Automatic path planning method and device for railway tunnel unmanned aerial vehicle and medium

Abstract

The invention relates to the technical field of unmanned aerial vehicles, and discloses an automatic path planning method and device for a railway tunnel unmanned aerial vehicle and a medium, and the method comprises the steps: obtaining a railway tunnel environment model, a preset take-off and landing point, an operation starting point and an operation ending point; and planning a take-off path, a landing path and an operation path by using an A * algorithm and a bidirectional search strategy, guiding a search direction by using a comprehensive distance heuristic function in the bidirectional search process, and introducing a ratio of a first distance between a current node and a target node to a second distance between a starting node and the target node into the comprehensive distance heuristic function. By considering the relative distance between the current node and the target node and the overall distance relationship between the starting node and the target node, the importance and potential value of the current node can be evaluated more accurately, so that the priority search node is determined more efficiently in the search, the optimal solution is found more quickly, and the search accuracy is effectively improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of unmanned aerial vehicles (UAVs), and in particular to an automatic path planning method, device and medium for an unmanned aerial vehicle (UAV) in a railway tunnel. Background Art

[0002] Drones, as an efficient and flexible tool, have significant advantages when conducting inspections and filming in railway tunnels. However, due to the small space and complex environment in tunnels, the take-off and landing of drones and the planning of operation paths face many challenges. Traditional methods of drone take-off and landing and operation path planning cannot meet the needs of operations in tunnels. Therefore, there is an urgent need for an automatic path planning solution for drones in railway tunnels to meet the needs of operations in tunnels. Summary of the invention

[0003] The purpose of the present invention is to at least provide an automatic path planning method, device and medium for a railway tunnel drone, which can at least solve the problems of automatic take-off and landing and operation path planning of the railway tunnel drone in a small space, and at least achieve the effect of safe and efficient operation of the drone in the tunnel.

[0004] To solve the above technical problems, at least one embodiment of the present application provides an automatic path planning method for a UAV in a railway tunnel, comprising:

[0005] Obtain the railway tunnel environment model, preset take-off and landing points, and operation start and end points;

[0006] The A* algorithm is used and a bidirectional search strategy is adopted to plan the take-off path between the preset take-off and landing point and the operation starting point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation starting point and the operation end point. During the bidirectional search process, a comprehensive distance heuristic function is used to guide the search direction. The comprehensive distance heuristic function introduces the ratio of the first distance between the current node and the target node to the second distance between the starting node and the target node.

[0007] In some optional embodiments, the preset take-off point and the preset landing point are located in a refuge tunnel of a railway tunnel or around an entrance or exit of a railway tunnel.

[0008] In some optional embodiments, before obtaining the railway tunnel environment model, the preset take-off and landing points, and the operation start point and the operation end point, the method further includes:

[0009] Obtain environmental data for railway tunnels;

[0010] Based on the environmental data, an environmental model of the railway tunnel is constructed and key elements in the railway tunnel are identified.

[0011] In some optional embodiments, the A* algorithm is used to plan the take-off path between the preset take-off and landing point and the operation start point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation start point and the operation end point by adopting a bidirectional search strategy, including:

[0012] According to the key elements in the railway tunnel and the operation requirements of the UAV, the take-off path, the landing path, and the operation path are planned respectively by using the A* algorithm and a bidirectional search strategy;

[0013] The key elements in the railway tunnel include at least one of cables, rails, and escape tunnels.

[0014] In some optional embodiments, the expression of the comprehensive distance heuristic function is as follows:

[0015] h(n)=1.2×ω(n)×d(n)-u(n)

[0016]

[0017] x(n)=|x g -x n |

[0018] y(n)=|y g -y n |

[0019] min x,y =min(|x g -x n |,|y g -y n |)

[0020]

[0021] In the formula, h(n) represents the comprehensive distance heuristic function, ω(n) represents the dynamic weight, d(n) represents the comprehensive distance, u(n) represents the ratio of the first distance between the current node and the target node to the second distance between the starting node and the target node, (x g ,y g ) represents the coordinates of the target node, (x n ,y n ) represents the coordinates of the current node, x(n) represents the distance between the coordinates of the current node and the target node in the horizontal direction, and y(n) represents the distance between the coordinates of the current node and the target node in the vertical direction.

[0022] In some optional embodiments, when the comprehensive distance is greater than a preset value, the dynamic weight ω(n) is set to a first value; when the comprehensive distance is less than or equal to a preset value, the dynamic weight ω(n) is set to a second value; the first value is greater than the second value.

[0023] In some optional embodiments, the automatic path planning method for a railway tunnel UAV further includes:

[0024] Remove redundant turning points from the planned takeoff path, landing path, and operation path;

[0025] The Bézier curve is used to smooth the takeoff path, landing path and operation path after removing redundant inflection points.

[0026] At least one embodiment of the present application further provides a method for operating a UAV in a railway tunnel, comprising:

[0027] Obtaining a take-off path, a landing path, and an operation path according to the automatic path planning method for the railway tunnel UAV;

[0028] Using the line-biting flight algorithm to control the UAV to start operation according to the take-off path, the operation path and the landing path;

[0029] In case of need to return during operation, the drone descends below the contact network plane;

[0030] The UAV returns according to the take-off path, the operation path and the landing path.

[0031] At least one embodiment of the present application further provides an automatic path planning device for a railway tunnel drone, comprising:

[0032] An acquisition module is used to acquire the railway tunnel environment model, preset take-off and landing points, and the operation start point and operation end point;

[0033] The first planning module is used to plan the take-off path between the preset take-off and landing point and the operation start point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation start point and the operation end point by using the A* algorithm and a two-way search strategy. During the two-way search, a comprehensive distance heuristic function is used to guide the search direction. The comprehensive distance heuristic function introduces the ratio of the first distance between the current node and the target node to the second distance between the start node and the target node.

[0034] At least one embodiment of the present application further provides a railway tunnel drone operation device, comprising:

[0035] A second planning module is used to obtain a take-off path, a landing path, and an operation path according to the automatic path planning device of the railway tunnel UAV;

[0036] The operation module is used for the UAV to start operation according to the take-off path, the operation path and the landing path.

[0037] At least one embodiment of the present application further provides an electronic device, including:

[0038] at least one processor; and,

[0039] a memory communicatively connected to the at least one processor; wherein,

[0040] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor so that the at least one processor can perform the above method.

[0041] At least one embodiment of the present application further provides a computer-readable storage medium storing a computer program, wherein the computer program implements the above method when executed by a processor.

[0042] In the inspection path planning of railway tunnel drones, a comprehensive distance heuristic function is adopted. By considering the relative distance between the current node and the target node and the overall distance relationship between the starting node and the target node, the importance and potential value of the current node can be evaluated more accurately, thereby more efficiently determining the priority search nodes in the search, finding the optimal solution faster, and effectively improving the search accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] One or more embodiments are exemplarily described by the pictures in the corresponding drawings, and these exemplary descriptions do not constitute limitations on the embodiments.

[0044] Figure 1 It is a schematic flow chart of an automatic path planning method for a railway tunnel drone provided in an embodiment of the present application;

[0045] Figure 2 It is a schematic diagram of the flow of the railway tunnel drone operation method provided by the embodiment of the present application;

[0046] Figure 3 It is a schematic diagram of a railway tunnel drone automatic inspection system provided by an embodiment of the present application;

[0047] Figure 4 It is a schematic diagram of the path planning and flight control process provided in the embodiment of the present application. DETAILED DESCRIPTION

[0048] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the present application, many technical details are proposed in order to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present application can also be implemented. The division of the following embodiments is for the convenience of description, and the specific implementation of the present application should not be construed as any limitation, and the various embodiments can be combined and referenced with each other under the premise of no contradiction.

[0049] In order to solve the problem of automatic take-off and landing and operation path planning of a UAV in a narrow space in a railway tunnel, the present invention provides an automatic path planning method, device and medium for a UAV in a railway tunnel. The following is a detailed description of the implementation details of the embodiment of the present invention. The following content is only for the convenience of understanding the implementation details provided and is not necessary for the implementation of this solution.

[0050] Embodiment 1

[0051] The automatic path planning method for a railway tunnel drone of this embodiment can be applied to electronic devices with communication, computing and data storage capabilities. The specific process can be as follows: Figure 1 As shown, including:

[0052] Step 101: Obtain a railway tunnel environment model, preset take-off and landing points, and an operation start point and an operation end point.

[0053] In some specific implementations, the preset take-off point and the preset landing point are located in a refuge tunnel of a railway tunnel or around an entrance or exit of the railway tunnel.

[0054] In some examples, a take-off and landing platform is set up in the shelter of the railway tunnel, and a preset take-off and landing point is selected on the take-off and landing platform. In other examples, a take-off and landing site is set up in a relatively open area near the entrance and exit of the railway tunnel, and a preset take-off and landing point is selected within the take-off and landing site. In the case of setting up a take-off and landing site, protective facilities can also be built at the take-off and landing site to ensure that the take-off and landing of the drone are not interfered with by airflow and other factors in the railway tunnel. At the same time, it can also be equipped with take-off and landing guidance equipment, such as a lighting guidance system or radio navigation equipment, to help the drone take off and land accurately.

[0055] In some specific implementations, the operation start point and the operation end point can be set below the contact network plane, for example, at a height of about 1m below the contact network plane, and set at a position at the height that is the same distance from the vertical walls on both sides of the railway tunnel. In actual flight, the drone takes off from the preset take-off and landing point to the operation start point and completes take-off. Starting from the operation start point, the inspection operation task is performed, first flying upward from the operation start point to an open and safe area above the contact network plane, and then completing the inspection operation along the operation path. After reaching the operation end point, it flies downward to a height of about 1m below the contact network plane, then flies back to the operation start point, and then returns to the preset take-off and landing point. It should be understood that the height of about 1m below the contact network plane is only an example, and can be set according to the tunnel environment in actual applications, and is not limited to this.

[0056] Step 102, using the A* algorithm, a two-way search strategy is adopted to plan the take-off path between the preset take-off and landing point and the operation start point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation start point and the operation end point. During the two-way search process, a comprehensive distance heuristic function is used to guide the search direction. The comprehensive distance heuristic function introduces the ratio of the first distance between the current node and the target node to the second distance between the start node and the target node.

[0057] The method of this embodiment can reduce the human error in the operation of the drone in the railway tunnel, and improve the safety of the operation. After the drone flies to the starting point of the operation, the drone can operate in a relatively open environment, with a relatively high fault tolerance rate, which improves the operation efficiency. After the operation is completed, the drone only needs to be lowered to below the contact network plane, and the drone can be recovered through the return command, which has a high degree of automation and simple operation.

[0058] In some specific implementations, before obtaining the railway tunnel environment model, the preset take-off and landing points, and the operation start point and the operation end point, the method of this embodiment further includes:

[0059] Obtaining environmental data for railway tunnels; and

[0060] Based on the environmental data, an environmental model of the railway tunnel is constructed and key elements inside the railway tunnel are identified.

[0061] The drone is equipped with a variety of sensors, including laser radar, camera, inertial measurement unit (IMU), etc. These sensors are used to comprehensively scan the environment inside the railway tunnel to obtain environmental data; among them, the laser radar obtains the three-dimensional spatial point cloud data of the tunnel, the camera captures the internal image of the railway tunnel, and the inertial measurement unit records the attitude information of the drone. The collected environmental data is preprocessed, including filtering, denoising, and registration of the three-dimensional point cloud data to improve the accuracy and availability of the data. At the same time, the images taken by the camera are subjected to feature extraction and image enhancement processing to prepare for the key element identification and mapping of the railway tunnel.

[0062] The 3D environment model of the railway tunnel is constructed based on the pre-processed LiDAR point cloud data and camera image data. The key elements in the tunnel, such as cables, rails, and shelters, are identified using point cloud processing algorithms and image recognition technology. For cables, the location, direction, and type of the cables are accurately determined by analyzing the texture features and geometric shapes in the image and combining the spatial position information in the 3D point cloud data. For rails, the shape features and continuity of the point cloud data are used for identification. The identification of shelters is based on their specific spatial shapes and positional relationships, and the location and size of the shelters are marked in the 3D environment model.

[0063] Image processing algorithms are used to identify key elements in railway tunnels, including at least one of the cables, rails, and escape hatches. This helps drones accurately determine their own positions during flight and avoid collisions with important facilities, while also providing an important reference for subsequent path planning. For example, identifying escape hatches can determine suitable take-off and landing locations, and identifying rails and cables can ensure that the flight path is safe and meets operational requirements.

[0064] In some specific implementations, the A* algorithm is used to plan the take-off path between the preset take-off and landing point and the operation start point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation start point and the operation end point using a bidirectional search strategy, including:

[0065] Step 102a: According to the key elements in the railway tunnel and the operation requirements of the UAV, the A* algorithm is used to adopt a two-way search strategy to plan the take-off path, landing path, and operation path respectively.

[0066] Based on the identified key elements and the operational requirements of the UAV inspection mission, the key inspection areas can be determined. Then, using the constructed three-dimensional environmental model of the railway tunnel, the safe space area suitable for UAV flight can be calculated to avoid obstacles and restricted areas in the railway tunnel. At the same time, the flight performance and operation requirements of the UAV are taken into consideration to provide constraints and reference basis for path planning.

[0067] When planning the take-off path, landing path and operation path, feasible paths are searched in the three-dimensional environment model according to the starting point and the target point (the next location to be reached), and two types of paths are generated. One type is a high-speed track (operation path), which is used for the UAV to fly quickly in a relatively open and safe area to improve inspection efficiency; the other type is a transition path (take-off path, landing path), which enables the UAV to fly smoothly from the preset take-off and landing point to the high-speed track or from the high-speed track to the preset take-off and landing point, ensuring flight stability and safety.

[0068] The A* algorithm combines the advantages of the greedy strategy and the Dijkstra algorithm, and uses the evaluation function f(n) = g(n) + h(n) to screen nodes, where n represents the current node, g(n) is the actual cost from the starting point to the current node, and h(n) is used as a heuristic function (cost function) to estimate the cost from the current node to the target node. The A* algorithm constructs and continuously updates the open list and the closed list, and always selects the node with the lowest cost as the current node for expansion until the target node is found or a specific stopping condition is met. Its heuristic function f(n) is the key factor in guiding the search towards the target node.

[0069] In order to overcome the problems of too many traversed nodes, long paths, low efficiency, etc. in the search process of the A* algorithm, this embodiment adopts a bidirectional search strategy, a comprehensive distance heuristic function, and a global path screening strategy to improve efficiency.

[0070] The bidirectional search strategy searches both the starting node and the target node to speed up the path planning process. It includes the following 7 steps:

[0071] Step 1: Initialize two open lists openlist1 and openlist2, and two closed lists closelist1 and closelist2, which are used to store the nodes to be searched and the nodes that have been searched, respectively.

[0072] Step 2: Add the start node to openlist1 and the target node to openlist2, and then start the search from both the start and target nodes. Each time, select the node with the lowest cost from the two open lists for forward and reverse search. When searching forward, update the heuristic function of the current node and determine whether the optimal node is on the reverse search path. If it exists, the path is found; otherwise, add the node to closelist1.

[0073] Step 3: Process the current optimal node. Find the connection point m of the current node. If node m is in closelist1, search for the next node m. If node m is not in closelist1, determine whether node m is in the forward search openlist1. If node m is not in openlist1, calculate the actual cost g(m), estimated cost h(m) and cost function f(m) of the point.

[0074] Step 4: Set the parent node of node m to n, and add node m to the openlist1 of the forward search in order of cost function f(m) from small to large.

[0075] Step 5: Determine whether the actual cost g(n) plus the distance from parent node n to node m is greater than g(m). If it is greater than g(m), further search for the next connection point of parent node n; if it is less than g(m), recalculate g(m), h(m) and f(m). This process continues until all connection points with parent node n have been searched.

[0076] Step 6: Perform a reverse search while performing the forward search. The steps are the same as the forward search.

[0077] Step 7: Determine the return values ​​of step 4 and step 5. If both openlist1 and openlist2 are empty and the target node has not been reached, it means that the pathfinding has failed. If a node in the start node and the target node search queue meets on the map, a path from the start node to the target node is found. At this time, the complete path can be constructed by backtracking based on the encounter node and its predecessor node.

[0078] In path planning, the guiding role of the heuristic function is crucial. The comprehensive distance heuristic function in this embodiment makes the search path more uniform and smooth by adjusting the distance in the diagonal direction during the calculation process, balancing the cost of straight-line and diagonal movement. Given that the actual costs of the two may vary in different scenarios, this balance can better guide the search direction and improve search efficiency and accuracy.

[0079] In order to improve the flexibility and adaptability of the search, the comprehensive distance heuristic function introduces dynamic weights to control the search speed. In complex environments, the search speed is reduced to carefully consider various possibilities; in simple environments, the search speed is appropriately accelerated to quickly locate the target path, thereby significantly enhancing the efficiency and performance of the algorithm. In addition, the function also incorporates the ratio u(n) of the first distance between the current node and the target node and the second distance between the starting node and the target node to further optimize the search accuracy. By considering the relative distance between the current node and the target node and the overall distance relationship between the starting node and the target node, the algorithm can more accurately evaluate the importance and potential value of the current node, thereby more efficiently determining the priority search nodes in the search and finding the optimal solution faster.

[0080] In some specific implementations, the expression of the comprehensive distance heuristic function is as follows:

[0081] h(n)=1.2×ω(n)×d(n)-u(n)

[0082]

[0083] x(n)=|x g -x n |

[0084] y(n)=|y g -y n |

[0085] min x,y =min(|x g -x n |,|y g -y n |)

[0086]

[0087] In the formula, h(n) represents the comprehensive distance heuristic function, ω(n) represents the dynamic weight, d(n) represents the comprehensive distance, u(n) represents the ratio of the first distance between the current node and the target node to the second distance between the starting node and the target node, (x g ,y g ) represents the coordinates of the target node, (x n ,y n ) represents the coordinates of the current node, x(n) represents the distance between the coordinates of the current node and the target node in the horizontal direction, and y(n) represents the distance between the coordinates of the current node and the target node in the vertical direction.

[0088] When the comprehensive distance is greater than the preset value, the dynamic weight ω(n) is set to a first value; when the comprehensive distance is less than or equal to the preset value, the dynamic weight ω(n) is set to a second value; the first value is greater than the second value. By setting different dynamic weights for different comprehensive distances, the search speed can be effectively controlled.

[0089] In one example, the preset value is 5, the first numerical value is 2.8, and the second numerical value is 0.9.

[0090] In some specific implementations, the automatic path planning method for a railway tunnel drone also includes:

[0091] Remove redundant turning points from the planned takeoff path, landing path, and operation path; and

[0092] The Bézier curve is used to smooth the takeoff path, landing path and operation path after removing redundant inflection points.

[0093] In order to eliminate redundant inflection points in the planned global path, a global path screening strategy is introduced to optimize the path planning results. The specific steps are as follows:

[0094] Step 1: Get the coordinates of each turning point in each path planned by the A* algorithm of the bidirectional search strategy:

[0095] {p1,p2,p3,…pn},

[0096] Place the starting node ps (xs, ys) at the first position of the inflection point coordinate set, and place the target node pg (xg, yg) at the last position to form the set of all key nodes A{ps, p1, p2, p3, …, pn, pg}.

[0097] Step 2: Create an empty set B{SetB} to store the optimized coordinate points.

[0098] Step 3: Determine whether the line segment formed by the first coordinate ps and the last coordinate pg in set A intersects with the obstacle. If not, determine the line segment formed by ps and the second-to-last coordinate pn; if still not, continue to determine the line segment formed by the third-to-last coordinate pn-1, and so on. Until a line segment formed by a certain coordinate is found that does not intersect with the obstacle, add ps to set B, and update set A to {px,…pn,pg}.

[0099] Step 4: Repeat step 3 for the updated set A, and continue to update sets A and B. When the number of coordinates in set A is two, add the last two coordinates to set B, and set B is now the optimized path.

[0100] In some examples, the line segment intersection detection in the path optimization strategy uses the property of vector cross product for judgment, p1 and p2 are two coordinates in the inflection point set, and p3 and p4 are two coordinates in the obstacle coordinate set.

[0101] The process of determining whether line segments intersect is specifically divided into the following two steps:

[0102] Step 1: cp1 is the cross product of vector p1p2 and vector p1p3; cp2 is the cross product of vector p1p2 and vector p1p4; cp3 is the cross product of vector p3p4 and vector p3p1; cp4 is the cross product of vector p3p4 and vector p3p2.

[0103] Step 2: If cp1×cp2≤0 and cp3×cp4≤0, it means that the line segment formed between the inflection points intersects with the line segment formed between the obstacle points. Otherwise, it means that the line segment formed between the inflection points does not intersect with the line segment formed between the obstacle points.

[0104] The Bézier curve is defined by a set of control points, and its shape is determined by the location and number of these control points. Since there are many coordinate points for the drone to fly in the railway tunnel, the calculation is performed recursively.

[0105] For a given control point p0,p1,p2,...,p n and parameter t:

[0106] Initial conditions:

[0107] P i (0) =P i fori=0,1,...,n

[0108] Recursive calculation:

[0109]

[0110] final, is the point of the Bézier curve at parameter t. The parameter t ranges from 0 to 1. By adjusting the value of t, points at different positions on the curve can be obtained.

[0111] Embodiment 2

[0112] The railway tunnel drone operation method of this embodiment can be applied to electronic devices with communication, computing and data storage capabilities. The specific process can be as follows: Figure 2 As shown, including:

[0113] Step 201: Obtain a take-off path, a landing path, and an operation path according to the automatic path planning method for a railway tunnel UAV provided in the previous embodiment.

[0114] Step 202: Use the line-biting flight algorithm to control the UAV to start operation according to the take-off path, operation path and landing path.

[0115] The core idea of ​​the line-biting flight algorithm is to make the drone fly along a predefined path (straight or curved), and keep it on the center line of the target path by continuously adjusting the position and direction of the drone.

[0116] The main steps are as follows:

[0117] Step 1: After obtaining the take-off path, operation path, and landing path of the UAV in the railway tunnel through the A* algorithm and the comprehensive distance heuristic function, the operation begins;

[0118] Step 2: During the operation, the current position of the drone is obtained through the high-precision positioning module, laser radar and camera;

[0119] Step 3: Calculate the deviation between the current position of the drone and the flight path in real time;

[0120] Step 4: Use PID control or pure tracking algorithm to generate control instructions to adjust the attitude and speed of the drone so that it flies along the center line of the railway tunnel.

[0121] The PID control algorithm is as follows:

[0122] PID control is a classic feedback control algorithm that adjusts the control quantity through three links: proportional, integral and differential.

[0123] Proportional (P) control: Generates control quantity according to the deviation size.

[0124] Integral (I) control: Generates control quantity based on the accumulation of historical deviations.

[0125] Derivative (D) control: Generates the control amount based on the rate of change of the deviation.

[0126] PID control formula:

[0127]

[0128] Where:

[0129] u(t) is the control quantity;

[0130] e(t) is the deviation;

[0131] Kp, Ki, Kd are the proportional, integral and differential coefficients respectively.

[0132] Step 203: When the UAV needs to return during the operation, it descends below the contact network plane.

[0133] Step 204: The UAV returns according to the take-off path, the operation path and the landing path.

[0134] During the operation, when encountering an emergency, the operator only needs to lower the drone to below the contact network plane and start the one-key return command. The drone can automatically return to the preset take-off and landing point (evacuation hole) and land safely and accurately according to the previously planned path (including take-off path, operation path and landing path, etc.). This temporary one-key return function greatly improves the convenience and safety of the operation, reduces the possibility of human operation errors, and also reflects the high degree of automation of the entire inspection method.

[0135] In practical applications, the above operation method is based on Figure 3 The railway tunnel UAV automatic inspection system shown is implemented, which includes a UAV platform, a data processing center, a ground control station and a communication link.

[0136] The UAV platform is the core equipment for performing inspection tasks, including the UAV body, flight control system, positioning and navigation module, environmental perception module, mission payload module, data processing center and energy supply module. The UAV body adopts a structural design suitable for railway tunnel environment operation, with good maneuverability and stability; the flight control system is responsible for accurately controlling the flight attitude, speed, position and other parameters of the UAV to ensure safe and stable flight; the positioning and navigation module is mainly a combination system that integrates multiple positioning technologies. In view of the fact that satellite signals may be blocked in some areas of the railway tunnel, the UAV is provided with accurate real-time position and attitude information through the complementarity of multiple positioning technologies; the environmental perception module contains multiple sensors such as laser radar, camera, millimeter wave radar, etc., which are used to detect the environmental conditions in the railway tunnel in real time, including the location of obstacles, the distance from the tunnel wall, the status of cables and rails, etc., to provide key data support for flight safety and mission execution; the mission payload module is equipped with corresponding equipment according to the inspection needs, such as high-definition cameras for shooting the internal facilities of the tunnel, thermal imagers for detecting abnormal cable temperature, etc.; the energy supply module provides continuous and stable power for the UAV to ensure sufficient energy throughout the inspection task. The data processing center is the data processing and storage core of the entire system, responsible for in-depth processing and analysis of the data collected by the drone. It includes data storage servers, data processing software modules, etc. The data storage server is used to store the original data collected by the drone, the processed result data, and the tunnel environment model and other information. The data processing software module further processes the collected images, point clouds and other data, such as more accurate identification of tunnel elements.

[0137] The ground control station is an operating platform for operators to interact with drones. It has functions such as human-machine interaction interface, flight command control module, data receiving and display module, and mission planning module. Operators remotely control drones through the ground control station, including sending instructions such as takeoff, landing, inspection path planning, and mission execution. At the same time, the ground control station receives real-time data sent back by the drone, such as flight status data, sensor collection data, mission execution data, etc., and displays it intuitively, so that operators can monitor the operation of the drone in real time and perform manual intervention when necessary.

[0138] The communication link is responsible for data transmission between the UAV platform and the ground control station and data processing center. Wireless communication technology, such as high-speed Wi-Fi communication or dedicated wireless data transmission links, is used to ensure the real-time, stability and reliability of data transmission. In some areas with weak signals in railway tunnels, signal relay equipment or distributed antenna systems are set up to ensure smooth communication links, so that the UAV and the ground control station can exchange commands and data in real time, and the data processing center can also obtain the data collected by the UAV in a timely manner for processing.

[0139] The path planning and flight control process describes in detail how Figure 4 The drone shown here goes through the entire process from preparing to perform an inspection mission to landing after completing the mission. The following are the detailed steps and explanations:

[0140] Step 1: System startup and self-test. Start the UAV platform, ground control station and communication link, and perform initialization and self-test operations on each device. The UAV platform checks the working status of the flight control system, sensors, mission payloads and other equipment, as well as the server operation status and software module availability of the data processing center. The ground control station checks whether the functions of the command control module, data receiving and display module are normal. The communication link performs signal strength test and data transmission stability test to ensure that the entire system is in normal working condition.

[0141] Step 2: Tunnel environment data collection and preprocessing. Before taking off or flying in the railway tunnel according to the preset route, the drone uses the environmental perception module to collect environmental data in the tunnel, including lidar point cloud data, camera image data, millimeter wave radar echo data, etc. The collected data is transmitted to the data processing center in real time through the communication link. The data processing center preprocesses the data, such as registration, denoising, and simplification of point cloud data, feature extraction and enhancement of image data, etc., to provide a high-quality data foundation for subsequent environmental modeling and element recognition.

[0142] Step 3: Tunnel environment modeling and accurate element identification. Based on the pre-processed environmental data, the data processing center uses advanced algorithms to build a three-dimensional environmental model of the railway tunnel and accurately identify various key elements in the tunnel, such as cables, rails, escape tunnels, and disease characteristics. Through deep learning algorithms, image data and point cloud data are analyzed to accurately determine the type, direction, and connection status of cables, the degree of wear and flatness of rails, and cracks, water leakage, and other diseases of the tunnel wall. The identification results are integrated with the three-dimensional environmental model to provide detailed and accurate environmental information for path planning.

[0143] Step 4: Inspection path planning and optimization. According to the inspection task requirements and the tunnel environment model, the data processing center uses the A* algorithm for path planning to plan the inspection path (take-off path, operation path, and landing path) of the drone. Considering factors such as the flight performance of the drone, mission load restrictions, and obstacle distribution in the tunnel, one or more safe and efficient inspection paths are generated. In the path planning process, technologies such as bidirectional search strategies and dynamic weight adjustment are used to optimize path length, smoothness, and safety.

[0144] Step 5: UAV takeoff and initial path flight. After receiving the takeoff command, the UAV takes off from the takeoff and landing platform (tunnel shelter) according to the pre-set takeoff procedure and initial planned path. The flight control system controls the UAV to fly along the initial path based on the location information and path planning instructions provided by the positioning and navigation module. At the same time, the environmental perception module monitors the surrounding environment in real time to ensure the safety of the takeoff process. The initial path is smoothed using algorithms such as the Bézier curve to make the UAV fly more smoothly.

[0145] Step 6: Inspection operation execution and dynamic obstacle avoidance. The drone flies along the planned path. After arriving at the inspection area, it performs inspection tasks according to the instructions of the mission payload module, such as taking high-definition images, recording videos, collecting temperature data, etc. During the entire flight process, the environmental perception module continuously monitors the environment inside the railway tunnel and detects obstacles (including dynamic and static obstacles) in real time. Once an obstacle is found, the dynamic obstacle avoidance mechanism is immediately triggered to avoid the obstacle by adjusting the flight attitude, speed and path. Combined with sensor fusion technology, the data of sensors such as lidar, camera, and millimeter-wave radar are integrated to improve the accuracy and reliability of obstacle detection.

[0146] Step 7: Flight status monitoring and path adjustment. The ground control station receives the flight status data sent back by the drone in real time, including information such as position, attitude, speed, and power, and monitors the execution of the task. The data processing center analyzes the data collected by the drone in real time to determine whether the inspection task is proceeding normally and whether there are any abnormalities in the tunnel facilities. If the drone is found to deviate from the predetermined path, encounter sudden environmental changes, or abnormal task execution, the ground control station or data processing center will generate adjustment instructions in a timely manner, and the drone will adjust the flight path and task execution strategy in real time according to the instructions to ensure the smooth completion of the inspection task.

[0147] Step 8: Mission completion and landing. When the drone completes all inspection tasks, it will fly to the take-off and landing platform and land according to the pre-planned landing path or temporary instructions. After landing, the system will summarize the tasks and organize the data, archive and store the data collected during the inspection and the execution of the tasks, providing a basis for subsequent data analysis and tunnel maintenance.

[0148] In practical applications, in addition to using the laser radar module in the environmental perception module for positioning, vision-based positioning methods can also be used, such as simultaneous localization and mapping (SLAM) technology. The camera on the drone is used to capture images of the interior of the railway tunnel. By extracting and matching image features, a tunnel map is constructed and the drone's position in the map is determined at the same time. Combined with sensor data such as the inertial measurement unit (IMU), the accuracy and stability of positioning can be improved.

[0149] In practical applications, in addition to laser radar and cameras as obstacle avoidance sensors, millimeter wave radar can also be used as the main obstacle avoidance sensor. Millimeter wave radar can emit electromagnetic waves in the millimeter wave frequency band and receive reflected waves to detect the distance, speed and angle information of obstacles. Through multi-beam scanning technology, all-round monitoring of the surrounding environment of the tunnel can be achieved.

[0150] Embodiment 3

[0151] The automatic path planning device for a railway tunnel drone of this embodiment includes:

[0152] An acquisition module is used to acquire the railway tunnel environment model, preset take-off and landing points, and the operation start point and operation end point;

[0153] The first planning module is used to plan the take-off path between the preset take-off and landing point and the operation start point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation start point and the operation end point by using the A* algorithm and a two-way search strategy. During the two-way search process, a comprehensive distance heuristic function is used to guide the search direction. The comprehensive distance heuristic function introduces the ratio of the first distance between the current node and the target node to the second distance between the start node and the target node.

[0154] The specific implementation of this device can be found in Example 1, which will not be described in detail here.

[0155] It is worth mentioning that all modules involved in this embodiment are logic modules. In practical applications, a logic unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed by this application, but this does not mean that there are no other units in this embodiment.

[0156] Embodiment 4

[0157] The railway tunnel drone operation device of this embodiment includes:

[0158] A second planning module is used to obtain a take-off path, a landing path, and an operation path according to the automatic path planning device of the railway tunnel UAV;

[0159] The operation module is used for the UAV to start operation according to the take-off path, the operation path and the landing path.

[0160] The specific implementation of this device can be found in Example 2, which will not be described in detail here.

[0161] It is worth mentioning that all modules involved in this embodiment are logic modules. In practical applications, a logic unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed by this application, but this does not mean that there are no other units in this embodiment.

[0162] Embodiment 5

[0163] Another embodiment of the present application relates to an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the methods in the above-mentioned embodiments.

[0164] Among them, the memory and the processor are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor.

[0165] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0166] Embodiment 6

[0167] Another embodiment of the present application relates to a computer-readable storage medium storing a computer program, which implements the above method embodiment when executed by a processor.

[0168] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as: ROM), random access memory (Random Access Memory, referred to as: RAM), disk or optical disk and other media that can store program codes.

[0169] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present application, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present application.

Claims

1. An automatic path planning method for a UAV in a railway tunnel, characterized in that: include: Obtain the railway tunnel environment model, preset take-off and landing points, and operation start and end points; The A* algorithm is used and a bidirectional search strategy is adopted to plan the take-off path between the preset take-off and landing point and the operation starting point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation starting point and the operation end point. During the bidirectional search process, a comprehensive distance heuristic function is used to guide the search direction. The comprehensive distance heuristic function introduces the ratio of the first distance between the current node and the target node to the second distance between the starting node and the target node.

2. The automatic path planning method for a railway tunnel UAV according to claim 1 is characterized in that: The preset take-off point and the preset landing point are located in the escape tunnel of the railway tunnel or around the entrance and exit of the railway tunnel.

3. The automatic path planning method for a railway tunnel UAV according to claim 1 is characterized in that: Before obtaining the railway tunnel environment model, preset take-off and landing points, and the operation start point and operation end point, it also includes: Obtain environmental data for railway tunnels; Based on the environmental data, an environmental model of the railway tunnel is constructed and key elements in the railway tunnel are identified.

4. The automatic path planning method for a railway tunnel UAV according to claim 3 is characterized in that: The A* algorithm is used to plan the take-off path between the preset take-off and landing point and the operation start point, the landing path between the operation end point and the preset take-off and landing point, and the operation path between the operation start point and the operation end point by adopting a bidirectional search strategy, including: According to the key elements in the railway tunnel and the operation requirements of the UAV, the take-off path, the landing path, and the operation path are planned respectively by using the A* algorithm and a bidirectional search strategy; The key elements in the railway tunnel include at least one of cables, rails, and escape tunnels.

5. The automatic path planning method for a railway tunnel UAV according to claim 1, characterized in that: The expression of the comprehensive distance heuristic function is as follows: h(n)=1.2×ω(n)×d(n)-u(n) x(n)=|x g -x n | y(n)=|y g -and n | min x,y =min(|x g -x n |,|y g -y n |) In the formula, h(n) represents the comprehensive distance heuristic function, ω(n) represents the dynamic weight, d(n) represents the comprehensive distance, u(n) represents the ratio of the first distance between the current node and the target node to the second distance between the starting node and the target node, (x g ,y g ) represents the coordinates of the target node, (x n ,y n ) represents the coordinates of the current node, x(n) represents the distance between the coordinates of the current node and the target node in the horizontal direction, and y(n) represents the distance between the coordinates of the current node and the target node in the vertical direction.

6. The automatic path planning method for a railway tunnel UAV according to claim 5 is characterized in that: When the comprehensive distance is greater than a preset value, the dynamic weight ω(n) is set to a first value; When the comprehensive distance is less than or equal to a preset value, the dynamic weight ω(n) is set to a second value; The first value is greater than the second value.

7. The automatic path planning method for a railway tunnel UAV according to claim 1, characterized in that: Also includes: Remove redundant turning points from the planned takeoff path, landing path, and operation path; The Bézier curve is used to smooth the takeoff path, landing path and operation path after removing redundant inflection points.

8. A method for operating a UAV in a railway tunnel, characterized in that: include: Obtaining a take-off path, a landing path, and an operation path according to the automatic path planning method for a railway tunnel UAV according to any one of claims 1 to 7; Using the line-biting flight algorithm to control the UAV to start operation according to the take-off path, the operation path and the landing path; In case of need to return during operation, the drone descends below the contact network plane; The UAV returns according to the take-off path, the operation path and the landing path.

9. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images