A tunnel path planning method and system of clearance corridor and elastic geodesic refinement

Abstract

The present application relates to the technical field of tunnel and underground space robot autonomous navigation, and provides a tunnel path planning method and system of clearance corridor and elastic geodesic refinement, which comprises the following steps: calculating the Euclidean distance to the nearest obstacle or wall as the clearance for the structural terrain, generating the direction-independent cost, constructing the dual metric, running the Riemannian geometry fast marching and backtracking geodesic under the condition of given start and end point pixel coordinates to obtain a coarse path, calculating the distance from the pixel to the coarse path after gridding, obtaining the widening radius along the coarse path with the clearance, and performing the nearest neighbor propagation to the whole domain to obtain a radius map, defining the corridor, propagating the approximate curvature field from the coarse path corner, calculating the score, carrying out the segmented linear penalty after partitioning through thresholding, updating the direction-dependent total cost, and taking the reciprocal of the cost after re-labeling according to the partition as a speed field, and obtaining the final path through a solver. The present application solves the problem of simultaneously meeting the requirements of safe remote obstacle, controlled curvature and efficient solution in engineering.

Description

Technical Field

[0001] This invention belongs to the field of autonomous navigation technology for tunnel and underground space robots, and particularly relates to a tunnel path planning method and system with clear corridor and flexible geodesy refinement. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Tunnel scenarios are characterized by strong boundary constraints, narrow passable zones, and large variations in terrain slope and roughness.

[0004] Traditional global search methods (such as A*, Dijkstra, and fast travel methods) are costly on high-resolution maps and are prone to "wall-hugging" solutions; using only the cost field with invariant direction is difficult to express direction-sensitive risks such as uphill or side slope.

[0005] The existing fixed radius corridor method cannot adaptively shrink or widen according to the "clearance from the wall", resulting in insufficient safety-efficiency trade-off; while the SE(2) elastic geodesy can constrain curvature, but if it is directly refined in the whole domain, the computational cost and local extremum problem are prominent. Summary of the Invention

[0006] To address the technical problems mentioned above, this invention provides a tunnel path planning method and system with clearance corridor and flexible geodesy refinement. By considering only the location cost across the entire domain, a two-dimensional coarse geodesy path is obtained. Then, a corridor is generated around the coarse path using a variable radius driven by clearance. Finally, pitch, tilt penalties, and difficult / easy recalibration are superimposed on the corridor, and flexible geodesy is used to obtain the final path with continuous curvature. This solves the problem in engineering that requires simultaneous satisfaction of safe obstacle avoidance, curvature control, and efficient solution.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] The first aspect of the present invention provides a tunnel path planning method with clearance corridor and flexible geodesy refinement, comprising:

[0009] For the target tunnel, obtain a raster map of the structural terrain;

[0010] For structural terrain, calculate the Euclidean distance from each grid point to the nearest obstacle or wall to obtain the clearance distance field. Based on the clearance distance field, construct the wall penalty factor and combine it with the location features according to weights to generate the orientation-independent cost.

[0011] Based on the direction-independent cost, a dual metric is constructed. Given the starting and ending pixel coordinates, a Riemannian geometric fast traversal and backtracking geodesy are performed to obtain a coarse path. After rasterizing the coarse path, the distance from each pixel to the coarse path is calculated. The widening radius is obtained along the coarse path using the clearance, and the nearest neighbor is propagated to the entire domain to obtain a radius map. Based on the distance from each pixel to the coarse path and the radius map, a corridor is defined.

[0012] For the corridor, the approximate curvature field is obtained by propagating from the coarse path corner, and the score is calculated by combining the clearance. After being partitioned by thresholding, piecewise linear penalty is applied, and the direction-related total cost is updated based on the penalty. After being relabeled by partition, its reciprocal is used as the velocity field. The final path is obtained by solving the ordinary differential equation.

[0013] Furthermore, the location features include slope intensity and roughness.

[0014] Furthermore, the penalty for leaning against the wall is: Where r0 is the safety radius and r1 is the penalty band width. Let represent the net air volume, and t represent the normalized parameter of the net air volume. It represents a very small positive number.

[0015] Furthermore, the widening radius: , where r max and r min These are the maximum and minimum widths of the corridor. It is the airspace adaptation coefficient. Represents pixels Clear space This represents the constant offset in a linear mapping.

[0016] Furthermore, the corridor is ,in, For pixels, The distance from the pixel to the coarse path. This refers to the radius in the radius diagram.

[0017] Furthermore, the fraction is: ,in, For a curvature field, For purification, For slope, For roughness, , , and All are weights.

[0018] Furthermore, the piecewise linear penalty includes:

[0019] Punishment for looking up and down: ;

[0020] Lateral tilt penalty: ;

[0021] in, It is the roll penalty factor. It is the roll threshold. It is the uphill penalty coefficient. It is the uphill threshold. It is the coefficient for steep downhill slopes. It is the downhill threshold, pitch. Lateral tilt ,course The unit vector is , ( ) represents the gradient.

[0022] A second aspect of the present invention provides a tunnel path planning system with clearance corridor and flexible geodesy refinement, comprising:

[0023] The environment modeling module is configured to: for the target tunnel, acquire a raster map of the structural terrain;

[0024] The cost generation module is configured to: for structural terrain, calculate the Euclidean distance from each grid point to the nearest obstacle or wall to obtain the net distance field, construct a wall penalty factor based on the net distance field, and combine it with the location features according to weights to generate a direction-independent cost.

[0025] The corridor generation module is configured to: construct a dual metric based on orientation-independent cost; run Riemannian geometric fast travel and backtrack geodesy given the starting and ending pixel coordinates to obtain a coarse path; rasterize the coarse path; calculate the distance from each pixel to the coarse path; obtain a variable-width radius along the coarse path using the clearance; and propagate to the global area using the nearest neighbor to obtain a radius map; and define the corridor based on the distance from each pixel to the coarse path and the radius map.

[0026] The path planning module is configured as follows: For a corridor, an approximate curvature field is obtained by propagating from the coarse path corners, and a score is calculated by combining the clearance. After being partitioned by thresholding, a piecewise linear penalty is applied, and the direction-related total cost is updated based on the penalty. After being relabeled by partition, its reciprocal is used as the velocity field, and the final path is obtained through an ordinary differential equation solver.

[0027] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the tunnel path planning method for clearance corridor and flexible geodesy refinement as described above.

[0028] A fourth aspect of the present invention provides a computer device including a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, wherein the processor executes the program to implement the steps in the tunnel path planning method for clearance corridor and flexible geodesy refinement as described above.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] This invention obtains a coarse two-dimensional geodesic path by considering only the location cost across the entire domain, then generates a corridor around the coarse path with a variable radius driven by clearance, and finally superimposes pitch and roll penalties and difficult / easy recalibration in the corridor, using elastic geodesy to obtain a final path with continuous curvature, thus solving the problem in engineering that requires simultaneous satisfaction of safe obstacle avoidance, curvature control, and efficient solution. Attached Figure Description

[0031] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0032] Figure 1 This is a flowchart of a tunnel path planning method with clearance corridor and flexible geodesy refinement according to Embodiment 1 of the present invention;

[0033] Figure 2 This is a schematic diagram of the structure of a computer device according to Embodiment 4 of the present invention. Detailed Implementation

[0034] 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.

[0035] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0036] Example 1

[0037] This embodiment provides a tunnel path planning method with clear corridor and flexible geodesy refinement.

[0038] This embodiment provides a tunnel path planning method with clearance corridor and flexible geodesy refinement. Through a two-stage process that combines global coarse path, clearance adaptive corridor trimming, difficult / easy partitioning and SE(2) flexible geodesy refinement, it achieves an engineering path that is "safe, far from obstacles, with fewer bends and smoother flow, and faster calculation".

[0039] This embodiment provides a tunnel path planning method with clearance corridor and elastic geodesy refinement. In the global stage, a direction-independent cost field is constructed based on the location characteristics such as clearance, slope, and roughness. A two-dimensional coarse path is obtained through rapid travel using Riemann geometry. Then, a corridor is generated around the coarse path using a variable radius driven by clearance. Finally, pitch, tilt penalties and difficult / easy recalibration are superimposed in the corridor. The final path with curvature continuity is obtained using SE(2) elastic geodesy (Elastica). This solves the problem in engineering that requires simultaneous satisfaction of "safety and remote obstacle avoidance, curvature control and efficient solution".

[0040] This embodiment provides a tunnel path planning method based on clearance corridors and flexible geodesy refinement, proposing:

[0041] (1) Two-stage planning and unique path strategy: First, construct the Riemann metric in the whole domain at the cost of direction independence to obtain the coarse path as the skeleton, and then perform SE (2) elastic geodesic refinement only in the corridor; the endpoint is set without direction and the speed outside the corridor is set to near zero; the linkage mechanism of decoupling and constraint ensures that the curvature is controlled and away from the obstacle, while significantly shrinking the refinement domain, suppressing multiple solutions and "sticking to the wall", and obtaining a single smooth and stable passage path;

[0042] (2) Clearance-driven widening corridor: The Euclidean distance to the obstacle / boundary is used as the clearance metric. The clearance is sampled along the coarse path and mapped to the radius. A pixel-by-pixel continuous radius map is formed through nearest neighbor propagation. Based on this, an irregular corridor that "widens in open areas and narrows in cramped areas" is defined, and strong velocity suppression is applied outside the corridor. This method is different from fixed buffer or regular corridor. It can significantly reduce the computational domain while ensuring safety margin, and improve overall efficiency and robustness.

[0043] (3) Difficult / easy partitioning and zone relabeling based on coarse path curve prior: The curve prior is constructed from the discrete turning angle of the coarse path and propagated to the whole map. It is integrated with the clearance, slope and coarsness to form a score D. The threshold is used to obtain the hard / easy mask. In the corridor, the speed cost is relabeled by zone: the price is increased in the hard zone and decreased in the easy zone, forming interpretable soft guidance, reducing meaningless bends and boundary contact, and improving the path quality and the controllability of parameter tuning.

[0044] (4) Direction-sensitive penalty for attitude safety: Pitch and roll indices are calculated based on heading and terrain gradient. Piecewise linear penalties are applied to uphill, steep downhill and unfavorable roll directions and coupled with direction cost. This mechanism further constrains attitude safety in addition to geometric constraints. Combined with the aforementioned zone guidance and corridor constraints, the path achieves a better balance between accessibility, comfort and safety.

[0045] The above-mentioned aspects work together in a predetermined order: "overall rough path → clear corridor → zoned guidance → SE(2) refinement", achieving a comprehensive effect of fast calculation, stable movement, distance from obstacles, and smooth curves in engineering.

[0046] This embodiment provides a tunnel path planning method with clear corridor and flexible geodesy refinement, such as... Figure 1 As shown, it includes the following steps:

[0047] Step 1: Construct the environmental field.

[0048] For the target tunnel, obtain the height field or raster map of the structural terrain.

[0049] Alternatively, Gaussian superposition can be used to generate structural terrain (valleys and ridges on both sides), and multi-scale smooth white noise can be superimposed to form fBm (fractional Brownian motion)-like elevation undulations.

[0050] The gradient is calculated after slightly smoothing the terrain height field Z. ), slope strength and the steepest downhill direction ;

[0051] A binary mask is generated based on obstacles and boundaries, using the Euclidean distance from each grid point to the nearest obstacle / wall. Define the clearance and construct the penalty for being against the wall: Where r0 is the safety radius, r1 is the penalty zone width, and t represents the clearance distance. A normalized parameter linearly scaled to [0,1] in the interval [r0, r1] is used to construct the transition zone for the wall penalty. This represents extremely small positive numbers, used to avoid division by zero or numerical instability when r1-r0 is too small or equal to 0; its only function is to stabilize the numerical value.

[0052] Based on the wall penalty, and by normalizing the roughness and slope, we obtain... , The cost of forming a direction-independent process: And thus obtain the velocity map. To characterize accessibility, among which, , and It is a weighting factor that determines the importance of adjusting clearance, slope, and roughness.

[0053] Among them, the roughness field R 01The texture components r1, r2, and r3 at coarse, medium, and fine scales are obtained by multi-scale smoothing noise superposition: three independent Gaussian white noise fields are generated at the height field size, and convolved with two-dimensional Gaussian kernels with different standard deviations, respectively. These are then linearly superimposed with preset weights of 0.6, 0.3, and 0.1 to obtain rough_raw(x,y), and first normalized by the maximum absolute value. Finally, the smoothing is achieved using... The roughness is mapped to the [0,1] interval and used as the roughness intensity field.

[0054] Among them, the slope intensity field The gradient magnitude of the height field Z is obtained by: first, performing a slight Gaussian smoothing Zs=G*Z on the height field to suppress high-frequency noise; then, calculating the gradient components in the horizontal and vertical directions according to the grid step size. , Slope intensity is defined as gradient modulus: and using the same linear normalization Map S to [0,1] as the slope intensity field.

[0055] In direction-independent cost, roughness R 01 Slope S 01 Penalty C for leaning against the wall clear A linear combination with weights yields: And take F(x,y)=1 / c2d(x,y) as the velocity field to characterize accessibility.

[0056] Step 2: Generate a global coarse path.

[0057] Construct a dual metric based on direction-independent cost. Given the start and end pixel coordinates, a coarse path is obtained by performing a Riemannian geometric fast traversal (solving the eikonal equations using the Fast Marching method under Riemannian metric) and backtracking geodesy. , where I represents the identity matrix.

[0058] Step 3: Generate an adaptive corridor based on clearance.

[0059] coarse path Rasterize (convert to pixel-level routes) and calculate the distance from each pixel to the coarse path. The widening radius is obtained by using the clearance along the coarse path: And the radius graph is obtained by propagating to the nearest neighbor region. ; where r max and r min These are the maximum and minimum widths of the corridor. It is the airspace adaptation coefficient. This represents a constant offset in the linear mapping, used to adjust the linear relationship between "clearance distance and corridor radius": when When smaller, It may be negative, through comparison with The clamping mechanism ensures that the corridor radius is not too small; when As the net clearance increases, r increases approximately linearly with the net clearance, but the overall value still depends on [r]. min ,r max Limitation; physically, this can be understood as a "width offset / safety margin" parameter used to control the basic width of the corridor under limited clearance conditions;

[0060] Based on the distance and radius map from each pixel to the coarse path, define the corridor: Only pixels whose distance to the coarse path is less than or equal to the corresponding radius in the radius graph belong to the corridor, and a light morphological cleanup is performed. The velocity outside the corridor is close to zero to suppress out-of-bounds searches.

[0061] Step 4: Cutting and dividing the corridor into difficult / easy sections.

[0062] The existing direction-related cost field , and gradient Fields such as [field name] are clipped to the corridor bounding box; among them, [field name] is [field name]. Explanation: Specifically, the Euclidean distance C to the nearest obstacle. pix (x, y) represents the clearance; first pass through... The clearance is normalized to the [0,1] interval and smoothed using a Gaussian filter; then... Map it to [ξ] min ,ξ max ], where ξmin and ξmax are the minimum and maximum curvature scales in pixel units; the larger the clearance, the larger ξ, and the more the path tends to maintain a larger turning radius; the smaller the clearance, the smaller ξ, and the path can allow a smaller radius turn in narrow areas; but finally, in the SE(2) refinement, the median of ξ in the corridor is taken and converted into a physical scalar according to gs to enhance stability. First, there is the ξ(x,y) field, and then the median is taken in the ROI.

[0063] Within the corridor bounding box, an approximate curvature field is obtained by propagation from the coarse path corner. , and the clear sky ,slope Roughness Combined into a score : Thresholding is used to obtain hard / easy masks (i.e., pixels with a score D above the threshold are hard regions, and pixels below the threshold are easy regions), and small-scale closing operations and overlap band expansion are performed. It's the weight.

[0064] Step 5, Direction-sensitive penalty (pitch / roll).

[0065] Command heading unit vector Based on gradient, pitch is defined as: Gradient along the direction of travel (p positive = uphill, p negative = downhill); Lateral tilt The slope perpendicular to the direction of travel (the larger the |r|, the more severe the lateral tilt, such as when walking along the side of a slope).

[0066] Piecewise linear penalty is applied:

[0067] Punishment for looking up and down ,in, It is the uphill penalty coefficient. It is the uphill threshold (uphill sections exceeding the threshold are penalized). It is the coefficient for steep downhill slopes. It is the downhill threshold (a penalty is also added if the downhill slope is too steep).

[0068] Side tilt penalty ,in, It is the roll penalty factor. It is the roll threshold (a penalty is added if the roll exceeds the threshold);

[0069] Multiply the pitch and roll penalties by the existing direction-dependent cost field c(x,y,θ) to obtain the updated total direction-dependent cost. (It can also be used as the reciprocal of the velocity field within the corridor) to enhance sensitivity to uphill, steep downhill, and lateral tilt.

[0070] Step 6: Partition relabeling and numerical stability.

[0071] Directional costs within the corridor Relabel by partition:

[0072] hard zone: ;

[0073] Easy section: ;

[0074] In the hard zone, the cost (i.e., the reciprocal of speed) is increased proportionally to "promote straight travel", and in the easy zone, it is decreased to "promote turning". In the SE(2) refinement, the median of ξ in the corridor is taken and converted into a physical scalar according to gs (physical scale coefficient) to enhance stability.

[0075] Step 7: Use SE(2) for flexible geodesy refinement in the corridor.

[0076] by For the velocity field, the velocity outside the corridor is close to zero (unreachable). Only a directionless endpoint is provided so that the algorithm can automatically select the optimal heading. The unique refined path (i.e., the final path) is obtained by using ODE (Ordinary Differential Equation Solver). Since the velocity outside the corridor is near zero, the path is naturally confined to the ROI (Region of Interest, which is a refined subdomain defined by the corridor tube and its bounding box driven by the clearance, i.e. the area where the path is actually allowed to exist and undergoes elastic geodesic refinement), avoiding multiple approximate equivalent solutions.

[0077] This embodiment provides a tunnel path planning method with clearance corridor and elastic geodesy refinement. For tunnel environments with walls, obstacles, undulating terrain and roughness conditions, it combines clearance constraints, global coarse path, adaptive corridor and SE(2) elastic geodesy refinement to achieve the generation of a unique smooth passage path in a confined space with remote obstacle safety, curvature control and computational efficiency, suppressing wall-hugging and meaningless bends, and taking into account accessibility and attitude safety.

[0078] Example 2

[0079] This embodiment provides a tunnel path planning system with clearance corridor and flexible geodesy refinement, including:

[0080] The environment modeling module is configured to: for the target tunnel, acquire a raster map of the structural terrain;

[0081] The cost generation module is configured to: for structural terrain, calculate the Euclidean distance from each grid point to the nearest obstacle or wall to obtain the net distance field, construct a wall penalty factor based on the net distance field, and combine it with the location features according to weights to generate a direction-independent cost.

[0082] The corridor generation module is configured to: construct a dual metric based on orientation-independent cost; run Riemannian geometric fast travel and backtrack geodesy given the starting and ending pixel coordinates to obtain a coarse path; rasterize the coarse path; calculate the distance from each pixel to the coarse path; obtain a variable-width radius along the coarse path using the clearance; and propagate to the global area using the nearest neighbor to obtain a radius map; and define the corridor based on the distance from each pixel to the coarse path and the radius map.

[0083] The path planning module is configured as follows: For a corridor, an approximate curvature field is obtained by propagating from the coarse path corners, and a score is calculated by combining the clearance. After being partitioned by thresholding, a piecewise linear penalty is applied, and the direction-related total cost is updated based on the penalty. After being relabeled by partition, its reciprocal is used as the velocity field, and the final path is obtained through an ordinary differential equation solver.

[0084] It should be noted that each module in this embodiment corresponds one-to-one with each step in Embodiment 1, and their specific implementation processes are the same, so they will not be repeated here.

[0085] Example 3

[0086] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the tunnel path planning method for clearance corridors and flexible geodesy refinement as described in Embodiment 1 above.

[0087] Example 4

[0088] This embodiment provides a computer device, such as... Figure 2 As shown, the system includes a computer-readable storage medium 1003, a processor 1001, a communication interface 1002, and a computer program stored on the computer-readable storage medium 1003 and executable on the processor 1001. The processor 1001, communication interface 1002, and computer-readable storage medium 1003 can be connected via a bus or other means. The communication interface 1002 is used to receive and send data. When the processor 1001 executes the program, it implements the steps in the tunnel path planning method for clearance corridors and flexible geodesy refinement as described in Embodiment 1 above.

[0089] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A tunnel path planning method with clearance corridor and flexible geodesy refinement, characterized in that, include: For the target tunnel, obtain a raster map of the structural terrain; For structural terrain, calculate the Euclidean distance from each grid point to the nearest obstacle or wall to obtain the clearance distance field. Based on the clearance distance field, construct the wall penalty factor and combine it with the location features according to weights to generate the orientation-independent cost. Based on the direction-independent cost, a dual metric is constructed. Given the starting and ending pixel coordinates, a Riemannian geometric fast traversal and backtracking geodesy are performed to obtain a coarse path. After rasterizing the coarse path, the distance from each pixel to the coarse path is calculated. The widening radius is obtained along the coarse path using the clearance, and the nearest neighbor is propagated to the entire domain to obtain a radius map. Based on the distance from each pixel to the coarse path and the radius map, a corridor is defined. For the corridor, the approximate curvature field is obtained by propagating from the coarse path corner, and the score is calculated by combining the clearance. After being partitioned by thresholding, piecewise linear penalty is applied, and the direction-related total cost is updated based on the penalty. After being relabeled by partition, its reciprocal is used as the velocity field. The final path is obtained by solving the ordinary differential equation.

2. The tunnel path planning method with clearance corridor and flexible geodesy refinement as described in claim 1, characterized in that, The location features include slope intensity and roughness.

3. The tunnel path planning method with clearance corridor and flexible geodesy refinement as described in claim 1, characterized in that, The wall-adjacent penalty factor is: Where r0 is the safety radius and r1 is the penalty band width. Let represent the net air volume, and t represent the normalized parameter of the net air volume. It represents a very small positive number.

4. The tunnel path planning method with clearance corridor and flexible geodesy refinement as described in claim 1, characterized in that, The widening radius: , where r max and r min These are the maximum and minimum widths of the corridor. It is the airspace adaptation coefficient. Represents pixels Clear space This represents the constant offset in a linear mapping.

5. The tunnel path planning method with clearance corridor and flexible geodesy refinement as described in claim 1, characterized in that, The corridor is ,in, For pixels, The distance from the pixel to the coarse path. This refers to the radius in the radius diagram.

6. The tunnel path planning method with clearance corridor and flexible geodesy refinement as described in claim 1, characterized in that, The fraction is: ,in, For a curvature field, For purification, For slope, For roughness, , , and All are weights.

7. The tunnel path planning method with clearance corridor and flexible geodesy refinement as described in claim 1, characterized in that, The piecewise linear penalty includes: Punishment for looking up and down: ; Lateral tilt penalty: ; in, It is the roll penalty factor. It is the roll threshold. It is the uphill penalty coefficient. It is the uphill threshold. It is the coefficient for steep downhill slopes. It is the downhill threshold, pitch. Lateral tilt ,course The unit vector is , ( ) represents the gradient.

8. A tunnel path planning system with clearance corridor and flexible geodesy refinement, characterized in that, include: The environment modeling module is configured to: for the target tunnel, acquire a raster map of the structural terrain; The cost generation module is configured to: for structural terrain, calculate the Euclidean distance from each grid point to the nearest obstacle or wall to obtain the net distance field, construct a wall penalty factor based on the net distance field, and combine it with the location features according to weights to generate a direction-independent cost. The corridor generation module is configured to: construct a dual metric based on orientation-independent cost; run Riemannian geometric fast travel and backtrack geodesy given the starting and ending pixel coordinates to obtain a coarse path; rasterize the coarse path; calculate the distance from each pixel to the coarse path; obtain a variable-width radius along the coarse path using the clearance; and propagate to the global area using the nearest neighbor to obtain a radius map; and define the corridor based on the distance from each pixel to the coarse path and the radius map. The path planning module is configured as follows: For a corridor, an approximate curvature field is obtained by propagating from the coarse path corners, and a score is calculated by combining the clearance. After being partitioned by thresholding, a piecewise linear penalty is applied, and the direction-related total cost is updated based on the penalty. After being relabeled by partition, its reciprocal is used as the velocity field, and the final path is obtained through an ordinary differential equation solver.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the tunnel path planning method with clear corridor and flexible geodesy refinement as described in any one of claims 1-7.

10. A computer device comprising a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the tunnel path planning method with clearance corridor and flexible geodesy refinement as described in any one of claims 1-7.

Images