Path planning method for mine rescue robot

Abstract

The invention discloses a path planning method for a mine rescue robot, relates to the field of path planning of an intelligent algorithm, plans an optimal path of the mine rescue robot on a rasterized map, and aims to solve the problems of low search efficiency, multiple turning times, path redundancy and easiness in falling into a local optimal solution in the initial stage of iteration. The method comprises the following steps: establishing a grid map environment model; changing a heuristic function of a pseudo-random probability ACS algorithm to obtain an initial path; performing non-uniform distribution on initial pheromones through the initial path, and performing iteration; obtaining a neighborhood search mechanism according to a map environment, and selecting path nodes to form a path; smoothing each path successfully reaching the end point; a pheromone reward and punishment mechanism is added, and arrangement of pheromones is changed; and next iteration is carried out to obtain a final optimal path. The method is suitable for path planning in the fields of Internet of Things, cloud computing and the like, and an innovative solution is provided for path planning of the mine rescue robot.

Description

Technical Field

[0001] The present application relates to the field of path planning of intelligent algorithms, and specifically to a path planning method for a mine rescue robot. Background Art

[0002] Mines are well-known disaster-prone areas, which may cause accidents such as landslides due to various reasons. After a mine disaster occurs, it is crucial to quickly launch rescue work, but the environment under the mine may be cluttered with obstacles and secondary landslides. If rescuers enter rashly, it may cause secondary disasters and increase the risk of casualties. Therefore, when a disaster occurs, if advanced emergency rescue equipment can be used to carry out efficient rescue at the best rescue time after the accident, the difficulty of rescue can be greatly reduced and the success rate of rescue can be increased. Emergency rescue robots have gradually entered people's field of vision. Rescue robot technology mainly includes environmental perception, positioning, path planning, decision-making and other parts. Among them, in order to enable the rescue robot to reach the disaster site safely and efficiently, the first problem is how to analyze and calculate the optimal motion path in the map, so the research on accurate and efficient path planning algorithms is particularly important. The performance of the algorithm directly determines the robot's autonomous navigation ability and task completion efficiency in a complex environment.

[0003] The path planning problem of the rescue robot can be simply described as finding an optimal and collision-free path in the workspace. The optimal path mainly achieves the goal of shortest time consumption, generally including the shortest path and the least number of turns. Path planning mainly includes two parts: environment modeling and path search. The grid method is one of the most widely used methods in robot path planning. Generally speaking, it can significantly reduce the complexity of the environment model. Summary of the invention

[0004] The present invention solves the problems of slow search efficiency, many turns, redundant paths, and easy falling into local optimal solutions in the initial iteration, and proposes a path planning method for mine rescue robots. This is achieved by changing the heuristic function of the pseudo-random probability ACS algorithm. The improved ACS algorithm plans the initial path and obtains the corresponding initial pheromone distribution to prevent blind search in the early stage of the algorithm; the directionality of the algorithm is improved and the deadlock rate is reduced by improving the heuristic function and neighborhood search mechanism; a smoothing mechanism is introduced to reduce the redundant nodes and the number of turns on the path, reduce the turning angle, and shorten the turning time of the rescue robot; a pheromone reward and punishment mechanism is introduced to improve the search efficiency.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: a mine rescue robot path planning method, comprising the following steps: Establish a raster map environment model; Change the heuristic function of the pseudo-random ACS algorithm to obtain the initial path; Non-uniformly distributing the initial pheromone through the initial path, and iterating; According to the map environment, a neighborhood search mechanism is derived to select path nodes and form a path; Smooth each path that successfully reaches the end point; Add pheromone reward and punishment mechanism and change the arrangement of pheromones; Carry out the next iteration to obtain the final optimal path.

[0006] In this technical solution, the heuristic function is changed to make it suitable for grid maps. First, the pseudo-random ACS algorithm, a variant of the traditional ACO algorithm, is improved by changing its heuristic function and changing the iterative algorithm to a deterministic algorithm, thereby obtaining an initial path. Then, the initial pheromones are non-uniformly distributed according to the initial path, the initial parameters are defined, and iteration begins.

[0007] During iteration, path nodes are selected based on the improved heuristic function that introduces endpoint and directional information and the six-neighborhood search mechanism to form a complete path, which is then compared with the optimal path in the current iteration and updated to prevent blind search in the early stage of the algorithm and reduce the deadlock rate.

[0008] Finally, the path is smoothed according to the smoothing mechanism, and the current optimal smoothed path is updated. After completing one iteration, the pheromone matrix is ​​updated according to the pheromone reward and punishment mechanism that introduces the adaptive dynamic pheromone volatilization factor and the unsmoothed optimal path to start the next iteration. After reaching the number of iterations, the optimal smoothed path is output, which shortens the turning time of the rescue robot and improves the search efficiency.

[0009] Furthermore, the map environment is a grid map, which makes it easier to select the nearest node.

[0010] Furthermore, in the pheromone reward and punishment mechanism, an adaptive dynamic pheromone volatilization factor is introduced to change the arrangement of pheromones, making it easier to find an excellent path.

[0011] Furthermore, the heuristic function is applied to the ACO algorithm to introduce endpoint and directionality information, thereby improving the directionality of the algorithm.

[0012] Furthermore, the nearest node is selected by the pseudo-random probability, and the pseudo-random probability can automatically select the best channel. The pseudo-random probability state transition rule is: Where j is the next mobile node, q is a random variable parameter between [0,1], q0 is a predefined initial parameter, α is the pheromone importance factor, β is the heuristic function importance factor, J m is the set of free nodes that ant m can choose.

[0013] Furthermore, the initial path distributes the initial pheromones non-uniformly, thereby reducing the blindness of the search.

[0014] Furthermore, the search mechanism is a six-neighborhood search mechanism. The six-neighborhood search mechanism is derived according to the map environment, and the six-neighborhood search mechanism reduces the possibility of the path falling into deadlock.

[0015] Furthermore, the pheromone reward and punishment mechanism includes comparing the path length and the number of turns of the path, and introducing a smoothing mechanism to reduce redundant nodes and the number of turns of the path.

[0016] Furthermore, the heuristic function converts the iterative algorithm into a deterministic algorithm to generate a better initial path.

[0017] Furthermore, the improved heuristic information formula of the heuristic function is: Where, d jT′ Indicates the distance between the optional node j at the current position i and the virtual node T′, where the virtual node T′ is the node adjacent to T on the extension line of the line connecting the actual end point T and the starting point S.

[0018] Furthermore, the reward and punishment mechanism increases the pheromone concentration of the unsmoothed path with the shortest length and the least number of turns in each iteration, and the reward and punishment mechanism rules are: Where ρ is the adaptive dynamic pheromone volatility coefficient, L Pb is the length of the shortest path in the current iteration, t b is the number of turns of the path with the least number of turns in the current iteration, t w is the number of turns of the path with the most turns in the iteration, L Tb is the length of the path with the least number of turns in that iteration.

[0019] Furthermore, the path node selection rule is: The path node selection retains randomness to expand the search space.

[0020] Compared with the prior art, the present invention has the following beneficial effects: A new heuristic function is proposed to improve the pseudo-random ACS algorithm into a deterministic algorithm suitable for grid maps, and the suboptimal solution is obtained and used to arrange the initial pheromones. Introducing endpoint and directionality information, changing the neighborhood search mechanism, making it more suitable for actual situations and reducing unnecessary exploration directions; Smooth each path that successfully reaches the end point, reduce unnecessary turning nodes, reduce turning angles, make the path smoother, and conform to the actual situation of the rescue robot; Apply pheromone reward and punishment mechanism and adaptive pheromone volatility coefficient to enhance the guidance ability of subsequent iterations. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The figure is a flow chart of the path planning method for a mine rescue robot based on a pseudo-random initial path.

[0022] Figure 2 This is the expansion processing diagram of irregular obstacles in the raster map.

[0023] Figure 3 Two route maps from the starting point to the end point.

[0024] Figure 4 is a pseudo-random probability initial path map.

[0025] Figure 5 The suboptimal paths and their initial pheromone distributions for a 20×20 map.

[0026] Figure 6 The 8-neighborhood mechanism was changed to a 6-neighborhood mechanism diagram.

[0027] Figure 7 Schematic diagram of path smoothing mechanism. DETAILED DESCRIPTION

[0028] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0029] Embodiment 1: like Figure 1 As shown, a flow chart of a mine rescue robot path planning method based on a pseudo-random initial path according to Embodiment 1 of the present invention is shown. As shown in the figure, the process mainly includes the following steps.

[0030] Establishing a grid map simulation environment model: In actual rescue scenarios, the obstacles encountered by rescue robots are often irregular. In a two-dimensional grid map, the robot is simplified to a point moving in a plane space. Therefore, in order to ensure the safety of the planned path, the actual irregular obstacles must be expanded and turned into regular square obstacles. This increases the safe distance between the robot and the obstacles and avoids collisions. In the grid method modeling map, the expansion of irregular obstacles is as follows: Figure 2 shown.

[0031] Get the initial path: The ACS algorithm is proposed to solve the TSP problem. Therefore, in the heuristic function and state transition probability mechanism, the ant is biased to select the next nearest node at the current node. Figure 3 As shown, when the ant starts from the yellow starting point, there are three directions to choose from, and the path lengths are 1, 1. When applying the pseudo-random probability state transition rules of the ACS algorithm, equations 1 and 2, the probability of selecting a node with path 1 is much greater than that of selecting a node with path The probability of selecting nodes at 45° is 0.04, but it is obvious that the path formed by selecting nodes in the 45° direction is shorter and more in line with the purpose of path optimization. Therefore, the heuristic function is improved to make it suitable for grid maps. This function is also used in the improved ACO algorithm proposed later. The proposed improved heuristic information is shown in Formula 4.

[0032]

[0033]

[0034]

[0035] Among them, j is the next mobile node, q is a random variable parameter between [0,1], q0 is a predefined initial parameter, argmax(x) is a function that selects the node corresponding to the maximum x value, and J is a movable node randomly selected by roulette. is the transition probability of each optional path of ant m from node i to node j, τ(i,j) is the pheromone concentration, η(i,j) is the heuristic function, α is the pheromone importance factor, β is the heuristic function importance factor, J m is the set of free nodes that ant m can choose.

[0036]

[0037] Where, d jT′It indicates the distance between the optional node j at the current position i and the virtual node T'. The virtual node T' is the node adjacent to T on the extension line of the line connecting the actual end point T and the starting point S. Rows indicates the total number of rows in the grid map. The improved heuristic information enhances the guidance of the end point to the ant colony, reduces the blindness of the search, and improves the search efficiency of the algorithm.

[0038] In the pseudo-random state transition probability, there is an initial threshold q0. Generally, q0 is predefined to be close to 1, so that it is easy to select the node with the largest heuristic function value when selecting a node, while retaining a certain degree of randomness to expand the search space. In the present invention, the initial parameter q0 is defined as 1, that is, the random algorithm is changed to a deterministic heuristic algorithm, the iteration is cancelled, and the path solution is obtained. The node selection rule is shown in Formula 5, as Figure 4 , in a 10×10 grid graph, a better initial path is obtained using the improved deterministic algorithm.

[0039]

[0040] Initial pheromone distribution: The initial pheromone is redistributed according to the initial path, which increases the guidance of the ants in the algorithm to find the route, and the pheromone concentration on the feasible nodes near the suboptimal solution is linearly distributed according to the distance, so that in the path-finding process, there is still a good search ability for other path solutions, and it is not easy to fall into the possibility of taking the suboptimal path as the local optimal solution. The calculation rule of the pheromone concentration on the feasible edge near the suboptimal path is shown in Formula 6.

[0041]

[0042] In the formula, Δτ ik is the pheromone concentration on the edge between node i and its surrounding feasible node j on the suboptimal path, τ sec is the pheromone concentration on the suboptimal path, and its magnitude is Q / L sec ,δ is the pheromone concentration deviation constant, rows is the total number of rows in the grid map, d ik is the Euclidean distance between node i and node k. It should be noted that the starting point in the suboptimal path does not release pheromones, which reduces the probability of ants returning to the starting point in subsequent algorithm iterations and increases the outward search space. In the 20×20 random obstacle grid map, the initial pheromone distribution obtained according to the suboptimal path is as follows: Figure 5 As shown, the pheromone concentration is proportional to the brightness of red. In the edge blue area, the pheromone value is the initial pheromone value "1" in the traditional ACO algorithm.

[0043] Neighborhood search mechanism: In traditional grid map path planning, most of them adopt an 8-neighborhood search mechanism, that is, in the current grid, the eight surrounding grid nodes are all transferable nodes. Figure 5In the figure, the direction information between the starting point S and the target point T is represented by a dotted line and an arrow. If there are no obstacles, the straight line between the two points is the shortest path. The closer the path is to the ST line, the shorter it is; conversely, the further it deviates from the ST line, the longer the path length is. Therefore, the present invention makes the route as close to the line between the two points as possible, that is, the ant needs to walk in the upper right direction. In this case, a rectangular coordinate system is established with the starting point S as the origin, which requires that the horizontal and vertical coordinates of the next free node selected by the ant cannot all be smaller than the horizontal and vertical coordinates of the node where the ant is located. Figure 6 In (b), when selecting the next node at point P, there should be a tendency to move to the upper right along the ST line. If the lower left and lower directions, that is, the direction towards the origin, are selected, the possibility of ants returning to the starting point, resulting in a longer redundant path and a "deadlock" will be increased. In addition, the number of iterations to find the optimal solution will increase greatly due to the reduction in the number of successful ants and the increase in pheromones on unnecessary paths. After removing these two search directions, the search efficiency is improved, which will significantly reduce the number of iterations required to reach the optimal path.

[0044] Each path that successfully reaches the destination is smoothed according to the path smoothing mechanism: In the 8-neighborhood algorithm of the traditional grid, the turning angles at the turning nodes of the path are only 45° and 90°, which not only leads to the redundancy of the number of turns and increases the path length, but also causes the actual rescue robot to turn too long, dragging down the timeliness of the search and rescue. When the path lengths are equal, the path with more turns will take longer. Therefore, the present invention proposes a smoothing mechanism that greatly reduces the turning nodes and turning angles, making the rescue path smoother, taking less time, and saving rescue time. The pseudo code of the path smoothing mechanism is as follows: As shown in Table 1. Table 1 Pseudo code of path smoothing mechanism

[0045] like Figure 7 As shown, the black dotted line is the unsmoothed path, and the red solid line is the smoothed path. It is obvious that the number of turns has been reduced from 5 to 2, and the turning angle has also been greatly reduced.

[0046] Pheromone reward and punishment mechanism and adaptive dynamic pheromone volatility coefficient: The pheromone update of the traditional ACO algorithm is to increase the pheromone concentration on the edge of each path through equations 7 and 8 according to the length of all the paths walked by the ants, including the paths that reach the end point and the paths that do not reach the end point, after each iteration.

[0047]

[0048]

[0049] The traditional pheromone update mechanism does not highlight good paths and has no guidance for the optimal path of the current iteration. Instead, the short paths that are partially trapped in "deadlock" accumulate higher pheromone concentrations on this path due to their short length, which provides the wrong direction for subsequent iterative optimization, causing the algorithm to fall into local optimality and greatly increasing the number of failed ants in the ant colony. The present invention proposes a new pheromone reward and punishment mechanism, which uses a reward rule to increase the pheromone concentration on the path with the shortest length and the least number of turns in the unsmoothed path obtained in each iteration, and guides the direction for subsequent iterations. In addition, the ants that are trapped in "deadlock" in the iteration are abandoned, and their paths are not updated except for global volatilization, so as to avoid blind search of the ant colony in the next iteration. For other common paths, the pheromone update rules still refer to Formula 7 and Formula 8. The reward and punishment mechanism is shown in Formula 9 and Formula 10.

[0050]

[0051]

[0052] Where ρ is the adaptive dynamic pheromone volatility coefficient, calculated as in Equation 11, ε is a random number in the interval [1.05, 1.2], used to increase the pheromone concentration of the shortest path within the iteration, L Pb is the length of the shortest path in the current iteration, t b is the number of turns of the path with the least number of turns in the current iteration, t w is the number of turns of the path with the most turns in the iteration, L Tb is the length of the path with the least number of turns in that iteration.

[0053]

[0054] Among them, rows is the total number of rows in the grid map, γ is a constant coefficient with a value of 0.6, and route m is the walking path of the ants that reach the end point in each iteration, and M is the number of ants that successfully reach the end point. The shorter the average path of the iteration, the larger ρ is, the greater the positive feedback effect is, and the better the path of this iteration is. The optimal path alone increases pheromones, and other suboptimal paths can also obtain greater positive feedback effects to avoid falling into local optimality. In the early stage of the algorithm, the average path is long and ρ is small. Then, the difference in pheromone concentrations of each path in the iteration is small, and the guiding role of the ant colony is weakened, which enhances the global search range of the ant colony and the accuracy of the algorithm; as the average path in the iteration becomes shorter and shorter, the pheromone volatilization factor ρ also increases, which increases the difference in pheromone concentrations of each path, improves the search speed of the ant colony, and makes the algorithm converge quickly.

[0055] The present invention changes the pheromone volatilization factor with a constant coefficient in the traditional ACO algorithm to one that changes size adaptively according to the path length, and designs a pheromone reward and punishment mechanism based on the path length and the number of turns of the path, highlighting excellent paths. Compared with the previous traditional ant colony algorithm, the invention has fewer iterations, a shorter path length, and fewer turns.

[0056] Embodiment 2: Based on Example 1, the above-mentioned mine rescue robot path planning method is described in more detail.

[0057] like Figure 1 As shown, the following steps are included: The heuristic function used for selecting nodes in the present invention adds end point and directionality information according to the characteristics of the grid graph, and changes the eight-neighborhood search mechanism into a six-neighborhood search mechanism. The six-neighborhood search mechanism reduces the amount of data that needs to be processed during the indexing and retrieval process, reduces the complexity of the search and the difficulty of selection, helps to improve the search efficiency, and reduces the possibility of the path falling into a deadlock. After the environmental model is established with the grid graph, the parameters and conditions are initialized. First, the variant of the traditional ACO algorithm, the pseudo-random probability ACS algorithm, is improved. The ACO algorithm is a swarm intelligence optimization algorithm that simulates the foraging behavior of ants in nature. The algorithm has a high complexity, which leads to a relatively slow convergence speed of the algorithm, and it is easy to fall into a local optimum. Stagnation is prone to occur during the search process, that is, all ants are concentrated on a certain path and cannot further search the solution space, thus falling into a local optimal solution. It is difficult to find a global optimal solution and is easily affected by multiple parameters, such as the pheromone importance factor, the heuristic function importance factor, and the information The selection of these parameters usually depends on experience, and it is difficult to achieve the optimal setting, which in turn affects the overall performance of the algorithm. In the initial stage of the algorithm, due to the same or low pheromone concentration, the selection of ants is more random, which may lead to low search efficiency. The characteristic of the pseudo-random probability ACS algorithm is that pseudo-random probability is added when selecting path nodes, which can make it easier to select the nearest node. After improving the pseudo-random probability ACS algorithm, its heuristic function is changed, and the iterative algorithm is changed to a deterministic algorithm, thereby obtaining an initial path. The ACS algorithm can automatically select the best channel, automatically evaluate the load and interference of different channels, and select the most suitable channel to provide better wireless connection without manual intervention. By selecting channels with low utilization and low interference, it helps to reduce network congestion and interference, thereby improving the overall network performance and stability. The automatic channel selection feature eliminates the need for users to have an in-depth understanding of the complex configuration of the wireless network, which can reduce the complexity of the configuration.

[0058] The initial pheromone of the traditional ACO algorithm is uniformly distributed as "1". After obtaining the initial path, the present invention distributes the initial pheromone non-uniformly according to the initial path to obtain a non-uniformly distributed initial pheromone matrix. After defining the initial parameters, the iteration begins. According to the improved heuristic function that introduces the end point and directional information and the six-neighborhood search mechanism, the path nodes are selected to form a complete path, which is then compared with the optimal path in the current iteration and updated. The grid map is composed of pixel points, each pixel point carries specific information, making the data structure intuitive and easy to understand. The information is displayed through the arrangement and combination of pixel points, which is convenient for quickly understanding and comparing the differences between different data or images. By encoding squares of different colors, sizes or filling methods, the grid map can display complex data distribution and characteristics, and its resolution is Depending on the size of the cell, it can be flexibly adjusted to meet different display and analysis requirements. Under high resolution, the calculation and processing complexity of the grid map will increase significantly, and there are certain requirements for hardware performance. Due to the characteristics of the grid map, the path angle formed is 45° or 90°, and there is room for further shortening the path. The present invention designs a smoothing mechanism to make the path shorter, with fewer turns and smaller turning angles without touching obstacles. Then, the path is smoothed according to the smoothing mechanism, and the current optimal smoothed path is updated; after completing one iteration, the pheromone matrix is ​​updated in combination with the unsmoothed optimal path according to the pheromone reward and punishment mechanism that introduces the adaptive dynamic pheromone volatilization factor, so as to start the next iteration; after reaching the number of iterations, the optimal smoothed path is output.

[0059] Among them, at the beginning of the iteration, the ants are released, and the ants start to find the next node from the starting point according to the heuristic function to determine whether the ants have reached the end point. When the ants reach the end point, the path length and the number of turns are compared with the optimal path of the current iteration. When the ants have not reached the end point, the ants are released again. After comparing the path length and the number of turns, the path is smoothed according to the smoothing mechanism, and the optimal path after smoothing is updated. It is determined whether the number of ants is less than the population number. If it is less than the population number, the ants are released again. If it is greater than the population number, the pheromone matrix is ​​updated according to the reward and punishment mechanism in combination with the unsmoothed optimal path, the current global optimal path is updated, and the next iteration is performed until the optimal path is obtained.

[0060] The ACS algorithm provided by the present invention can be evaluated and selected according to different environments and conditions and has strong adaptability. These advantages enable the present invention to be widely used in wireless LAN management, especially in scenarios requiring automatic optimization of network performance and stability.

[0061] Verify the effectiveness and superiority of the present invention. Table 2 Experimental results of each algorithm in simulation diagram

[0062] Table 2 shows the performance of each algorithm in terms of optimal path degree, average path length, path length standard deviation, minimum number of iterations, average number of iterations, standard deviation of number of iterations and number of path turns. The bold numbers indicate the best. All of the above indicate the effectiveness and superiority of the PRIP-AACO algorithm model proposed in the present invention.

[0063] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc., which contain computer-usable program codes.

Claims

1. A mine rescue robot path planning method, characterized in that: The following steps are involved: Establish a raster map environment model; Change the heuristic function of the pseudo-random probability ACS algorithm to obtain the initial path; Non-uniformly distributing the initial pheromone through the initial path, and iterating; According to the map environment, a search mechanism is derived, and path nodes are selected to form a path; Smooth each path that successfully reaches the end point; Add pheromone reward and punishment mechanism and change the arrangement of pheromones; Carry out the next iteration to obtain the final optimal path.

2. A mine rescue robot path planning method according to claim 1, characterized in that: A pseudo-random probability is added when selecting the path node, and the map environment is a grid map.

3. A mine rescue robot path planning method according to claim 1, characterized in that: In the pheromone reward and punishment mechanism, an adaptive dynamic pheromone volatilization factor is introduced.

4. A mine rescue robot path planning method according to claim 1, characterized in that: The heuristic function is applied to the ACO algorithm to introduce endpoint and directionality information.

5. A mine rescue robot path planning method according to claim 2, characterized in that: The nearest node is selected by the pseudo-random probability, and the pseudo-random probability can automatically select the best channel.

6. A mine rescue robot path planning method according to claim 3, characterized in that: The pheromone reward and penalty mechanism includes comparing the path length and the number of turns in the path.

7. A mine rescue robot path planning method according to claim 4, characterized in that: The heuristic function converts the iterative algorithm into a deterministic algorithm to generate a better initial path.

8. A mine rescue robot path planning method according to claim 1, characterized in that: The search mechanism is a six-neighborhood search mechanism.

9. A mine rescue robot path planning method according to claim 3 or 6, characterized in that: The reward and penalty mechanism improves the pheromone concentration of the unsmoothed path with the shortest length and the least number of turns in each iteration.

10. A mine rescue robot path planning method according to claim 2, characterized in that: The path node selection rule is to select and retain randomness to expand the search space.

Images