surrounding environment perception data, and fiber optic cable path information, a terrain-obstacle coupling model is constructed to generate an environmental constraint set, based on which an initial navigation trajectory is generated. During execution, attitude deviation and environmental change information are collected in real time to correct the model. After updating the constraints, the trajectory is regenerated and the control is executed, forming a continuously optimized closed loop, which improves the robot's adaptability to navigation in complex terrain and obstacle avoidance safety.

Description

Technical Field

[0001] This invention relates to the field of navigation and control technology, and in particular to an autonomous navigation and obstacle avoidance control method and system for fiber optic cable inspection robots in complex terrain. Background Technology

[0002] The autonomous navigation and obstacle avoidance problem of fiber optic cable inspection robots in complex terrain environments has long relied on the conventional approach of combining single sensor fusion with offline path planning. Existing technologies typically utilize LiDAR or stereo vision to acquire 3D point clouds, identify obstacles through segmentation algorithms, and then classify the terrain using elevation maps. Robot platforms often employ differential GPS or inertial navigation systems for positioning. Based on this, path planning algorithms such as A* and RRT are used to generate traversable paths from the starting point to the destination. Obstacle avoidance strategies are independent of the terrain model, triggering local replanning only for dynamic obstacles. This approach treats terrain undulations and obstacle distribution as two independent constraints, failing to consider their spatial coupling.

[0003] The aforementioned conventional approaches have significant drawbacks. First, the combined effects of terrain undulations on obstacle occlusion and robot mobility are treated separately. In rugged mountainous areas, steep slopes or ravines can place obstacles in blind spots, causing the perception system to delay capturing dangerous targets. Simultaneously, terrain slope is directly related to the robot's chassis mobility, but existing methods often rely on posterior mobility assessments to temporarily correct paths, failing to balance accessibility and obstacle avoidance during the planning phase. Second, there is a lack of incremental feedback correction mechanisms for real-time environmental changes and robot posture deviations. Once global path planning is complete, traditional methods primarily rely on local obstacle avoidance to evade temporary obstacles, unable to dynamically adjust the spatial mapping between terrain and obstacles. When the robot experiences posture shifts due to ground subsidence or slippage while climbing, the original perception model's correspondence with the actual environment quickly becomes invalid, leading to uncontrollable deviations in the navigation trajectory and even collision risks. Summary of the Invention

[0004] This invention provides an autonomous navigation and obstacle avoidance control method and system for optical cable inspection robots in complex terrain, which can solve the problems in the prior art.

[0005] A first aspect of the present invention provides an autonomous navigation and obstacle avoidance control method for a fiber optic cable inspection robot in complex terrain, comprising:

[0006] Acquire the robot's current location information, surrounding environment perception data, and fiber optic cable path information;

[0007] Based on the surrounding environment perception data, a terrain-obstacle coupling model is constructed. The terrain-obstacle coupling model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing passage safety and accessibility.

[0008] Based on the set of environmental constraints and the optical cable path information, an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety is generated under the constraints of the joint constraint relationship.

[0009] During the execution of the initial navigation trajectory, robot posture deviation information and environmental change information are collected in real time. The posture deviation information and environmental change information are used as feedback inputs to incrementally correct the spatial correlation mapping between terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model, update the joint constraint relationship between terrain accessibility and obstacle occlusion, and obtain the updated environmental constraint set.

[0010] Based on the updated set of environmental constraints, a new navigation trajectory is generated and the robot is controlled to execute it. The newly generated navigation trajectory is then used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop.

[0011] Based on the surrounding environment perception data, a terrain-obstacle coupling model is constructed. This model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing accessibility and reachability, including:

[0012] The surrounding environment perception data is spatially gridded, and the terrain undulation features and obstacle distribution features are extracted in each spatial grid. Based on the terrain continuity and obstacle spatial extension between adjacent spatial grids, the relationship between grids is established to form a spatial relationship mapping.

[0013] Based on the inter-mesh association relationship in the spatial association mapping, by analyzing the constraints of terrain slope change on robot mobility and the limitation of obstacle occlusion range on passable area, a joint constraint relationship between terrain mobility and obstacle occlusion is established, and the spatial association mapping and the joint constraint relationship together constitute a terrain-obstacle coupling model.

[0014] The accessibility and accessibility of each spatial grid are evaluated based on the joint constraint relationship in the terrain-obstacle coupling model. The evaluation results are used as the constraint attributes of the corresponding spatial grid, and the constraint attributes of all spatial grids are summarized to generate an environmental constraint set.

[0015] Based on the inter-mesh relationships in the spatial association mapping, by analyzing the constraints of terrain slope changes on robot mobility and the limitations of obstacle occlusion range on passable areas, a joint constraint relationship between terrain mobility and obstacle occlusion is established, including:

[0016] Based on the inter-grid relationship, the transmission path of terrain slope change between adjacent spatial grids is identified, and the cumulative constraint effect of terrain slope change on robot mobility is determined according to the transmission path, thus obtaining the terrain mobility constraint description.

[0017] Based on the inter-mesh relationship, the diffusion area of ​​obstacle occlusion between adjacent spatial meshes is identified, and the coverage limit of obstacle occlusion on the passable area is determined according to the diffusion area, thus obtaining the obstacle occlusion constraint description.

[0018] The terrain accessibility constraint description and the obstacle occlusion constraint description are coupled and fused to establish a joint constraint relationship between terrain accessibility and obstacle occlusion.

[0019] Based on the set of environmental constraints and the optical cable path information, an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety is generated under the constraints of the joint constraint relationship, including:

[0020] Based on the optical cable path information, a target path framework for the inspection task is constructed. Under the guidance of the target path framework, feasible passage areas are determined according to the set of environmental constraints. Based on the joint constraint relationship, a comprehensive evaluation of the terrain accessibility and obstacle occlusion of the feasible passage areas is carried out to screen out candidate passage areas that meet the requirements of terrain adaptability and obstacle avoidance safety.

[0021] Within the candidate passage area, plan local trajectory segments that connect adjacent target points in the target path framework, ensuring that the local trajectory segments simultaneously meet the requirements of terrain adaptability and obstacle avoidance safety;

[0022] All local trajectory segments are spliced ​​and smoothed according to the order of target points in the target path framework to generate the initial navigation trajectory.

[0023] Using the attitude deviation information and the environmental change information as feedback inputs, the spatial correlation mapping between terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model is incrementally corrected, and the joint constraint relationship between terrain accessibility and obstacle occlusion is updated to obtain the updated set of environmental constraints, including:

[0024] The deviation between the actual representation of terrain undulation features and the predicted representation of terrain undulation features in the spatial association mapping is quantified based on attitude deviation information. The deviation between the actual representation of obstacle distribution features and the predicted representation of obstacle distribution features in the spatial association mapping is quantified based on environmental change information. Each deviation is used as a correction increment to locally adjust the inter-grid association relationship in the spatial association mapping, resulting in the corrected spatial association mapping.

[0025] Based on the adjusted inter-mesh correlation in the modified spatial correlation mapping, the constraints of terrain slope change on robot mobility and the limitation of obstacle occlusion range on passable area are re-analyzed, and the joint constraint relationship between terrain mobility and obstacle occlusion is updated.

[0026] The accessibility and traversability of the spatial grid are reassessed based on the updated joint constraint relationships. The constraint attributes of the corresponding spatial grid are updated, and all updated constraint attributes are summarized to obtain the updated set of environmental constraints.

[0027] Based on the updated set of environmental constraints, a new navigation trajectory is generated and the robot is controlled to execute it. The regenerated navigation trajectory is then used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop, including:

[0028] Based on the updated set of environmental constraints and optical cable path information, the navigation trajectory is regenerated under the constraints of the updated joint constraint relationship. The regenerated navigation trajectory is then compared with the initial navigation trajectory before execution to identify the trajectory adjustment area.

[0029] The robot is controlled to execute according to the navigation trajectory. Within the trajectory adjustment area, robot posture deviation information and environmental change information are collected. The navigation trajectory is used as a new initial navigation trajectory, and the posture deviation information and environmental change information are used as new feedback inputs to trigger the correction of the terrain-obstacle coupling model.

[0030] The updated set of environmental constraints is obtained through correction. The navigation trajectory is regenerated based on the updated set of environmental constraints. The trajectory adjustment area identification and key data acquisition process is repeated to form a continuously optimized navigation control closed loop.

[0031] Based on the updated set of environmental constraints and fiber optic cable path information, a new navigation trajectory is generated under the updated joint constraint relationships. The regenerated navigation trajectory is then compared with the initial navigation trajectory before execution, and the trajectory adjustment area is identified, including:

[0032] Based on the updated set of environmental constraints and optical cable path information, local trajectory segments connecting adjacent target points are replanned under the constraints of the updated joint constraint relationship, and the local trajectory segments are spliced ​​together to generate a navigation trajectory.

[0033] Extract the trajectory segments between the corresponding target points in the navigation trajectory and the initial navigation trajectory, calculate the spatial position offset and path shape difference between the corresponding trajectory segments, determine the spatial region where the trajectory is adjusted based on the spatial position offset and the path shape difference, and mark the spatial region as the trajectory adjustment region.

[0034] A second aspect of the present invention provides an autonomous navigation and obstacle avoidance control system for a fiber optic cable inspection robot in complex terrain, comprising:

[0035] The information acquisition unit is used to acquire the robot's current position information, surrounding environment perception data, and optical cable path information;

[0036] The coupling modeling unit is used to construct a terrain-obstacle coupling model based on the surrounding environment perception data. The terrain-obstacle coupling model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing passage safety and accessibility.

[0037] The trajectory generation unit is used to generate an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety under the constraints of the joint constraint relationship, based on the set of environmental constraints and the optical cable path information.

[0038] The feedback correction unit is used to collect robot posture deviation information and environmental change information in real time during the execution of the initial navigation trajectory, use the posture deviation information and environmental change information as feedback input, perform incremental correction on the spatial correlation mapping of terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model, update the joint constraint relationship of terrain accessibility and obstacle occlusion, and obtain the updated environmental constraint set.

[0039] The closed-loop control unit is used to regenerate the navigation trajectory based on the updated set of environmental constraints and control the robot to execute it. The regenerated navigation trajectory is used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop.

[0040] A third aspect of the present invention provides an electronic device, comprising:

[0041] processor;

[0042] Memory used to store processor-executable instructions;

[0043] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0044] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0045] The terrain-obstacle coupling model integrates terrain undulation features with obstacle distribution features through spatial correlation mapping, establishing a joint constraint relationship between accessibility and safety, significantly improving the terrain adaptability and obstacle avoidance safety of navigation trajectories in complex terrains. The initial navigation trajectory generated based on the set of environmental constraints effectively avoids local dangerous areas, reduces the risk of path failure caused by terrain occlusion or sudden obstacle changes, lowers the probability of robot tipping over or collision, and ensures the continuity of inspection operations.

[0046] Real-time collected robot posture deviation information and environmental change information serve as feedback inputs to incrementally correct spatial correlation mappings, avoiding resource waste caused by full recalculation. The updated joint constraint relationships accurately reflect the impact of dynamic terrain changes and newly added obstacles, ensuring that the environmental constraint set always fits the actual operation scenario. This overcomes the shortcomings of traditional fixed models in coping with environmental evolution and enhances the method's robustness to unstructured terrain.

[0047] The continuously optimized navigation control closed loop uses the corrected navigation trajectory as the new initial trajectory for further feedback correction, forming an adaptive iterative mechanism. Each closed-loop cycle reduces the cumulative effect of sensor noise and execution deviation, gradually improving trajectory accuracy and control stability. The entire process requires no manual intervention, reducing maintenance costs and significantly improving the success rate of long-distance autonomous navigation and real-time obstacle avoidance response speed of the inspection robot in complex terrains such as hills and canyons. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the autonomous navigation and obstacle avoidance control method for a fiber optic cable inspection robot in complex terrain, according to an embodiment of the present invention.

[0049] Figure 2 This is a flowchart illustrating the navigation trajectory adjustment area identification process based on environmental constraints, as described in an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0052] Figure 1 This is a flowchart illustrating the autonomous navigation and obstacle avoidance control method for a fiber optic cable inspection robot in complex terrain, according to an embodiment of the present invention.

[0053] The autonomous navigation and obstacle avoidance control methods for fiber optic cable inspection robots in complex terrain include:

[0054] Acquire the robot's current location information, surrounding environment perception data, and fiber optic cable path information;

[0055] Based on the surrounding environment perception data, a terrain-obstacle coupling model is constructed. The terrain-obstacle coupling model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing passage safety and accessibility.

[0056] Based on the set of environmental constraints and the optical cable path information, an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety is generated under the constraints of the joint constraint relationship.

[0057] During the execution of the initial navigation trajectory, robot posture deviation information and environmental change information are collected in real time. The posture deviation information and environmental change information are used as feedback inputs to incrementally correct the spatial correlation mapping between terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model, update the joint constraint relationship between terrain accessibility and obstacle occlusion, and obtain the updated environmental constraint set.

[0058] Based on the updated set of environmental constraints, a new navigation trajectory is generated and the robot is controlled to execute it. The newly generated navigation trajectory is then used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop.

[0059] In one optional implementation, a terrain-obstacle coupling model is constructed based on the surrounding environment perception data. This model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features through spatial correlation mapping, thereby generating a set of environmental constraints describing accessibility and reachability, including:

[0060] The surrounding environment perception data is spatially gridded, and the terrain undulation features and obstacle distribution features are extracted in each spatial grid. Based on the terrain continuity and obstacle spatial extension between adjacent spatial grids, the relationship between grids is established to form a spatial relationship mapping.

[0061] Based on the inter-mesh association relationship in the spatial association mapping, by analyzing the constraints of terrain slope change on robot mobility and the limitation of obstacle occlusion range on passable area, a joint constraint relationship between terrain mobility and obstacle occlusion is established, and the spatial association mapping and the joint constraint relationship together constitute a terrain-obstacle coupling model.

[0062] The accessibility and accessibility of each spatial grid are evaluated based on the joint constraint relationship in the terrain-obstacle coupling model. The evaluation results are used as the constraint attributes of the corresponding spatial grid, and the constraint attributes of all spatial grids are summarized to generate an environmental constraint set.

[0063] When spatially meshing the surrounding environment perception data, the robot's working area is divided into several uniform 3D spatial grid cells with a fixed resolution. The selection of the grid resolution needs to comprehensively consider the robot's body size and the frequency of terrain changes. Too low a resolution will lead to the loss of detailed terrain features, while too high a resolution will result in an excessive computational burden. In practical applications, the grid size can be adaptively adjusted according to the sensor point cloud density to ensure that each grid contains a sufficient number of perception data points to support feature extraction. For raw perception data collected by multiple sensors such as LiDAR and depth cameras, coordinate system alignment and timestamp synchronization need to be completed first, and then the aligned point cloud data is projected onto the corresponding grid cells.

[0064] Within each spatial grid, terrain undulation features and obstacle distribution features are extracted. The extraction of terrain undulation features is based on the elevation statistics of the point cloud data within the grid, calculating the mean elevation, elevation variance, and slope value along the principal direction within the grid. Let the th grid be... The elevation of each point is There are a total of If there are valid points, then the grid elevation mean for Elevation variance for Slope value The obstacle distribution characteristics are obtained by fitting a local plane to the point cloud within the grid and calculating the angle between the normal vector and the vertical direction. The obstacle distribution characteristics are obtained by performing cluster analysis on the point cloud within the grid, identifying point cloud clusters that are above a certain threshold above the ground reference plane, marking them as potential obstacles, and recording the centroid coordinates, occupied volume, and spatial extension direction of each obstacle cluster.

[0065] When establishing inter-grid relationships based on terrain continuity and obstacle spatial extension between adjacent spatial grids, terrain continuity is quantified by calculating the difference in the mean elevation and slope between adjacent grids. Let adjacent grids... With grid The average elevation values ​​are respectively and The grid spacing is The transition slope between the two grids for .when When the slope exceeds the robot's maximum traversable threshold, the grid pair is marked as a terrain discontinuity, indicating that the robot cannot directly access the grid. Transition to grid The spatial extension association of obstacles is established by determining whether obstacle clusters in adjacent grids belong to the same physical obstacle entity. If obstacle clusters in two adjacent grids are spatially connected, they are merged into the same obstacle entity, and an obstacle extension association marker is established between the grids. The above terrain continuity association and obstacle extension association together constitute a spatial association mapping. The spatial association mapping is stored in the form of a graph structure, with grids as nodes and the relationship between grids as weighted edges. The edge weights encode attributes such as terrain transition slope and obstacle extension strength.

[0066] Based on the inter-mesh relationships in spatial correlation mapping, terrain mobility constraints are established by analyzing the constraints of terrain slope changes on robot mobility. The robot's mobility on terrain with different slopes is comprehensively limited by driving force, center of gravity stability, and adhesion. Regarding slope... Exceeding the robot's rated maximum ramp gradient The grid is used to mark terrain accessibility as impassable; for slopes with a safe accessibility gradient... and The grid between the slopes is marked as restricted passage, and a passage risk coefficient that monotonically increases with the slope is assigned. , ,when hour This indicates that terrain access is unrestricted. Obstacle occupancy constraints are established based on the volume occupied by the obstacle and its spatial extension direction. An obstacle occupancy influence area is generated by extending a certain safe distance outward from the obstacle's center. Mesh within this area is marked as occluded and assigned an occupancy risk coefficient based on its distance from the obstacle's center. The closer the distance The larger the value, the better. Terrain accessibility constraints and obstacle occlusion constraints are superimposed and integrated to form a joint constraint relationship. This joint constraint relationship is based on the joint risk coefficient of each grid cell. express, ,in and These are the weighting coefficients for terrain accessibility and obstacle occlusion, respectively, with a sum of 1, which can be dynamically adjusted according to specific application scenarios. The spatial association mapping and joint constraint relationships together constitute a terrain-obstacle coupling model. This model, using a mesh graph structure, fully describes the coupling relationship between terrain undulation distribution and obstacle spatial distribution.

[0067] Based on the joint constraints in the terrain-obstacle coupling model, the accessibility and safety of each spatial grid are evaluated. Accessibility is assessed using a joint risk coefficient. As the core indicator, when When the grid is below the safety threshold, it is rated as safe and passable; when... When the mesh is between the safe and dangerous thresholds, it is deemed conditionally passable, requiring additional constraints on velocity and attitude; when... When the danger threshold is exceeded, the grid is deemed impassable. The accessibility assessment, based on the accessibility safety assessment, combines the relationships between grids to perform a global connectivity analysis. A graph accessibility algorithm is used to determine whether a safe path exists from the current location to the target area. If a grid passes its own accessibility safety assessment but cannot connect to the target area because its neighboring grids are all marked as impassable, then the grid's accessibility assessment result is unreachable.

[0068] The accessibility and safety assessment results for each grid are combined as the constraint attributes for that grid. These constraint attributes are stored in structured data format, including fields such as safety level label, joint risk coefficient value, accessibility flag, and speed constraint upper limit. The constraint attributes of all spatial grids are aggregated to form an environmental constraint set covering the entire working area. This environmental constraint set is efficiently stored in a sparse matrix format, retaining only grid records with non-zero constraint attributes to reduce storage and query overhead. During subsequent navigation trajectory generation, the environmental constraint set serves as the planning constraint input, ensuring that the generated trajectory always meets the dual requirements of terrain adaptability and obstacle avoidance safety. It also provides a benchmark constraint reference for the incremental correction process, supporting continuous optimization of the navigation control closed loop.

[0069] In one optional implementation, based on the inter-mesh relationships in the spatial association mapping, by analyzing the constraints of terrain slope changes on robot mobility and the limitations of obstacle occlusion range on passable areas, a joint constraint relationship between terrain mobility and obstacle occlusion is established, including:

[0070] Based on the inter-grid relationship, the transmission path of terrain slope change between adjacent spatial grids is identified, and the cumulative constraint effect of terrain slope change on robot mobility is determined according to the transmission path, thus obtaining the terrain mobility constraint description.

[0071] Based on the inter-mesh relationship, the diffusion area of ​​obstacle occlusion between adjacent spatial meshes is identified, and the coverage limit of obstacle occlusion on the passable area is determined according to the diffusion area, thus obtaining the obstacle occlusion constraint description.

[0072] The terrain accessibility constraint description and the obstacle occlusion constraint description are coupled and fused to establish a joint constraint relationship between terrain accessibility and obstacle occlusion.

[0073] After obtaining the spatial correlation mapping between terrain undulation features and obstacle distribution features, it is necessary to further establish joint constraint relationships between terrain accessibility and obstacle occlusion based on the inter-grid correlation relationships to generate a complete set of environmental constraints for subsequent trajectory planning. The inter-grid correlation relationships describe the topological connections and attribute transfer relationships between adjacent spatial grids in dimensions such as elevation, slope, and obstacle distribution. It is the basic data structure for analyzing the transmission path of terrain slope changes and the area of ​​obstacle occlusion diffusion.

[0074] To establish the terrain mobility constraint description, the propagation path of terrain slope changes between adjacent spatial grids is identified based on the inter-grid relationships. Specifically, for any two adjacent grids... and Its transition slope The calculations have been completed in the spatial association mapping stage. Based on this, using the actual mesh sequence traveled by the robot as the transfer path, a cumulative analysis of the transition slope between adjacent mesh pairs is performed along the path direction. Continuity along the path is defined. The cumulative slope constraint effect of a transmission path segment composed of grid nodes. It is measured by weighted accumulation of the transition slope of each adjacent grid within the path segment:

[0075] ;

[0076] in, For the first Distance normalization weights of the path segments. For the first The grid and the first Transition slope between grids This represents the total number of grid nodes contained in the path segment. When Exceeding the robot's rated maximum traversable slope threshold When, the transmission path is marked as impassable; when Between the safe passage slope threshold and During this period, the path is marked as restricted and assigned a corresponding terrain access risk factor. This is used to describe the cumulative constraint degree of the path segment on the robot's mobility. By performing the above analysis on all transmission paths, a terrain mobility constraint description covering the global map is finally obtained. This description uses grids as the basic unit and records the mobility status and cumulative constraint quantization value of each grid in different directions.

[0077] After completing the terrain accessibility constraint description, based on the same inter-mesh correlation, the diffusion region of obstacle occlusion between adjacent spatial meshes is identified. In 3D space, obstacles not only occupy their own mesh but also exert an occlusion diffusion effect on surrounding adjacent meshes due to their volume, height, and the size requirements of the fiber optic cable inspection robot. For a mesh that detects an obstacle, the occlusion diffusion proceeds outward layer by layer along the adjacency direction defined by the inter-mesh correlation, with the diffusion radius... The actual dimensions of the obstacle are determined by both the measured dimensions of the obstacle and the safe clearance between the robot and the obstacle. Let the measured outer dimensions of the obstacle be... The safe distance between the robot body is Then the diffusion radius satisfies:

[0078] ;

[0079] In diffusion radius All grids within the range are included in the obstacle occlusion diffusion zone, and their values ​​are determined based on their actual spatial distance from the obstacle source point. Assign a corresponding obstacle occlusion risk factor The closer the distance, the higher the occlusion risk coefficient; the farther the distance, the lower the risk coefficient becomes, until it decays to zero at the diffusion boundary. This diffusion mechanism allows the obstacle occlusion constraint description to continuously and smoothly reflect the coverage limitation of obstacles on the passable area, avoiding the problem of abrupt constraint boundary changes caused by discretization. By performing diffusion analysis on all obstacle meshes and summarizing the occlusion coverage of each mesh, a complete obstacle occlusion constraint description is finally obtained.

[0080] After obtaining the terrain accessibility constraint description and obstacle occlusion constraint description separately, the two are coupled and fused to establish a joint constraint relationship between terrain accessibility and obstacle occlusion. The core of the coupling and fusion lies in the unified quantification and comprehensive expression of the two types of constraints on the same spatial grid dimension. For each spatial grid, its joint risk coefficient... Based on terrain access risk coefficient Risk factor of obstruction by obstacles Weighted and synthesized according to their respective weighting coefficients:

[0081] ;

[0082] in, This is the terrain accessibility weighting coefficient. These are obstacle occlusion weighting coefficients; the sum of these two coefficients is 1, and both are non-negative. In actual inspection scenarios, and The value can be dynamically adjusted according to the complexity of the terrain and the density of obstacles: in mountainous environments with undulating terrain and sparse obstacles, the value can be appropriately increased. The proportion; in urban utility tunnel environments with relatively flat terrain but dense obstacles, the proportion should be appropriately increased. The proportion of weight. This adaptive weight allocation mechanism enables the joint constraint relationship to more accurately reflect the traffic safety requirements in different scenarios.

[0083] After the joint constraint relationships are established, a joint risk coefficient is assigned to each grid in the global map. and according to The quantization value divides the grid into three states: when When the grid is below a safety threshold, it is marked as passable; when When the value is between the safety threshold and the danger threshold, the grid is marked as a restricted passage state and needs to be constrained in trajectory planning; when... When the danger threshold is exceeded, the grid is marked as impassable and treated as a prohibited area in trajectory planning. These three states together constitute a set of environmental constraints describing accessibility and safety, providing complete constraint input for the generation of the initial navigation trajectory. By coupling and fusing terrain slope transfer path analysis and obstacle occlusion diffusion analysis within a unified grid association framework, the established joint constraint relationship can simultaneously capture the combined impact of continuous terrain changes and partial obstacle occlusion on the robot's mobility, effectively supporting autonomous navigation and obstacle avoidance control in complex terrain fiber optic cable inspection scenarios.

[0084] In one optional implementation, generating an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety under the constraints of the joint constraint relationship, based on the set of environmental constraints and the optical cable path information, includes:

[0085] Based on the optical cable path information, a target path framework for the inspection task is constructed. Under the guidance of the target path framework, feasible passage areas are determined according to the set of environmental constraints. Based on the joint constraint relationship, a comprehensive evaluation of the terrain accessibility and obstacle occlusion of the feasible passage areas is carried out to screen out candidate passage areas that meet the requirements of terrain adaptability and obstacle avoidance safety.

[0086] Within the candidate passage area, plan local trajectory segments that connect adjacent target points in the target path framework, ensuring that the local trajectory segments simultaneously meet the requirements of terrain adaptability and obstacle avoidance safety;

[0087] All local trajectory segments are spliced ​​and smoothed according to the order of target points in the target path framework to generate the initial navigation trajectory.

[0088] like Figure 2 As shown, the method includes:

[0089] When constructing the target path framework for an inspection task based on fiber optic cable path information, it is necessary to align the geographical orientation of the fiber optic cable with the map coordinate system and discretize the fiber optic cable path into a series of ordered target point sequences. Specifically, path nodes are extracted along the fiber optic cable path at preset sampling intervals. Each node carries spatial coordinates and fiber optic cable orientation angle information, thus forming a target path framework describing the overall direction of the inspection task. This framework does not directly serve as the robot's execution trajectory but rather as a macroscopic guide for trajectory planning, ensuring that the final generated navigation trajectory remains consistent with the fiber optic cable path in its overall direction and does not deviate from the inspection task objective. The spacing between adjacent target points in the target path framework is dynamically adjusted according to the complexity of the terrain. In flat terrain areas, the sampling interval can be appropriately increased to reduce the amount of planning calculations, while in areas with drastic terrain undulations or dense obstacles, the sampling interval is reduced to improve the precision of path planning.

[0090] Guided by the target path framework, the process of determining feasible traversable areas based on the set of environmental constraints requires a grid-by-grid screening of the map grid covered by the terrain-obstacle coupling model. For each grid cell, it is determined whether its joint risk coefficient exceeds a preset feasible traversable threshold. Grids with joint risk coefficients below the threshold are marked as candidate feasible grids, while those exceeding the threshold are marked as prohibited grids. The set of feasible grids constitutes the initial feasible traversable area. Based on this, the feasible traversable area is further comprehensively evaluated using joint constraint relationships, focusing on two dimensions of constraints: terrain traversability constraints require that the slope of the grids traversed by the path does not exceed the robot's tolerance range, and that the transition slope between adjacent grids does not change too drastically, to avoid robot attitude instability in continuously undulating terrain; obstacle occlusion constraints require that the path maintain a sufficient safe distance from obstacles, and consider the impact of obstacle occlusion diffusion areas to prevent the robot from entering the occlusion influence range of obstacles. After comprehensive evaluation, candidate traversable areas that simultaneously satisfy both types of constraints are selected from the initial feasible traversable area for subsequent local trajectory planning.

[0091] When planning local trajectory segments connecting adjacent target points within a candidate passable region, a method combining graph search and cost function optimization is adopted. The grid within the candidate passable region is constructed as a weighted directed graph, where nodes correspond to grid center points, and edges between nodes represent passable connections between adjacent grids. The cost of an edge is determined by multiple components: Let... To from grid nodes to adjacent grid nodes Local path cost, The cost component is the Euclidean distance between the two nodes. The cost component is the terrain accessibility factor. If the cost is for obstacle occlusion, then the local path cost is calculated as follows:

[0092] ;

[0093] in For terrain cost weighting, These are obstacle cost weights, both preset positive real numbers, used to balance the relationship between path length, terrain adaptability, and obstacle avoidance safety. Terrain traversability cost component. According to the node With nodes The slope characteristics and elevation change range of the corresponding grid are calculated, and the greater the slope and the more drastic the elevation change, the higher the cost; obstacle occlusion cost component. According to the node The distance between the current grid and the nearest obstacle is calculated, with closer distances resulting in higher costs. Nodes within the obstacle's occlusion radius have their costs maximized to achieve effective avoidance. Based on this cost function, a heuristic search algorithm is used within the candidate passage area to search for the optimal path from the current target point to the next target point, obtaining local trajectory segments. The heuristic function uses the Euclidean distance from the current node to the target point as an estimate, guiding the search direction towards the target point and improving search efficiency.

[0094] For the planning of each local trajectory segment, it is also necessary to further verify the joint constraint satisfaction of each node on the trajectory segment. Specifically, for each node sequence of the searched path, it is checked whether the joint risk coefficient of its grid satisfies the preset constraints, and at the same time, it is checked whether the transition slope between adjacent nodes is within the robot's tolerance range. If there is a node on a certain path that does not meet the constraints, a higher penalty cost is imposed on the grid where the node is located, and the local search is retried to bypass the area until a local trajectory segment that fully meets the requirements of terrain adaptability and obstacle avoidance safety is obtained.

[0095] After all local trajectory segments are planned, they are sequentially spliced ​​together according to the order of the target points in the target path framework to form a continuous path node sequence covering the entire inspection task. Since there are abrupt changes in direction or curvature discontinuities at the splicing points of each local trajectory segment, the spliced ​​path needs to be smoothed. The smoothing process uses piecewise cubic spline interpolation, with the spatial coordinates of the path nodes as interpolation constraint points, generating smooth curve segments with continuous curvature between adjacent nodes to ensure that the robot does not make abrupt turns during trajectory execution. During the smoothing process, it is also necessary to ensure that the smoothed trajectory remains within the candidate passable area and does not encroach on restricted areas or obstacle-occluded areas due to smoothing interpolation. Therefore, the sampling points on the smoothed trajectory are constrained and verified point by point. If the smoothed curve deviates from the candidate passable area, constraint correction is introduced at the corresponding position to pull the deviated point back to the nearest feasible position within the candidate passable area, and smoothing interpolation is re-executed until the entire smoothed trajectory meets the constraint requirements.

[0096] After smoothing, a global consistency check is performed on the generated initial navigation trajectory. This check includes verifying whether the trajectory's start and end points are aligned with the first and last target points of the target path framework, whether the trajectory covers the vicinity of all intermediate target points, whether the maximum slope of each segment of the trajectory is within the robot's rated mobility, and whether the minimum distance between the trajectory and all known obstacles meets safety clearance requirements. Once the global check passes, this trajectory is output as the initial navigation trajectory for subsequent robot execution and feedback correction. If the global check identifies a segment that does not meet the requirements, local trajectory planning is re-triggered for that segment, the corresponding segment's local trajectory is updated, and smoothing and checking are performed again until the entire initial navigation trajectory passes the global consistency check.

[0097] In one optional implementation, the attitude deviation information and the environmental change information are used as feedback inputs to incrementally correct the spatial correlation mapping between terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model, updating the joint constraint relationship between terrain accessibility and obstacle occlusion, resulting in an updated set of environmental constraints including:

[0098] The deviation between the actual representation of terrain undulation features and the predicted representation of terrain undulation features in the spatial association mapping is quantified based on attitude deviation information. The deviation between the actual representation of obstacle distribution features and the predicted representation of obstacle distribution features in the spatial association mapping is quantified based on environmental change information. Each deviation is used as a correction increment to locally adjust the inter-grid association relationship in the spatial association mapping, resulting in the corrected spatial association mapping.

[0099] Based on the adjusted inter-mesh correlation in the modified spatial correlation mapping, the constraints of terrain slope change on robot mobility and the limitation of obstacle occlusion range on passable area are re-analyzed, and the joint constraint relationship between terrain mobility and obstacle occlusion is updated.

[0100] The accessibility and traversability of the spatial grid are reassessed based on the updated joint constraint relationships. The constraint attributes of the corresponding spatial grid are updated, and all updated constraint attributes are summarized to obtain the updated set of environmental constraints.

[0101] During the execution of the initial navigation trajectory, the robot collects real-time data from its own attitude sensors and the surrounding environment. These two types of data serve as the actual observation basis for terrain undulation features and obstacle distribution features, respectively. Attitude deviation information comes from the output of attitude sensing devices such as the inertial measurement unit (IMU) and tilt sensors, reflecting the robot's pitch angle, roll angle, and actual slope sensing values ​​as it moves on the actual terrain. Environmental change information comes from the real-time scanning results of sensing devices such as lidar and depth cameras, reflecting changes in the position, shape, or occlusion range of obstacles relative to the initial modeling. Comparing these two types of information with the predicted representations in the terrain-obstacle coupling model is a prerequisite for incremental correction.

[0102] The deviation quantification process for terrain undulation features uses attitude deviation information as the core input. In the initial modeling stage, the predicted terrain slope information for each grid is recorded in the spatial association mapping. During actual trajectory execution, when the robot passes through a grid area, the attitude sensor outputs the actual slope perception value for that area in real time. The difference between the actual slope perception value and the predicted slope value is defined as the terrain undulation deviation, denoted as […]. ,in This indicates the terrain slope prediction error for the current grid. When Exceeding the preset terrain perception error tolerance threshold When this occurs, a local correction process is triggered for the grid and its neighboring grids. The value is determined based on the slope response sensitivity of the robot's mechanical structure and the accuracy of the sensors, and is usually set as a certain percentage of the robot's rated traversable slope range.

[0103] The process of quantifying the deviation of obstacle distribution features takes environmental change information as input. In the initial construction stage of spatial association mapping, the position and occlusion range of obstacles have been marked and associated with the corresponding grid nodes. In actual execution, LiDAR or depth camera scans the area in front in real time. If the actual position or outer size of the detected obstacle differs from the initial record, the difference is quantified as the obstacle distribution deviation. ,in This indicates the positional or dimensional deviation between the measured and predicted distribution of obstacles. Exceeding the obstacle perception error tolerance threshold When this occurs, it triggers a correction of the obstacle occlusion range in the affected grid area. The value is determined by taking into account both the resolution of the sensing device and the safety distance requirements of the robot.

[0104] Terrain undulation deviation Deviation from obstacle distribution As a correction increment, the inter-grid relationships in the spatial correlation mapping are locally adjusted. Specifically, for the target grid that triggers the correction and its neighboring grids within a certain range, based on... The correlation parameters of the transition slope between adjacent grids are updated to make the corrected inter-grid slope correlation values ​​closer to the actual terrain conditions; based on The correlation parameters of the obstacle occlusion influence range are updated, and the difference between the actual occlusion diffusion range and the initial predicted value is superimposed on the occlusion attribute of the corresponding mesh. This local adjustment method only modifies the mesh region affected by the deviation, without reconstructing the global correlation mapping, thus ensuring the computational efficiency of the correction process and meeting the time constraints of real-time control. The corrected spatial correlation mapping retains the original global structure while significantly improving local accuracy.

[0105] Based on the corrected spatial correlation mapping, the constraints of terrain slope changes on robot mobility are re-analyzed. For the corrected grid region, the terrain mobility of the region is re-evaluated according to the updated transition slope values ​​between adjacent grids. If the corrected transition slope value exceeds the robot's navigable slope range, the corresponding grid is marked as impassable; if the corrected transition slope value is within the safe navigable range, the terrain mobility rating of the grid is maintained or improved. Simultaneously, based on the updated obstacle occlusion range, the set of grids affected by occlusion is re-determined, and the obstacle occlusion constraint values ​​are recalculated for the grids within the occlusion range, adding the occlusion effect to the constraint attributes of the corresponding grids. The updates to both terrain mobility constraints and obstacle occlusion constraints are completed within the local correction range, ensuring the real-time update capability of the joint constraint relationship.

[0106] The updated terrain accessibility constraints and obstacle occlusion constraints together form a new joint constraint relationship. The update of this joint constraint relationship follows the same coupling logic as the initial modeling phase: changes in terrain accessibility affect the effective spatial range of obstacle occlusion, and changes in obstacle occlusion range, in turn, limit the actual reachable area of ​​terrain accessibility. Within the local correction region, the coupling relationship between the two types of constraints is adjusted synchronously with the update of their respective parameters, ensuring the internal consistency of the joint constraint relationship. For mesh regions that have not triggered corrections, their joint constraint relationship remains unchanged, avoiding unnecessary computational overhead.

[0107] Based on the updated joint constraints, the accessibility and traversability of each spatial grid within the corrected area are reassessed. The accessibility assessment comprehensively considers the updated terrain slope constraints and obstacle occlusion constraints, assigning a new accessibility score to each grid. The accessibility assessment, based on the accessibility score, combines the path connectivity between the robot's current position and the target position to determine whether each grid meets the accessibility requirements under the new constraints. For grids whose accessibility or accessibility has changed, their constraint attributes are updated, including access status markers, risk scores, and accessibility indicators. For grids that have not changed, the original constraint attributes are retained, and no duplicate calculations are performed.

[0108] By aggregating all updated constraint attributes, an updated environmental constraint set is obtained. This set uses a grid as the basic unit and records the latest accessibility safety score, accessibility flag, and joint risk rating for each grid, providing a complete constraint basis for the regeneration of subsequent navigation trajectories. The updated environmental constraint set maintains the same data structure as the initial environmental constraint set, ensuring that the trajectory planning module can call the constraint data through the same interface without requiring additional adaptation to the planning algorithm. As the robot continues to advance, the above incremental correction process is executed cyclically in each control cycle. The environmental constraint set continuously approaches the real environment state with the accumulation of actual perception data, thereby supporting the continuous optimization and operation of the navigation control closed loop.

[0109] In one optional implementation, the navigation trajectory is regenerated based on the updated set of environmental constraints, and the robot is controlled to execute it. The regenerated navigation trajectory is then used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop, including:

[0110] Based on the updated set of environmental constraints and optical cable path information, the navigation trajectory is regenerated under the constraints of the updated joint constraint relationship. The regenerated navigation trajectory is then compared with the initial navigation trajectory before execution to identify the trajectory adjustment area.

[0111] The robot is controlled to execute according to the navigation trajectory. Within the trajectory adjustment area, robot posture deviation information and environmental change information are collected. The navigation trajectory is used as a new initial navigation trajectory, and the posture deviation information and environmental change information are used as new feedback inputs to trigger the correction of the terrain-obstacle coupling model.

[0112] The updated set of environmental constraints is obtained through correction. The navigation trajectory is regenerated based on the updated set of environmental constraints. The trajectory adjustment area identification and key data acquisition process is repeated to form a continuously optimized navigation control closed loop.

[0113] After incrementally refining the terrain-obstacle coupling model and obtaining the updated set of environmental constraints, this set of constraints, along with the current fiber optic cable path information, is input into the trajectory planning module. Under the updated joint constraint relationships, the navigation trajectory is regenerated. The regenerated navigation trajectory comprehensively reflects the latest perceived terrain undulations and obstacle distribution characteristics. Its path node sequence exhibits local spatial deviations from the initial navigation trajectory before execution. To efficiently identify these deviation areas, the regenerated navigation trajectory is compared segment by segment with the initial navigation trajectory before execution in a grid coordinate system. The spatial offset of the corresponding path segment is calculated. When the offset of a path segment exceeds a preset trajectory adjustment threshold, the area containing that path segment is marked as a trajectory adjustment area. The identification results of the trajectory adjustment areas are stored in the form of area boundary indexes for triggering judgments in subsequent key data acquisition processes.

[0114] The identification of trajectory adjustment regions employs a segmented comparison strategy. Both the initial navigation trajectory and the regenerated navigation trajectory are discretized into equally spaced path node sequences, and the 3D Euclidean distance deviation is calculated for each pair of corresponding nodes. When the deviation of several consecutive nodes exceeds a threshold, the entire consecutive segment is designated as a trajectory adjustment region, avoiding misjudgments caused by single-point noise. For path segments with small deviations, the trajectory adjustment amplitude is considered to be within the robot's motion tolerance range, and therefore does not require focused acquisition. This allows limited sensing resources to be concentrated on areas of significant change. This differentiated resource allocation method can reduce the computational and communication burden caused by repeated global acquisition while ensuring navigation safety.

[0115] When controlling the robot to execute the regenerated navigation trajectory, a regular data acquisition frequency is maintained outside the trajectory adjustment area to ensure basic attitude deviation monitoring and environmental perception. When the robot enters the trajectory adjustment area, a focused acquisition mode is triggered: the attitude data sampling frequency of the inertial measurement unit is increased, and the scanning density of the LiDAR or depth camera is increased to achieve high-density perception of the terrain undulation features and obstacle distribution features in this area. The robot attitude deviation information and environmental change information obtained from the focused acquisition are fed into the feedback data stream in real time, serving as new feedback input to trigger the correction of the terrain-obstacle coupling model. Since the trajectory adjustment area is precisely the region where the difference between the initial prediction and the actual environment is most significant, the perception data acquired intensively in this area has the highest information gain for correcting the spatial correlation mapping between terrain undulation features and obstacle distribution features, enabling accurate correction of the largest part of the error in the coupling model with minimal data volume.

[0116] Using the currently executed navigation trajectory as the new initial navigation trajectory means that each closed-loop iteration is based on the latest planning result, rather than always referencing the original trajectory. This design ensures that the identification of the trajectory adjustment region is always based on the difference between the planning results of two adjacent iterations, rather than the cumulative deviation, thus avoiding the problem of the comparison benchmark becoming invalid as the number of iterations increases. In each iteration, the range of the trajectory adjustment region gradually shrinks as the accuracy of the coupled model improves, eventually converging within the robot's motion tolerance range, indicating that the set of environmental constraints has fully reflected the actual environmental state and the navigation trajectory tends to stabilize.

[0117] By acquiring new feedback inputs through focused data collection, the terrain undulation characteristics and obstacle distribution characteristics of the corresponding grid regions in the terrain-obstacle coupling model are incrementally corrected. The joint constraint relationship between terrain accessibility and obstacle occlusion is updated, resulting in a new set of updated environmental constraints. Based on this constraint set, the navigation trajectory is regenerated and compared with the initial trajectory to identify the new trajectory adjustment area. If the newly identified trajectory adjustment area is significantly smaller than the previous one, it indicates that the closed-loop optimization process has converged well. If the trajectory adjustment area expands, it indicates that the robot has entered a new complex terrain section or encountered new dynamic obstacles during execution, requiring further focused data collection within the new trajectory adjustment area to drive further correction of the coupling model.

[0118] The entire continuously optimized navigation control closed loop operates continuously throughout the robot's navigation task, independent of external trigger signals, using the identification results of trajectory adjustment areas as intrinsic driving signals. In complex terrain fiber optic cable inspection scenarios, terrain undulations and obstacle distribution often change continuously as the path extends. The robot encounters environmental features that the initial model could not accurately predict in each newly entered path region. By organically linking trajectory comparison, key data acquisition, model correction, and trajectory replanning to form an adaptive closed loop, the navigation control capability continuously improves as the task progresses. Each iteration of the closed loop makes the terrain-obstacle coupling model more closely resemble the actual environment, making the navigation trajectory more consistent with the dual requirements of terrain adaptability and obstacle avoidance safety, ultimately achieving stable, efficient, and safe autonomous navigation control in complex and unknown terrain.

[0119] In the collaborative mechanism of trajectory adjustment area identification and key data acquisition, the timing matching problem between robot speed and perception delay also needs to be considered. When the robot passes through the trajectory adjustment area at a high speed, the key data acquisition mode needs to be triggered a certain distance before entering the boundary of the area to ensure that sufficient high-density perception data has been accumulated when the robot actually reaches the key area. The advance triggering distance is dynamically calculated based on the robot's current speed and perception processing delay to ensure the timeliness and completeness of the feedback data, thereby maintaining the stability and response speed of closed-loop control.

[0120] In one optional implementation, based on the updated set of environmental constraints and the optical cable path information, a navigation trajectory is regenerated under the constraints of the updated joint constraint relationship. The regenerated navigation trajectory is then compared with the initial navigation trajectory before execution, and the trajectory adjustment area is identified, including:

[0121] Based on the updated set of environmental constraints and optical cable path information, local trajectory segments connecting adjacent target points are replanned under the constraints of the updated joint constraint relationship, and the local trajectory segments are spliced ​​together to generate a navigation trajectory.

[0122] Extract the trajectory segments between the corresponding target points in the navigation trajectory and the initial navigation trajectory, calculate the spatial position offset and path shape difference between the corresponding trajectory segments, determine the spatial region where the trajectory is adjusted based on the spatial position offset and the path shape difference, and mark the spatial region as the trajectory adjustment region.

[0123] After obtaining the updated set of environmental constraints and fiber optic cable path information, it is necessary to regenerate the complete navigation trajectory under the updated joint constraint relationship and systematically compare it with the initial navigation trajectory before execution to identify the spatial areas where substantial adjustments have been made, thus providing a precise basis for local replanning for subsequent execution control.

[0124] The process of regenerating the navigation trajectory unfolds using local trajectory segment planning as the basic unit. A series of target point sequences distributed along the fiber optic cable route are predefined in the fiber optic cable path information; these target points constitute the skeleton constraints of the navigation trajectory. For each path segment between adjacent target points, local trajectory segments are independently replanned under the updated joint constraints. During local trajectory segment planning, the joint risk coefficient of each grid node in the updated environmental constraint set is used as the basis for the passage cost. Under the premise of satisfying terrain accessibility constraints and obstacle occlusion constraints, the optimal local path connecting adjacent target points is searched. After each local trajectory segment is planned, all local trajectory segments are sequentially spliced ​​according to the order of the target point sequence to form a new navigation trajectory covering the entire fiber optic cable inspection path. During the splicing process, the connection positions of adjacent local trajectory segments at target points are smoothly transitioned to avoid abrupt changes in direction caused by segmented planning affecting the robot's actual travel stability.

[0125] After generating the new navigation trajectory, it is compared and analyzed segment by segment with the initial navigation trajectory stored before execution. The comparison operation uses the corresponding target points as the alignment benchmark to extract the corresponding trajectory segments connecting the same pair of adjacent target points in the new navigation trajectory and the initial navigation trajectory, forming a set of paired trajectory segment sequences. For each pair of corresponding trajectory segments, the spatial position offset and path morphology difference between the two trajectory segments are calculated as quantitative criteria to determine whether the trajectory segment has undergone substantial adjustment.

[0126] The spatial position offset is calculated as follows: Several discrete corresponding points are sampled on the corresponding trajectory segment using an equal arc length parameterization method. The three-dimensional Euclidean distance between each sampled point on the new trajectory and its corresponding parameter position point on the initial trajectory is calculated. The maximum value of all sampled point distances is taken as the spatial position offset of the trajectory segment. This reflects the maximum deviation between two trajectory segments in spatial location. The path morphology difference is measured based on the overall morphological characteristics of the trajectory, calculating the difference in path length between corresponding trajectory segments. and the average angular deviation between the direction vector sequences of the two trajectories It comprehensively reflects the changes in the curvature and direction of the trajectory.

[0127] Based on the two types of metrics mentioned above, a spatial location offset threshold is set. Path length difference threshold With direction deviation threshold For each pair of corresponding trajectory segments, a judgment is made: if a trajectory segment satisfies ,or ,or If any of the conditions in the above conditions are met, the trajectory segment is considered to have undergone substantial adjustment, and the spatial area covered by the trajectory segment is marked as the trajectory adjustment region. The spatial extent of the trajectory adjustment region is determined by the three-dimensional bounding box between the corresponding target points. That is, the three-dimensional coordinates of all sampling points on the new trajectory segment and the initial trajectory segment are taken, and the minimum and maximum values ​​are taken in the three coordinate axis directions respectively to form an axis-aligned bounding box that can completely contain the two trajectory segments. This box represents the spatial boundary of the trajectory adjustment region.

[0128] In real-world complex terrain inspection scenarios, multiple areas of terrain undulation and obstacle distribution exist along the fiber optic cable route, causing simultaneous adjustments to multiple trajectory segments. When multiple adjacent trajectory segments are marked as trajectory adjustment areas, the spatial bounding boxes of these adjacent adjustment areas are merged and expanded to form continuous trajectory adjustment areas. This avoids frequent area switching due to small gaps between adjacent areas, improving the consistency of adjustment responses in subsequent execution control. For isolated single-segment trajectory adjustment areas, their independent marking is maintained to enable precise location and targeted handling of sudden obstacles or terrain anomalies.

[0129] The identification results of trajectory adjustment areas will serve as a key input for subsequent robot control. As the robot travels along its navigation trajectory, when its current position enters the spatial range of a trajectory adjustment area, a local trajectory segment switching command is triggered, switching the robot's target from the corresponding segment on the initial navigation trajectory to a newly generated trajectory segment. This enables dynamic response to environmental changes and terrain undulations. For trajectory segments not marked as trajectory adjustment areas, the robot continues along the initial navigation trajectory without switching, thus ensuring navigation safety while reducing unnecessary trajectory switching operations and mitigating the impact of frequent control command fluctuations on robot stability.

[0130] Through the complete process of local trajectory segment replanning, segment-by-segment comparative analysis, and trajectory adjustment area marking, refined management of dynamic updates to navigation trajectories in complex terrain fiber optic cable inspection scenarios is achieved. This mechanism ensures that after each environmental constraint update, only targeted adjustments are needed to the locally changed trajectory areas, rather than a complete replanning of the global trajectory. This effectively reduces the computational overhead of real-time navigation control while ensuring the accuracy and timeliness of trajectory adjustments, providing reliable trajectory management support for the robot to continuously and stably complete fiber optic cable inspection tasks in complex terrain.

[0131] A second aspect of the present invention provides an autonomous navigation and obstacle avoidance control system for a fiber optic cable inspection robot in complex terrain, comprising:

[0132] The information acquisition unit is used to acquire the robot's current position information, surrounding environment perception data, and optical cable path information;

[0133] The coupling modeling unit is used to construct a terrain-obstacle coupling model based on the surrounding environment perception data. The terrain-obstacle coupling model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing passage safety and accessibility.

[0134] The trajectory generation unit is used to generate an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety under the constraints of the joint constraint relationship, based on the set of environmental constraints and the optical cable path information.

[0135] The feedback correction unit is used to collect robot posture deviation information and environmental change information in real time during the execution of the initial navigation trajectory, use the posture deviation information and environmental change information as feedback input, perform incremental correction on the spatial correlation mapping of terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model, update the joint constraint relationship of terrain accessibility and obstacle occlusion, and obtain the updated environmental constraint set.

[0136] The closed-loop control unit is used to regenerate the navigation trajectory based on the updated set of environmental constraints and control the robot to execute it. The regenerated navigation trajectory is used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop.

[0137] A third aspect of the present invention provides an electronic device, comprising:

[0138] processor;

[0139] Memory used to store processor-executable instructions;

[0140] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0141] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0142] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0143] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An autonomous navigation and obstacle avoidance control method for a fiber optic cable inspection robot in complex terrain, characterized in that, include: Acquire the robot's current location information, surrounding environment perception data, and fiber optic cable path information; Based on the surrounding environment perception data, a terrain-obstacle coupling model is constructed. The terrain-obstacle coupling model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing passage safety and accessibility. Based on the set of environmental constraints and the optical cable path information, an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety is generated under the constraints of the joint constraint relationship. During the execution of the initial navigation trajectory, robot posture deviation information and environmental change information are collected in real time. The posture deviation information and environmental change information are used as feedback inputs to incrementally correct the spatial correlation mapping between terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model, update the joint constraint relationship between terrain accessibility and obstacle occlusion, and obtain the updated environmental constraint set. Based on the updated set of environmental constraints, a new navigation trajectory is generated and the robot is controlled to execute it. The newly generated navigation trajectory is then used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop.

2. The method according to claim 1, characterized in that, Based on the surrounding environment perception data, a terrain-obstacle coupling model is constructed. This model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing accessibility and reachability, including: The surrounding environment perception data is spatially gridded, and the terrain undulation features and obstacle distribution features are extracted in each spatial grid. Based on the terrain continuity and obstacle spatial extension between adjacent spatial grids, the relationship between grids is established to form a spatial relationship mapping. Based on the inter-mesh association relationship in the spatial association mapping, by analyzing the constraints of terrain slope change on robot mobility and the limitation of obstacle occlusion range on passable area, a joint constraint relationship between terrain mobility and obstacle occlusion is established, and the spatial association mapping and the joint constraint relationship together constitute a terrain-obstacle coupling model. The accessibility and accessibility of each spatial grid are evaluated based on the joint constraint relationship in the terrain-obstacle coupling model. The evaluation results are used as the constraint attributes of the corresponding spatial grid, and the constraint attributes of all spatial grids are summarized to generate an environmental constraint set.

3. The method according to claim 2, characterized in that, Based on the inter-mesh relationships in the spatial association mapping, by analyzing the constraints of terrain slope changes on robot mobility and the limitations of obstacle occlusion range on passable areas, a joint constraint relationship between terrain mobility and obstacle occlusion is established, including: Based on the inter-grid relationship, the transmission path of terrain slope change between adjacent spatial grids is identified, and the cumulative constraint effect of terrain slope change on robot mobility is determined according to the transmission path, thus obtaining the terrain mobility constraint description. Based on the inter-mesh relationship, the diffusion area of ​​obstacle occlusion between adjacent spatial meshes is identified, and the coverage limit of obstacle occlusion on the passable area is determined according to the diffusion area, thus obtaining the obstacle occlusion constraint description. The terrain accessibility constraint description and the obstacle occlusion constraint description are coupled and fused to establish a joint constraint relationship between terrain accessibility and obstacle occlusion.

4. The method according to claim 1, characterized in that, Based on the set of environmental constraints and the optical cable path information, an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety is generated under the constraints of the joint constraint relationship, including: Based on the optical cable path information, a target path framework for the inspection task is constructed. Under the guidance of the target path framework, feasible passage areas are determined according to the set of environmental constraints. Based on the joint constraint relationship, a comprehensive evaluation of the terrain accessibility and obstacle occlusion of the feasible passage areas is carried out to screen out candidate passage areas that meet the requirements of terrain adaptability and obstacle avoidance safety. Within the candidate passage area, plan local trajectory segments that connect adjacent target points in the target path framework, ensuring that the local trajectory segments simultaneously meet the requirements of terrain adaptability and obstacle avoidance safety; All local trajectory segments are spliced ​​and smoothed according to the order of target points in the target path framework to generate the initial navigation trajectory.

5. The method according to claim 1, characterized in that, Using the attitude deviation information and the environmental change information as feedback inputs, the spatial correlation mapping between terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model is incrementally corrected, and the joint constraint relationship between terrain accessibility and obstacle occlusion is updated to obtain the updated set of environmental constraints, including: The deviation between the actual representation of terrain undulation features and the predicted representation of terrain undulation features in the spatial association mapping is quantified based on attitude deviation information. The deviation between the actual representation of obstacle distribution features and the predicted representation of obstacle distribution features in the spatial association mapping is quantified based on environmental change information. Each deviation is used as a correction increment to locally adjust the inter-grid association relationship in the spatial association mapping, resulting in the corrected spatial association mapping. Based on the adjusted inter-mesh correlation in the modified spatial correlation mapping, the constraints of terrain slope change on robot mobility and the limitation of obstacle occlusion range on passable area are re-analyzed, and the joint constraint relationship between terrain mobility and obstacle occlusion is updated. The accessibility and traversability of the spatial grid are reassessed based on the updated joint constraint relationships. The constraint attributes of the corresponding spatial grid are updated, and all updated constraint attributes are summarized to obtain the updated set of environmental constraints.

6. The method according to claim 1, characterized in that, Based on the updated set of environmental constraints, a new navigation trajectory is generated and the robot is controlled to execute it. The regenerated navigation trajectory is then used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop, including: Based on the updated set of environmental constraints and optical cable path information, the navigation trajectory is regenerated under the constraints of the updated joint constraint relationship. The regenerated navigation trajectory is then compared with the initial navigation trajectory before execution to identify the trajectory adjustment area. The robot is controlled to execute according to the navigation trajectory. Within the trajectory adjustment area, robot posture deviation information and environmental change information are collected. The navigation trajectory is used as a new initial navigation trajectory, and the posture deviation information and environmental change information are used as new feedback inputs to trigger the correction of the terrain-obstacle coupling model. The updated set of environmental constraints is obtained through correction. The navigation trajectory is regenerated based on the updated set of environmental constraints. The trajectory adjustment area identification and key data acquisition process is repeated to form a continuously optimized navigation control closed loop.

7. The method according to claim 6, characterized in that, Based on the updated set of environmental constraints and fiber optic cable path information, a new navigation trajectory is generated under the updated joint constraint relationships. The regenerated navigation trajectory is then compared with the initial navigation trajectory before execution, and the trajectory adjustment area is identified, including: Based on the updated set of environmental constraints and optical cable path information, local trajectory segments connecting adjacent target points are replanned under the constraints of the updated joint constraint relationship, and the local trajectory segments are spliced ​​together to generate a navigation trajectory. Extract the trajectory segments between the corresponding target points in the navigation trajectory and the initial navigation trajectory, calculate the spatial position offset and path shape difference between the corresponding trajectory segments, determine the spatial region where the trajectory is adjusted based on the spatial position offset and the path shape difference, and mark the spatial region as the trajectory adjustment region.

8. An autonomous navigation and obstacle avoidance control system for a fiber optic cable inspection robot in complex terrain, used to implement the method as described in any one of claims 1-7, characterized in that, include: The information acquisition unit is used to acquire the robot's current position information, surrounding environment perception data, and optical cable path information; The coupling modeling unit is used to construct a terrain-obstacle coupling model based on the surrounding environment perception data. The terrain-obstacle coupling model establishes a joint constraint relationship between terrain undulation features and obstacle distribution features by spatially associating and mapping them, thereby generating a set of environmental constraints describing passage safety and accessibility. The trajectory generation unit is used to generate an initial navigation trajectory that satisfies terrain adaptability and obstacle avoidance safety under the constraints of the joint constraint relationship, based on the set of environmental constraints and the optical cable path information. The feedback correction unit is used to collect robot posture deviation information and environmental change information in real time during the execution of the initial navigation trajectory, use the posture deviation information and environmental change information as feedback input, perform incremental correction on the spatial correlation mapping of terrain undulation features and obstacle distribution features in the terrain-obstacle coupling model, update the joint constraint relationship of terrain accessibility and obstacle occlusion, and obtain the updated environmental constraint set. The closed-loop control unit is used to regenerate the navigation trajectory based on the updated set of environmental constraints and control the robot to execute it. The regenerated navigation trajectory is used as the new initial navigation trajectory to continue the feedback correction process, forming a continuously optimized navigation control closed loop.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images