A Robot Path Planning Method and System Based on Tunnel Feature Constraints

Abstract

This invention discloses a robot path planning method and system based on tunnel feature constraints, belonging to the field of autonomous navigation for tunnel robots. The method involves acquiring 3D point cloud data of the tunnel to generate an initial 3D tunnel map model; acquiring tunnel surface images and inputting them into a YOLOv5 deep learning model for identification to obtain damaged areas; mapping these damaged areas onto the initial 3D tunnel map model; defining impassable areas and safe distances based on the damaged areas; generating a 3D safety map based on the impassable areas and safe distances; obtaining a 3D tunnel map with superimposed multi-constraint fields based on multi-feature constraint joint modeling and the 3D safety map; generating a global time field from the starting point to each grid based on the superimposed multi-constraint field 3D tunnel map; generating path points from the endpoint backward along the time field gradient; verifying the distances of the path points; and outputting the planned path if the distance exceeds a set limit. This ensures that the tunnel robot can complete its task safely and efficiently.

Description

Technical Field

[0001] This invention belongs to the field of autonomous navigation for tunnel robots, and particularly relates to a robot path planning method and system based on tunnel feature constraints. Background Technology

[0002] The statements in this section are merely background information relating to this disclosure and do not necessarily constitute prior art.

[0003] Tunnel construction often faces challenges such as variable geological conditions, complex structural forms, and harsh construction environments. Tunnel robots, with their advantages of autonomous operation and the ability to replace human labor in high-risk environments, have become core equipment in tunnel inspection, equipment installation, and emergency rescue scenarios. Their level of intelligence directly affects the efficiency and safety of underground engineering construction.

[0004] As the underlying technology supporting autonomous navigation of robots, path planning algorithms are a key technical bottleneck in improving the environmental adaptability and task reliability of tunnel robots. Existing global path planning methods include Dijkstra's algorithm, ant colony algorithm, and A* algorithm, which have been widely applied in other fields. However, traditional algorithms are mostly designed for ground environments and there is no suitable robot path planning method for tunnel feature constraints; moreover, the identification of structurally damaged areas mostly relies on manual annotation or simple geometric judgment, making it difficult to achieve precise division. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, this invention provides a robot path planning method and system based on tunnel feature constraints, aiming to ensure that tunnel robots can complete tasks safely and efficiently.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions: In a first aspect, the present invention provides a robot path planning method based on tunnel feature constraints, comprising: Acquire 3D point cloud data of the tunnel and generate an initial 3D tunnel map model; The tunnel surface image is acquired and input into the YOLOv5 deep learning model for identification to obtain the damaged area. The damaged area is then mapped onto the initial 3D tunnel map model, and impassable areas and safe distances are defined based on the damaged area. Generate a 3D safety map based on impassable areas and safe distances; Based on multi-feature constraint joint modeling and combined with a 3D safety map, a 3D tunnel map with superimposed multi-constraint fields is obtained; Based on a 3D tunnel map with superimposed multi-constraint fields, a global time field is generated from the starting point to each grid. Path points are generated in reverse along the time field gradient from the endpoint. The distance of the path points is checked, and if the distance is greater than the set limit, the planned path is output.

[0007] A further technical solution defines the damaged and deformed areas as impassable areas, calculates the minimum bounding box for each damaged or deformed area, and expands the bounding box according to the safety distance.

[0008] A further technical solution involves setting the constraint strength of the impassable area to infinity and defining a safe distance from the impassable area.

[0009] A further technical solution involves jointly establishing a tunnel passable area model using multiple feature constraints, including equipment clearance, structural damage, traffic rules, and other constraint fields.

[0010] A further technical solution involves using an exponential decay model to quantify the constraint field, where the constraint strength decreases with increasing distance.

[0011] In a further technical solution, the constraint strength is adjusted according to the structural damage risk level, and the constraint field generates a gradient field according to the risk level, with the constraint strength being greater for higher regional risks.

[0012] A further technical solution involves generating a global time field from the starting point to each grid using an improved fast travel method, whereby the multi-constraint field intensity is converted into wavefront propagation speed.

[0013] Secondly, the present invention provides a robot path planning system based on tunnel feature constraints, comprising: The tunnel model construction module is configured to: acquire 3D point cloud data of the tunnel and generate an initial 3D tunnel map model; The damaged area identification module is configured to: acquire tunnel surface images, input them into the YOLOv5 deep learning model for identification, obtain damaged areas, map the damaged areas onto the initial 3D tunnel map model, and define impassable areas and safety distances based on the damaged areas; The safety map generation module is configured to generate a 3D safety map based on impassable areas and safety distances. The constraint overlay module is configured to: obtain a three-dimensional tunnel map with overlaid multi-constraint fields based on multi-feature constraint joint modeling and combined with a three-dimensional safety map; The path planning module is configured to: generate a global time field from the starting point to each grid based on a 3D tunnel map with superimposed multi-constraint fields; generate path points from the endpoint in reverse along the time field gradient; perform distance verification on the path points; and output the planned path if the distance is greater than the set limit.

[0014] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the robot path planning method based on tunnel feature constraints as described in the first aspect.

[0015] Fourthly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the robot path planning method based on tunnel feature constraints as described in the first aspect.

[0016] The above one or more technical solutions have the following beneficial effects: This invention addresses the inherent constraints of tunnels, such as equipment clearance, traffic rules, structural damage, and equipment avoidance rules, by establishing a multi-constraint function model. This model integrates tunnel constraints into the path planning calculation process, thereby ensuring that tunnel robots can complete their tasks safely and efficiently.

[0017] This invention introduces the YOLOv5 model to identify tunnel damage features. Based on this, the constraint strength is divided according to the structural damage risk level, enabling path planning to avoid risk areas differently according to the severity of damage. Compared with damage handling methods such as manual annotation or simple geometric judgment, this invention improves the identification accuracy, performs fine division, and makes constraint modeling more accurate. Attached Figure Description

[0018] 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.

[0019] Figure 1 This is a flowchart of a robot path planning method based on tunnel feature constraints according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the three-dimensional multi-constraint field superposition of the tunnel according to an embodiment of the present invention. Detailed Implementation

[0020] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration 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.

[0021] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0022] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0023] Example 1 like Figure 1 As shown, this embodiment discloses a robot path planning method based on tunnel feature constraints, which includes the following steps: S1: Acquire 3D point cloud data of the tunnel and generate an initial 3D tunnel map model; In this embodiment, the tunnel is scanned by a 3D LiDAR to obtain 3D point cloud data of the tunnel. The point cloud data is then imported into Blender software to generate an initial 3D tunnel map model that includes the tunnel walls and the outline of the construction equipment.

[0024] After determining the scanning range and planning the station locations, a 3D LiDAR is used to scan at multiple stations, with each station acquiring high-density point cloud data of a local area of ​​the tunnel. The point clouds acquired from multiple stations are aligned to a unified coordinate system, and the point cloud data is denoised and filtered to obtain denoised point cloud data. Based on the denoised point cloud data, the tunnel structure is extracted, segmenting the tunnel walls, arch, road surface, and other structures. Based on point cloud clustering, equipment point clouds (such as ventilation ducts, fire boxes, cables, etc.) are separated to extract equipment outlines. Surface reconstruction (triangular mesh reconstruction) is performed on the tunnel walls and equipment point clouds. After network optimization, an initial 3D tunnel map model is generated.

[0025] S2: Acquire the tunnel surface image, input it into the YOLOv5 deep learning model for identification, obtain the damaged area, map the damaged area onto the initial 3D tunnel map model, and define the impassable area and safety distance based on the damaged area; In this embodiment, images of the tunnel surface are acquired, and damaged areas such as cracks, seepage, and spalling are marked. A dataset is constructed based on the acquired data, and the YOLOv5 deep learning model is trained on the dataset using the CLoU Loss loss function to obtain the trained YOLOv5 deep learning model.

[0026] Images of the tunnel surface are acquired, and a YOLOv5 deep learning model is applied for damage identification, outputting the damage category, bounding box coordinates, and confidence score. Pixels within the detection boxes are back-projected into 3D space, and the defect point cloud mapped into 3D space is clustered to obtain a 3D tunnel map model.

[0027] Define impassable areas, specifically: Expand damaged areas: Calculate the minimum bounding box for each damaged or deformed area, and expand the bounding box according to the safety distance. Similarly, perform the same expansion operation on tunnel equipment (such as ventilation ducts, fire extinguisher boxes, etc.).

[0028] Calculate the accessible space, specifically: define the free space and calculate the reachable space for the robot.

[0029] Based on the YOLOv5 deep learning model, tunnel image datasets were trained to identify damage features such as cracks and water seepage. Uneven wall surfaces and structural deformation areas were delineated using LiDAR point cloud data. The identification results were simultaneously marked on a 3D tunnel map, allowing the extraction of coordinates for equipment and structurally damaged areas within the tunnel from the delineated point cloud data. This area was defined as impassable, with an infinite constraint strength, and safe distances from equipment and damaged structures were defined, such as a minimum horizontal safety distance of 0.5m and a minimum vertical safety distance of 0.3m between the robot and equipment. Figure 2 The diagram shows a 3D map of the tunnel. Based on manually set parameters, the risk intensity of the structurally damaged areas is defined and divided into three levels: low risk, medium risk, and high risk.

[0030] S3: Generate a 3D safety map based on impassable areas and safe distances; In this embodiment, a binary security map is constructed. ,in:

[0031] 0 and 1 represent impassable areas and free areas, respectively. Impassable areas are determined by the height of the tunnel ground. Areas higher than the height of the tunnel robot chassis are set as impassable, while those lower are set as free space.

[0032] S4: Based on multi-feature constraint joint modeling, combined with a 3D safety map, a 3D tunnel map with superimposed multi-constraint fields is obtained; In this embodiment, a tunnel passable area model is jointly established using multiple feature constraints, including equipment clearance, structural damage, traffic rules, and other constraint fields. Equipment clearance constraints ensure a safe distance between the robot and tunnel equipment. The constraint field uses an exponential decay model to quantify the equipment, with constraint strength decreasing as distance increases. The constraint strength is defined as follows. The constraint strength is adjusted according to the structural failure risk level. A gradient field is generated for the constraint field based on the risk level; the higher the risk in a region, the greater the constraint strength. (Definition...) for:

[0033] Traffic rules constraints are the traffic rules that tunnel robots must follow when navigating tunnels, such as keeping to the right and speed limits. In actual engineering projects, the rules vary from tunnel to tunnel. Taking keeping to the right as an example, let's define what happens when a robot mistakenly enters the oncoming lane. =1.2, which makes the path planning more biased towards the correct driving area. Other constraints represent other factors that need to be considered in different tunnel path planning, such as special construction terrain, moving construction personnel, and obstacles.

[0034] After setting each constraint, the constraints are dynamically weighted and fused to form the tunnel feature constraint strength, which is expressed as:

[0035] in, , , , Different constraint coefficients are used to represent varying degrees of constraint strength required in different tunnel construction environments. For example, in areas with dense equipment, the adjustment value is given... Higher weighting coefficients are applied to key transportation routes and high-risk structural areas, respectively. and Higher weighting coefficients are set. The final result is a 3D tunnel map with superimposed multi-constraint fields.

[0036] S5: Based on the superimposed multi-constraint field 3D tunnel map, generate the global time field from the starting point to each grid, generate path points from the endpoint in reverse along the time field gradient, perform distance verification on the path points, and output the planned path if the distance is greater than the set limit.

[0037] In this embodiment, the improved fast travel path is calculated. The multi-constraint field intensity is converted into wavefront propagation velocity; the stronger the constraint, the lower the velocity.

[0038] in, As the reference speed, This is the normalized constraint threshold. Constraint strength. The higher the speed of transmission The lower the value, the lower the accessibility of that location, and areas with high accessibility will be prioritized when planning routes.

[0039] After obtaining the improved velocity function, the global time field is calculated according to the following formula. For ease of calculation, the overall tunnel model is divided into several rectangular grids. A higher grid size results in higher calculation accuracy; the grid size can be adjusted according to actual needs. Starting from the starting point, the shortest arrival time T for each grid point is calculated based on the velocity function F, satisfying the following Eikonal equation:

[0040] After generating the global time field, reverse the time field gradient from the endpoint. Extract the initial path P. And assign path points... Perform distance verification; if the distance to the device... If the equipment is less than the specified clearance, cubic spline interpolation is used for extrapolation correction, resulting in:

[0041] in, For the newly generated path points, For the original path points, The distance from the device.

[0042] This invention converts the intensity of multiple constraint fields into wavefront propagation speed based on an improved fast travel method, constructing a constraint-aware global time field. In this invention, regions with stronger constraints have lower propagation speeds, and regions with high passability are prioritized during path planning. The global time field is generated by solving the Eikonal equation starting from the origin, and the path is extracted backward along the time field gradient from the destination. Distance verification and cubic spline interpolation extrapolation correction mechanisms are introduced to ensure that the generated path maintains a safe distance from the equipment. This method integrates multiple constraint fields into the path generation process, ensuring both safety and travel efficiency.

[0043] The system continuously monitors environmental changes and updates the constraint field. When a path is detected to be blocked, it triggers local replanning and outputs the final optimized path for robot trajectory tracking.

[0044] The method described in this invention can adapt to complex tunnel environments, generating optimal collision-free paths based on tunnel characteristic constraints and driving tasks. Simulation tests in actual tunnel environments demonstrate that this invention can adapt to complex tunnel conditions.

[0045] Example 2 This embodiment discloses a robot path planning system based on tunnel feature constraints, including: The tunnel model construction module is configured to: acquire 3D point cloud data of the tunnel and generate an initial 3D tunnel map model; The damaged area identification module is configured to: acquire tunnel surface images, input them into the YOLOv5 deep learning model for identification, obtain damaged areas, map the damaged areas onto the initial 3D tunnel map model, and define impassable areas and safety distances based on the damaged areas; The safety map generation module is configured to generate a 3D safety map based on impassable areas and safety distances. The constraint overlay module is configured to: obtain a three-dimensional tunnel map with overlaid multi-constraint fields based on multi-feature constraint joint modeling and combined with a three-dimensional safety map; The path planning module is configured to: generate a global time field from the starting point to each grid based on a 3D tunnel map with superimposed multi-constraint fields; generate path points from the endpoint in reverse along the time field gradient; perform distance verification on the path points; and output the planned path if the distance is greater than the set limit.

[0046] Example 3 The purpose of this embodiment is to provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method of Embodiment 1.

[0047] Example 4 The purpose of this embodiment is to provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the method of Embodiment 1.

[0048] The steps and methods involved in the apparatuses of Embodiments 3 and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0049] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0050] 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.

[0051] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A robot path planning method based on tunnel feature constraints, characterized in that, include: Acquire 3D point cloud data of the tunnel and generate an initial 3D tunnel map model; The tunnel surface image is acquired and input into the YOLOv5 deep learning model for identification to obtain the damaged area. The damaged area is then mapped onto the initial 3D tunnel map model, and impassable areas and safe distances are defined based on the damaged area. Generate a 3D safety map based on impassable areas and safe distances; Based on multi-feature constraint joint modeling and combined with a 3D safety map, a 3D tunnel map with superimposed multi-constraint fields is obtained; Based on a 3D tunnel map with superimposed multi-constraint fields, a global time field is generated from the starting point to each grid. Path points are generated in reverse along the time field gradient from the endpoint. The distance of the path points is checked, and if the distance is greater than the set limit, the planned path is output.

2. The robot path planning method based on tunnel feature constraints as described in claim 1, characterized in that, Define damaged and deformed areas as impassable areas, calculate the minimum bounding box for each damaged or deformed area, and expand the bounding box according to the safety distance.

3. The robot path planning method based on tunnel feature constraints as described in claim 2, characterized in that, Set the constraint strength of the impassable area to infinity and define a safe distance from the impassable area.

4. The robot path planning method based on tunnel feature constraints as described in claim 1, characterized in that, A model of the tunnel's passable area is jointly established using multiple feature constraints, including equipment clearance, structural damage, traffic rules, and other constraint fields.

5. The robot path planning method based on tunnel feature constraints as described in claim 4, characterized in that, The constraint field is quantized using an exponential decay model, and the constraint strength decreases with increasing distance.

6. The robot path planning method based on tunnel feature constraints as described in claim 4, characterized in that, The constraint strength is adjusted according to the structural damage risk level, and the constraint field generates a gradient field according to the risk level. The higher the risk of the area, the greater the constraint strength.

7. The robot path planning method based on tunnel feature constraints as described in claim 1, characterized in that, The improved fast travel method generates a global time field from the starting point to each grid, which converts the multi-constraint field intensity into wavefront propagation speed.

8. A robot path planning system based on tunnel feature constraints, characterized in that, include: The tunnel model construction module is configured to: acquire 3D point cloud data of the tunnel and generate an initial 3D tunnel map model; The damaged area identification module is configured to: acquire tunnel surface images, input them into the YOLOv5 deep learning model for identification, obtain damaged areas, map the damaged areas onto the initial 3D tunnel map model, and define impassable areas and safety distances based on the damaged areas; The safety map generation module is configured to generate a 3D safety map based on impassable areas and safety distances. The constraint overlay module is configured to: obtain a three-dimensional tunnel map with overlaid multi-constraint fields based on multi-feature constraint joint modeling and combined with a three-dimensional safety map; The path planning module is configured to: generate a global time field from the starting point to each grid based on a 3D tunnel map with superimposed multi-constraint fields; generate path points from the endpoint in reverse along the time field gradient; perform distance verification on the path points; and output the planned path if the distance is greater than the set limit.

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 robot path planning method based on tunnel feature constraints as described in any one of claims 1-7.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the robot path planning method based on tunnel feature constraints as described in any one of claims 1-7.

Images