Method and system for path planning based on her-aco

Abstract

The application provides a path planning method and system based on HER-ACO, belonging to the technical field of robot path planning. The method comprises the following steps: S1, obtaining an environment grid map, the grid map comprising a starting point, a target point and obstacle distribution information; S2, introducing a HER algorithm to improve a traditional ant colony algorithm ACO, and using the improved ant colony algorithm to search a feasible path from the starting point to the target point in the grid map as an initial planning path; the improvement comprises: introducing a HER idea and a dynamic pheromone evaporation mechanism; increasing a direction weight factor, a neighborhood obstacle factor and a future obstacle factor; S3, based on a straight line path feasibility check, smoothing the initial planning path, removing redundant intermediate nodes and outputting a final planning path. The simulation results show that the scheme has significant advantages in complex environments, especially in path quality and convergence speed, and is suitable for robot navigation, concrete robot path planning and other scenes.

Description

Technical Field

[0001] This application relates to the field of robot path planning technology, and in particular to a path planning method and system based on HER-ACO. Background Technology

[0002] Path planning plays a crucial role in many fields such as mobile robot control and autonomous driving. The Ant Colony Algorithm (ACO), as a heuristic intelligent algorithm, is widely used in path planning due to its parallelism and strong robustness.

[0003] However, traditional ant colony algorithms face the following challenges in robot path planning: the convergence speed drops sharply in complex obstacle environments, redundant path turns lead to mechanical losses, and the algorithms are not adaptable to dynamic obstacles and are prone to getting trapped in local optima.

[0004] Therefore, there is a need to provide an improved technical solution that addresses the shortcomings of the existing technology. Summary of the Invention

[0005] The purpose of this application is to provide a path planning method and system based on HER-ACO to solve or alleviate the problems existing in the prior art.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] This application provides a path planning method based on HER-ACO, including:

[0008] S1. Obtain an environmental grid map, the grid map including the starting point, the target point, and obstacle distribution information;

[0009] S2. The HER algorithm is introduced to improve the traditional Ant Colony Algorithm (ACO). The improved ant colony algorithm is then used to search for feasible paths from the starting point to the target point in a grid map, which are used as the initial planned paths. The specific improvements are as follows:

[0010] Based on the idea of ​​the HER algorithm, intermediate nodes that meet the preset evaluation conditions are selected as pseudo-targets. If the ant does not reach the target point, the effective sub-paths in the path from the starting point to the pseudo-target are identified, and the pheromone concentration is updated according to the effective sub-paths.

[0011] Calculate the direction weight factor based on the angle between the direction from the current node to the target point and the direction of movement of the candidate next node;

[0012] Calculate the neighborhood obstacle factor and the future obstacle factor; the neighborhood obstacle factor and the future obstacle factor reflect the obstacle distribution around the candidate node and the path in front of it, respectively;

[0013] The global pheromone is updated using a dynamically changing pheromone evaporation rate, which decreases as the number of iterations increases.

[0014] Based on the updated global pheromone, orientation weight factor, neighborhood obstacle factor, and future obstacle factor, the path selection probability of the traditional ant colony algorithm is improved to obtain a new path selection probability.

[0015] S3. Based on the feasibility check of the straight path, smooth the initial planned path, remove redundant intermediate nodes, and output the final planned path.

[0016] In some alternative embodiments, the new path selection probability is calculated as follows:

[0017] ,

[0018] In the formula, For ants From the current node Move to the candidate next node The probability of path selection. Ants In the current grid The set of selectable graticles from 8 neighboring graticles. This represents the pheromone enhancement coefficient. The larger the value, the deeper the impact on the initial pheromone, which in turn leads to a poorer global search capability of the algorithm. The expected heuristic information coefficient is the coefficient of interest; the larger the value, the stronger the ant's heuristic information. The closer the grid is to the target, the higher its tendency; express Current grid To the neighboring grid The concentration of pheromones along the connection path; express Current grid To the neighboring grid Heuristics on the connection path; Indicates the current grid To the neighboring grid Neighborhood obstacle factor; Indicates the directional weighting factor for normalization; This represents the future obstacle factor.

[0019] In some optional embodiments, selecting intermediate nodes that meet preset evaluation conditions as pseudo-targets specifically means:

[0020] The formula for calculating the pseudo-target evaluation value is as follows:

[0021] ,

[0022] In the formula, This is a pseudo-target evaluation value. To start from the beginning To the false target Euclidean distance, This represents the actual length of the sub-path the ant takes to reach the false target.

[0023] like If the value exceeds the preset evaluation threshold, the node is treated as a pseudo-target, and the length of the effective sub-path corresponding to the pseudo-target is recorded.

[0024] In some alternative embodiments, updating the pheromone concentration based on the effective sub-path specifically means:

[0025] Calculate the pheromone increment based on the length of the effective sub-path. The formula is as follows:

[0026] ,

[0027] In the formula, This is a constant gain coefficient, whose value is a constant greater than 0; This represents the actual length of the sub-path from which the ant reaches the false target.

[0028] In some alternative embodiments, the pheromone evaporation rate decreases exponentially with increasing iteration number, as expressed by:

[0029] ,

[0030] In the formula, The pheromone evaporation rate in the current iteration; It is the maximum evaporation rate. It is the volatility decay coefficient. It is the current iteration number. It is the maximum number of iterations.

[0031] In some optional embodiments, the direction weight factor is defined as the cosine of the angle between the direction from the current node to the target point and the motion direction of the next candidate node, and is normalized to obtain the normalized direction weight factor.

[0032] In some alternative embodiments, the neighborhood obstacle factor is specifically used to reflect the obstacle density in the local area surrounding the candidate next node, and is calculated as follows:

[0033] ,

[0034] In the formula, For the current node Candidate next node Neighborhood obstacle factor; The current grid The number of obstacle grids in the 8-neighborhood.

[0035] In some optional embodiments, the future obstacle factor is specifically used to evaluate the distribution of obstacles within a preset number of steps ahead of the candidate path, and the calculation formula is as follows:

[0036] ,

[0037] In the formula, Indicates the future obstacle factor. This represents the predicted penalty coefficient. The number of obstacles within a preset number of steps ahead of the candidate path.

[0038] In some optional embodiments, the smoothing process of the initial planned path based on the straight-line path feasibility check, removing redundant intermediate nodes, and outputting the final planned path includes:

[0039] For the initial planned path any intermediate node in ( If from arrive If all grid cells along the straight path are free grid cells, then delete the node. The straight path is generated using the Bresenham algorithm.

[0040] This embodiment provides a HER-ACO-based path planning system, which is used to execute the HER-ACO-based path planning method described in any of the above embodiments, including:

[0041] A construction unit is used to construct an environmental grid map, the grid map including a starting point, a target point, and obstacle distribution information;

[0042] The improved unit introduces the HER algorithm to improve the traditional Ant Colony Algorithm (ACO). The improved ACO algorithm is then used to search for feasible paths from the starting point to the target point in a grid map, serving as the initial planned path. Specific improvements are as follows:

[0043] Based on the idea of ​​the HER algorithm, intermediate nodes that meet the preset evaluation conditions are selected as pseudo-targets. If the ant does not reach the target point, the effective sub-paths in the path from the starting point to the pseudo-target are identified, and the pheromone concentration is updated according to the effective sub-paths.

[0044] Calculate the direction weight factor based on the angle between the direction from the current node to the target point and the direction of movement of the candidate next node;

[0045] Calculate the neighborhood obstacle factor and the future obstacle factor; the neighborhood obstacle factor and the future obstacle factor reflect the obstacle distribution around the candidate node and the path in front of it, respectively;

[0046] The global pheromone is updated using a dynamically changing pheromone evaporation rate, which decreases as the number of iterations increases.

[0047] Based on the updated global pheromone, orientation weight factor, neighborhood obstacle factor, and future obstacle factor, the path selection probability of the traditional ant colony algorithm is improved to obtain a new path selection probability.

[0048] The smoothing unit is used to smooth the initial planned path based on the feasibility check of the straight path, remove redundant intermediate nodes, and output the final planned path.

[0049] The technical solution of this application embodiment has the following beneficial effects:

[0050] By introducing several improvements to the traditional Ant Colony Algorithm (ACO), a comprehensive optimization of the pheromone update strategy, path selection mechanism, and path output quality in the path search process is achieved. Specifically, the improvements are as follows: Introducing the Herald Effect (HER) concept improves pheromone utilization efficiency during path planning, enhancing the rationality and directional guidance of path selection; the introduction of direction weight factors, neighborhood obstacle factors, and future obstacle factors improves obstacle avoidance capabilities and environmental adaptability; dynamically changing pheromone evaporation rates are used to update global pheromones, achieving a dynamic balance between exploration and development during pheromone updates, preventing premature or slow convergence; and straight-line connectivity detection is used to determine whether intermediate nodes can be skipped, eliminating unnecessary inflection points, shortening the actual walking path length, and optimizing the geometric quality of the final path, thus obtaining better path planning results in complex grid environments. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating a path planning method based on HER-ACO provided for some embodiments of this application.

[0052] Figure 2 This is a schematic diagram of the structure of a computer device.

[0053] Figure 3 The diagrams illustrate the construction of a raster map and path planning, where (a) is a diagram of a raster map, (b) is a diagram of a raster map where the target is unreachable, and (c) is a diagram of using the BFS algorithm to verify path reachability.

[0054] Figure 4This is a schematic diagram simulating the foraging process of ants using the traditional Ant Colony Algorithm (ACO). In the diagram, (a) represents ants starting from the nest to find food sources, (b) represents ants choosing paths as pheromone concentration changes, and (c) represents ants finding the optimal path.

[0055] Figure 5 Path planning results for Environment 1.

[0056] Figure 6 The convergence curve of the algorithm in Environment 1 is shown.

[0057] Figure 7 This is the path planning result for Environment 2.

[0058] Figure 8 The convergence curve of the algorithm in Environment 2 is shown.

[0059] Figure 9 This is the path planning result for Environment 3.

[0060] Figure 10 The algorithm convergence curve for Environment 3 is shown.

[0061] Figure 11 This is the path planning result for Environment 4.

[0062] Figure 12 The convergence curve of the algorithm in Environment 4 is shown. Detailed Implementation

[0063] For ease of explanation, the relevant concepts in the embodiments of this application are explained by way of example below, and such explanation does not constitute a limitation on the technical solution of this application.

[0064] Path planning refers to the process by which a mobile robot finds an optimal or feasible path from its starting point to its target point, avoiding obstacles and meeting certain constraints (such as the shortest path, the fewest turns, etc.).

[0065] The following is a brief explanation of the traditional ant colony algorithm.

[0066] Ant colony optimization (ACO) was first proposed in 1991. This algorithm simulates the foraging behavior of ants in nature. For example... Figure 4As shown, an obstacle exists between the ant nest and the food source. The ants have two paths to choose from: bypassing the top side of the obstacle (the upper path) and bypassing the bottom side (the lower path). When an ant sets off from its nest to find food, it releases a pheromone along its path and can sense pheromones released by other ants. The concentration of the pheromone indicates the distance of the path; a higher concentration means a shorter path. Typically, ants will preferentially choose the path with the higher pheromone concentration and release a certain amount of pheromone to increase the pheromone concentration along that path, creating a positive feedback loop. Ultimately, the ants find the optimal path from the nest to the food source—the shortest path.

[0067] When using the ant colony algorithm for robot path planning on a grid map, the path probability is selected as follows:

[0068] Assuming it is located in a grid (also known as a node). ants on Move within a raster environment (i.e., a raster map), based on the pheromones on each path within the raster environment. and the current grid Select the next grid cell Inspirational information , decide ants Probability of next path selection As shown in the following formula:

[0069] (2)

[0070] in, Ants In the current grid The set of selectable graticles from the 8 neighboring graticles; This represents the pheromone enhancement coefficient. The larger the value, the deeper the impact on the initial pheromone, which in turn leads to a poorer global search capability of the algorithm. This represents the expected heuristic information coefficient; the larger the value, the stronger the ant's heuristic information. The closer the grid is to the target, the higher its tendency; Indicates time, Indicates in Current grid To the next grid The concentration of pheromones along the connection path; Indicates in Current grid To the next grid Heuristics on the connection path are represented as follows:

[0071] (3)

[0072] In the formula, Represents grid To target grid The Euclidean distance is expressed as follows:

[0073] (4)

[0074] in, , For grid of , coordinate, , For target grid of , coordinate.

[0075] pheromone concentration The update is as follows:

[0076] Ant From the current grid To the neighboring grid This will update the local pheromone, as shown below:

[0077] (5)

[0078] in, For local pheromone evaporation rate, The pheromone concentration of each node's connection path at the initial moment is a very small positive constant.

[0079] When all ants reach the target grid When an iteration completes a search, the global pheromone value is updated using the following formula:

[0080] (6)

[0081] in, The global pheromone evaporation rate (abbreviated as global evaporation rate) represents the degree of pheromone evaporation. The formula for the global pheromone concentration increment is as follows:

[0082] , (7)

[0083] In the formula, The gain coefficient is a constant and A value greater than 0 indicates the total amount of pheromones released by the ant in one cycle; For ants The total length of the path traversed Indicates the current grid Next grid of candidates The edge.

[0084] The above is a brief description of the traditional ant colony algorithm.

[0085] The embodiments of this application will now be described with reference to the accompanying drawings.

[0086] The embodiments of this application can be applied to Figure 2 The computer device 200 shown may be, but is not limited to, mobile terminals such as mobile phones, tablets, handheld computers, and personal digital assistants (PDAs), smart home devices such as smart TVs and smart cameras, wearable devices such as smart bracelets, smartwatches, and smart glasses, or other desktop, laptop, notebook, ultra-mobile personal computer (UMPC), netbook, and smart screen computer devices.

[0087] like Figure 2 As shown, the computer device 200 may include one or more of the following components: a processor 201, a memory 203, a communication interface 202, and a communication bus 204. The memory 203 can be connected to the processor 201 via the bus 204. The bus can transfer data between the processor 201 and the memory 203. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0088] Processor 201 may include one or more processing cores. Processor 201 can connect to various parts within the electronic device 200 using various interfaces and lines. It performs various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 203, and by calling data stored in memory 203. For example, processor 201 may include an application processor (AP), a modem processor, a CPU, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a field-programmable gate array (FPGA), a programmable logic array (PLA), and / or a neural network processing unit (NPU). The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed; the NPU implements artificial intelligence (AI) functions; and the modem handles wireless communication. Different processing units can be independent devices or integrated into one or more processors. For example, the multiple processing units shown above are all integrated into a single SoC, or the AP is a separate semiconductor chip, while other processing units are integrated into a single SoC. This application does not limit this to any particular type.

[0089] The memory 203 may include random access memory (RAM), read-only memory (ROM), or non-transitory computer-readable storage medium. The memory 203 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 203 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function, such as a HER-ACO-based path planning method; the data storage area may store data created based on the use of the electronic device 200, such as raster map data.

[0090] In addition, those skilled in the art will understand that the structure of the computer device 200 shown in the above figures does not constitute a limitation on the computer device 200. Electronic devices may include more or fewer components than shown, or combine certain components, or have different component arrangements. For example, the computer device 200 also includes components such as microphones, speakers, radio frequency circuits, sensors, audio circuits, power supplies, and Bluetooth modules, which will not be described in detail here.

[0091] This embodiment provides a path planning method based on HER-ACO, such as... Figure 1 As shown, the method includes the following steps:

[0092] Step S1: Obtain an environmental grid map, which includes the starting point, the target point, and obstacle distribution information.

[0093] An environmental grid map is a robot's internal representation of its surroundings, used to assist in path planning. Common types include: grid maps, visibility graphs, and topological maps. A grid map divides the environment into uniformly sized grids, with each grid recording the presence or absence of obstacles. A visibility graph constructs a map by connecting obstacle vertices, used for precise path planning. A topological map records only the abstract structure of key locations and pathways.

[0094] Grid-based method: Discretizes a continuous space into two-dimensional or three-dimensional grid points, each with a state (e.g., "accessible" or "obstacle"), facilitating path calculation and search. The advantages of this method are its simple structure, ease of implementation, and compatibility with various search algorithms, such as A* and Dijkstra's algorithm.

[0095] In mobile robot path planning algorithms, the construction of the environment map plays a crucial role. Therefore, before performing mobile robot path planning, environment modeling is necessary. Currently, commonly used environment map modeling methods include grid-based methods, visual map methods, and topological methods. Among them, the grid-based method is widely used by many researchers due to its ease of construction, representation, and storage. Therefore, this embodiment also utilizes the grid-based method to establish the environment model.

[0096] like Figure 3 As shown in (a), assuming the starting grid is denoted as (Starting point, the blue grid in the image), target grid is denoted as... (End point / target point, red grid in the diagram) If an obstacle expands outwards with the radius of the mobile robot, and the obstacle and its expansion boundary occupy a portion of a grid, then it is considered to occupy the entire grid. In the diagram, black grids represent obstacles, which the mobile robot cannot pass through, and white grids represent feasible paths, which the mobile robot can pass through. Construct an N×N grid map, where obstacle generation satisfies the following conditions:

[0097] (1)

[0098] In the formula, Let be the state value of the grid cell in row i and column j. A state value of 1 indicates that the grid cell is an obstacle (the obstacle is randomly generated and satisfies the connectivity constraint), and a state value of 0 indicates that the grid cell is free space, i.e., there are no obstacles.

[0099] In this context, obstacle random generation refers to randomly placing obstacle grids (impassable areas) on the map. The path planning objective is to find the shortest collision-free path from the starting point to the destination; therefore, the randomly generated grid map must satisfy connectivity constraints. Connectivity constraints mean that there must be an unobstructed path between the starting and ending points (i.e., the map must be connected); otherwise, path planning is meaningless.

[0100] Because random generation methods may result in obstacles surrounding the start or end point, such as Figure 3 As shown in section (b), this is the situation that makes the goal unreachable, and the connectivity constraint cannot be satisfied.

[0101] Therefore, refer to Figure 3 In part (c), the following method can be used to ensure that the randomly generated raster map satisfies the connectivity constraint: limit the number of attempts to generate the map, and after each map generation, use the breadth-first search (BFS) algorithm to traverse the map and check whether it is possible to reach the destination from the starting point (i.e., perform path reachability verification). If it is reachable, exit the loop; otherwise, continue to try to generate a new map.

[0102] Step S2: Introduce the HER algorithm to improve the traditional Ant Colony Algorithm (ACO). Use the improved ant colony algorithm to search for feasible paths from the starting point to the target point in the grid map, which serve as the initial planned path. Specific improvements are as follows:

[0103] Step S21: Based on the idea of ​​the HER algorithm, select intermediate nodes that meet the preset evaluation conditions as pseudo-targets. If the ant does not reach the target point, identify the effective sub-paths in the path from the starting point to the pseudo-target, and update the pheromone concentration according to the effective sub-paths.

[0104] Step S22: Calculate the direction weight factor based on the angle between the direction from the current node to the target point and the movement direction of the candidate next node.

[0105] Step S23: Calculate the neighborhood obstacle factor and the future obstacle factor; the neighborhood obstacle factor and the future obstacle factor reflect the obstacle distribution around the candidate node and the path in front of it, respectively.

[0106] Step S24: Update the global pheromone using a dynamically changing pheromone evaporation rate, wherein the pheromone evaporation rate decreases as the number of iterations increases;

[0107] Step S25: Based on the updated global pheromone, direction weight factor, neighborhood obstacle factor, and future obstacle factor, improve the path selection probability of the traditional ant colony algorithm to obtain a new path selection probability.

[0108] The following provides a detailed explanation of each step from S21 to S25.

[0109] In step S21, the HER algorithm is first introduced to improve the traditional ant colony algorithm.

[0110] HER (Hindsight Experience Replay) is a technique for solving the problem of sparse reward signals in reinforcement learning (RL). The idea is to improve sample efficiency and avoid complex reward engineering by treating attempts that fail to achieve the goal during reinforcement learning as attempts to achieve other goals.

[0111] In this embodiment, drawing on the idea of ​​HER technology, in the path planning of the ant colony algorithm, path segments that have not reached the target grid are regarded as experience of successfully reaching the intermediate target (pseudo-target), and the pheromone distribution is optimized based on this. Step S21 specifically includes the following sub-steps:

[0112] Step S21a: Select a pseudo-target.

[0113] Nodes (grids) that are farthest from the starting point in Euclidean distance and have shortest path lengths are selected as pseudo-targets. Among the paths that do not reach the target grids, unobstructed and effective sub-paths from the starting point to the pseudo-targets are selected. The pseudo-target evaluation formula is as follows:

[0114] (8)

[0115] In the formula, To start from the beginning To the false target Euclidean distance, This represents the actual length of the sub-path from which the ant reaches the false target.

[0116] Based on the evaluation results of the pseudo-targets ( (Value selection) filters out false targets. For example, an evaluation threshold can be preset, when... If the value exceeds the evaluation threshold, the node is classified as a pseudo-target, and the length of the effective sub-path corresponding to the pseudo-target is recorded. If the value is less than the evaluation threshold, it is not considered a spurious target.

[0117] Step S21b: Calculate the pheromone increment of the pseudo-target path based on the length of the effective sub-path. The formula is as follows:

[0118] (9)

[0119] In the formula, This is the pheromone increment for the pseudo-target path, used to enhance the pheromone concentration of the effective sub-path.

[0120] The purpose of step S22 is to determine the angle between the direction from the current node to the target point and the direction of movement of the candidate next node. Calculate the directional weighting factor The directional weighting factor is defined as the included angle. The cosine value is given by the following formula:

[0121] (10)

[0122] In the formula, The angle between the vector from the current node to the target node and the vector from the current node to the next node.

[0123] To prioritize paths that approach the target direction and reduce ineffective exploration, the direction weighting factor is normalized to generate a new direction weighting factor, namely the normalized direction weighting factor. The formula is as follows:

[0124] (11)

[0125] In the formula, This is the new directional weighting factor.

[0126] Step S23 is used to set obstacle heuristic factors to achieve efficient obstacle avoidance and reduce redundant paths, including neighborhood obstacle factors and future obstacle factors.

[0127] Step S23a: Introduce a new heuristic factor, called the neighborhood obstacle factor.

[0128] Considering the heuristic information between grids and barrier grids, a new heuristic factor is introduced. Used to reflect the current grid Candidate next node Consider the density of obstacles in the surrounding local area to avoid choosing paths that are close to obstacles.

[0129] When an ant chooses the next node, it checks the number of obstacles in that node's 8-neighborhood. The more obstacle grids in the neighborhood, the lower the node's transition probability, thus preventing the ant from approaching the obstacles. The new heuristic factor is represented as:

[0130] (14)

[0131] In the formula, It represents the number of obstacle grids within the current grid's 8-neighborhood.

[0132] Step S23b: Introduce the future obstacle factor.

[0133] Obstacle statistics are performed on the path within a preset number of steps to the next node (e.g., 3 steps ahead) to evaluate the obstacle distribution within the preset number of steps ahead of the candidate path, thereby reducing the tendency to choose paths with obstacles ahead. The future obstacle factor is expressed as:

[0134] (15)

[0135] In the formula, This represents the predicted penalty coefficient. This indicates the number of obstacles in the predicted path, that is, the number of obstacles on the path within the preset number of steps.

[0136] The purpose of step S24 is to dynamically update the pheromone concentration. Specifically:

[0137] In traditional ant colony algorithms, the global pheromone evaporation rate The pheromone evaporation rate is fixed, which can lead to difficulties in balancing convergence speed and optimization ability. To address this issue, this embodiment proposes a method for dynamically adjusting the evaporation rate. Specifically, the pheromone evaporation factor decays exponentially, as shown in the following formula:

[0138] (12)

[0139] in, It is the pheromone evaporation rate (evaporation rate) of the current iteration. It is the maximum evaporation rate. It is the volatility decay coefficient. It is the current iteration number. It is the maximum number of iterations.

[0140] When the pheromone evaporation factor decays exponentially, the global pheromone update formula is as follows:

[0141] (13)

[0142] In the formula, It is a global pheromone that decays exponentially.

[0143] By dynamically adjusting the global evaporation rate, the improved ant colony algorithm can explore new paths faster in the early stages, avoiding getting trapped in local optima; and converge to the optimal path faster in the later stages.

[0144] Based on steps S21 to S24, in step S25, the path selection probability in the traditional ant colony algorithm is improved according to the updated global pheromone, orientation weight factor, neighborhood obstacle factor, and future obstacle factor. The new path selection probability is expressed as:

[0145] (16)

[0146] In the formula, Probability of choosing a new path. Ants In the current grid The set of selectable graticles from 8 neighboring graticles. This represents the pheromone enhancement coefficient. The larger the value, the deeper the impact on the initial pheromone, which in turn leads to a poorer global search capability of the algorithm. The expected heuristic information coefficient is the coefficient of interest; the larger the value, the stronger the ant's heuristic information. The closer the grid is to the target, the higher its tendency; express Current grid To the neighboring grid The concentration of pheromones along the connection path; express Current grid To the neighboring grid Heuristics on the connection path; Indicates the current grid To the neighboring grid Neighborhood obstacle factor; Indicates the directional weighting factor for normalization; This represents the future obstacle factor.

[0147] S3. Based on the feasibility check of the straight path, smooth the initial planned path, remove redundant intermediate nodes, and output the final planned path.

[0148] In this embodiment, smoothing is achieved by constructing a curve smoothing function.

[0149] Based on the feasibility check of straight paths, the Bresenham algorithm is used to generate straight grid points (i.e., straight paths) between two points, and it is verified that all these points are located in free space (without obstacles). Specifically, this includes the following sub-steps:

[0150] S31. Determine the unobstructed conditions for a straight line.

[0151] Given any two points and Its straight path Generated by Bresenham. The accessibility conditions for this path are:

[0152] (17)

[0153] Among them, the straight path Represented as from point arrive The set of all grid points traversed by the straight path trajectory. Represents any grid in the map state, This indicates that the grid is unobstructed, i.e., a free grid.

[0154] S32. Set path smoothing rules.

[0155] Path smoothing rules are used to optimize the generated path (i.e. the initial planned path) after path planning (such as ant colony algorithm generation), removing redundant intermediate nodes to make the path simpler and smoother.

[0156] For the initial planned path ,like If there are no obstacles, then remove them. .

[0157] ,

[0158] (18)

[0159] Formulas (17) and (18) together constitute a curve smoothing function, which is used to achieve path smoothing.

[0160] In summary, this embodiment improves upon the traditional Ant Colony Algorithm (ACO) by introducing the HER algorithm. It introduces a neighborhood obstacle factor to characterize the obstacle density in the current node's neighborhood and a future obstacle factor to reduce the tendency to choose paths with obstacles ahead, thus forming a three-dimensional obstacle perception. A dynamic evaporation mechanism based on an exponential decay function is used to adjust the pheromone evaporation intensity, and a smoothness function is designed to optimize path smoothness. These improvements alleviate the shortcomings of traditional ant colony algorithms in robot path planning, such as slow convergence speed, high mechanical losses, and poor adaptability to dynamic obstacles in complex environments, thereby improving the accuracy, adaptability, and efficiency of robot path planning.

[0161] The method proposed in this embodiment will be verified below with specific exemplary cases.

[0162] To verify the effectiveness of the HER-ACO-based path planning method provided in this embodiment, environmental maps of varying complexity were constructed. The method was then compared with traditional ant colony optimization (ACO) and breadth-first search (BFS) algorithms in the same grid environment. In all examples, the simulation algorithm was run on Matlab R2021a; all simulations were performed on this platform.

[0163] Environment 1:

[0164] The concrete robot was simulated and compared in a 20m×20m grid environment with an obstacle zone probability of 10%. The simulation results are as follows: Figure 5 , Figure 6 As shown. Figure 5 In the path planning results of Environment 1 shown, the red path is the path planned by the BFS algorithm, the green path is the path planned by the traditional ant colony algorithm, and the blue path is the path planned by the improved ant colony algorithm provided in this embodiment. Figure 5 It can be seen that the traditional ant colony algorithm has many twists and turns and redundant inflection points. The improved ant colony algorithm reduces the number of inflection points by 23.5% compared to the traditional ant colony algorithm, and the path is smoother, but it is still inferior to the path planned by the BFS algorithm. Figure 6As shown in the convergence curves, the improved ant colony algorithm exhibits a rapid decrease in path length during the initial iterations, stabilizing within a short period and demonstrating a faster convergence speed. In contrast, the traditional ant colony algorithm shows greater fluctuations in path length, slower convergence speed, and even significant optimization potential in later stages. The faster convergence speed of the improved ant colony algorithm is primarily attributed to the dynamic pheromone evaporation mechanism (exponential decay evaporation rate) and the HER technique. The combination of the HER technique and the dynamic evaporation mechanism improves exploration efficiency in the early stages of path planning, preventing getting trapped in local optima; while in the later stages, it enhances development capabilities, accelerating convergence to a better path. Regarding path length, although the BFS algorithm has fewer inflection points and faster planning speed, its planned paths are longer. The improved ant colony algorithm, on the other hand, converges quickly, resulting in the shortest final path with better smoothness, better balancing path optimization efficiency and accuracy.

[0165] Environment 2:

[0166] To ensure a fairer simulation comparison, this example modifies the grid map in Environment 1 and compares the path planning performance of each algorithm again. Specifically, the concrete robot is simulated in a 20m×20m grid environment with a 20% probability of obstacles. The comparison results are as follows: Figure 7 , Figure 8 As shown.

[0167] from Figure 7 As can be seen, despite the significant increase in obstacle density in Environment 2, the improved ant colony algorithm maintains high smoothness and simplicity in its path. In contrast, the traditional ant colony algorithm's path (green dashed line) becomes more tortuous and redundant, especially exhibiting significant detours and circuitous behavior in dense obstacle areas. This indicates that the path smoothing mechanism in the improved algorithm (such as removing redundant nodes based on Bresenham line detection) remains effective in high obstacle density environments, significantly reducing inflection points and improving path quality. The improved ant colony algorithm successfully bypasses all obstacles, and the path is closer to the optimal solution, further validating the design advantages of the improved algorithm in terms of neighborhood obstacle perception (neighborhood obstacle factor) and future obstacle prediction (future obstacle factor). Even with increased obstacle probability, the improved algorithm effectively avoids the path approaching obstacles or entering dead ends, demonstrating stronger environmental adaptability.

[0168] Environment 3:

[0169] To further increase the complexity of obstacles and the environment, the concrete robot was simulated in a 20m×20m grid environment with an obstacle zone probability of 40%. The comparison results are shown in Figure 9. Figure 10 As shown.

[0170] Environment 4:

[0171] To more comprehensively compare the path planning performance of each algorithm, the grid map was expanded to a more complex 30m×30m environment with a 20% probability of obstacle areas. The comparison results are as follows: Figure 11 , Figure 12 As shown.

[0172] Analysis of simulation results under different environments (environments 1-4) shows that, in terms of path length, the improved ant colony algorithm outperforms both the traditional ant colony algorithm and the BFS algorithm in 20×20 and 30×30 grid environments. Regarding convergence speed, the improved ant colony algorithm significantly outperforms the traditional ant colony algorithm but is inferior to the BFS algorithm. These results demonstrate that the HER-ACO-based path planning method provided in this embodiment, by introducing the HER concept, dynamic pheromone evaporation mechanism, multi-dimensional heuristic factors, and path smoothing processing, can significantly improve the planning accuracy and convergence efficiency of the algorithm in robot path planning. It successfully alleviates the problems of path redundancy, slow convergence speed, and susceptibility to local optima inherent in the traditional ant colony algorithm. Simulation results fully verify the significant advantages of the improved ant colony algorithm in high obstacle density environments, particularly its excellent performance in path quality, convergence speed, and stability, making it suitable for practical applications such as robot navigation and concrete robot path planning.

[0173] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A path planning method based on HER-ACO, characterized in that, include: S1. Obtain an environmental grid map, the grid map including the starting point, the target point, and obstacle distribution information; S2. The HER algorithm is introduced to improve the traditional ant colony algorithm (ACO). The improved ant colony algorithm is used to search for feasible paths from the starting point to the target point in the grid map as the initial planned path. The specific improvements are as follows: Based on the idea of ​​the HER algorithm, intermediate nodes that meet the preset evaluation conditions are selected as pseudo-targets. If the ant does not reach the target point, the effective sub-paths in the path from the starting point to the pseudo-target are identified, and the pheromone concentration is updated according to the effective sub-paths. Calculate the direction weight factor based on the angle between the direction from the current node to the target point and the direction of movement of the candidate next node; Calculate the neighborhood obstacle factor and the future obstacle factor; the neighborhood obstacle factor and the future obstacle factor reflect the obstacle distribution around the candidate node and the path in front of it, respectively; The global pheromone is updated using a dynamically changing pheromone evaporation rate, which decreases as the number of iterations increases. Based on the updated global pheromone, orientation weight factor, neighborhood obstacle factor, and future obstacle factor, the path selection probability of the traditional ant colony algorithm is improved to obtain a new path selection probability. S3. Based on the feasibility check of the straight path, smooth the initial planned path, remove redundant intermediate nodes, and output the final planned path. The selection of intermediate nodes that meet the preset evaluation conditions as pseudo-targets specifically involves: The formula for calculating the pseudo-target evaluation value is as follows: ， In the formula, This is a pseudo-target evaluation value. To start from the beginning To the false target Euclidean distance, This represents the actual length of the sub-path the ant takes to reach the false target; like If the value exceeds the preset evaluation threshold, the node is treated as a pseudo-target, and the length of the effective sub-path corresponding to the pseudo-target is recorded.

2. The method according to claim 1, characterized in that, The new path selection probability is calculated as follows: , In the formula, For ants From the current node Move to the candidate next node The probability of path selection. express It belongs to ants In the current grid The set of selectable grids in 8 neighboring grids Any grid in, This represents the pheromone enhancement coefficient. The larger the value, the deeper the impact on the initial pheromone, which in turn leads to a poorer global search capability of the algorithm. The expected heuristic information coefficient is the coefficient of interest; the larger the value, the stronger the ant's heuristic information. The closer the grid is to the target, the higher its tendency; express Current grid To the neighboring grid The concentration of pheromones along the connection path; express Current grid To the neighboring grid Heuristics on the connection path; Indicates the current grid To the neighboring grid Neighborhood obstacle factor; Indicates the directional weighting factor for normalization; Indicates the future obstacle factor; express Current grid To grid The concentration of pheromones along the connection path; express Current grid To grid Heuristics on the connection path; Indicates the current grid To grid The neighborhood obstacle factor.

3. The method according to claim 1, characterized in that, The process of updating the pheromone concentration based on the effective sub-path specifically involves: Calculate the pheromone increment based on the length of the effective sub-path. The formula is as follows: ， In the formula, This is a constant gain coefficient, whose value is a constant greater than 0; This represents the actual length of the sub-path from which the ant reaches the false target.

4. The method according to claim 1, characterized in that, The pheromone evaporation rate decreases exponentially with increasing iteration number, and its expression is as follows: ， In the formula, The pheromone evaporation rate in the current iteration; It is the maximum evaporation rate. It is the volatility decay coefficient. It is the current iteration number. It represents the maximum number of iterations.

5. The method according to claim 1, characterized in that, The direction weight factor is defined as the cosine of the angle between the direction from the current node to the target point and the direction of movement of the next candidate node, and is normalized to obtain the normalized direction weight factor.

6. The method according to claim 1, characterized in that, The neighborhood obstacle factor is used to reflect the obstacle density in the local area surrounding the candidate next node, and is calculated as follows: ， In the formula, For the current node Candidate next node Neighborhood obstacle factor; The current grid The number of obstacle grids in the 8-neighborhood.

7. The method according to claim 1, characterized in that, The future obstacle factor is used to evaluate the distribution of obstacles within a preset number of steps ahead of the candidate path, and is calculated as follows: ， In the formula, Indicates the future obstacle factor. This represents the predicted penalty coefficient. The number of obstacles within a preset number of steps ahead of the candidate path.

8. The method according to claim 1, characterized in that, The process of smoothing the initial planned path based on the feasibility check of the straight path, removing redundant intermediate nodes, and outputting the final planned path includes: For the initial planned path any intermediate node in ( If from arrive If all grid cells along the straight path are free grid cells, then delete the node. The straight path is generated using the Bresenham algorithm.

9. A HER-ACO-based path planning system, the system being used to execute the HER-ACO-based path planning method as described in any one of claims 1 to 8, comprising: A construction unit is used to construct an environmental grid map, the grid map including a starting point, a target point, and obstacle distribution information; The improved unit introduces the HER algorithm to improve the traditional ant colony algorithm (ACO), and uses the improved ant colony algorithm to search for feasible paths from the starting point to the target point in the grid map as the initial planned path. The specific improvements are as follows: Based on the idea of ​​the HER algorithm, intermediate nodes that meet the preset evaluation conditions are selected as pseudo-targets. If the ant does not reach the target point, the effective sub-paths in the path from the starting point to the pseudo-target are identified, and the pheromone concentration is updated according to the effective sub-paths. Calculate the direction weight factor based on the angle between the direction from the current node to the target point and the direction of movement of the candidate next node; Calculate the neighborhood obstacle factor and the future obstacle factor; the neighborhood obstacle factor and the future obstacle factor reflect the obstacle distribution around the candidate node and the path in front of it, respectively; The global pheromone is updated using a dynamically changing pheromone evaporation rate, which decreases as the number of iterations increases. Based on the updated global pheromone, orientation weight factor, neighborhood obstacle factor, and future obstacle factor, the path selection probability of the traditional ant colony algorithm is improved to obtain a new path selection probability. The smoothing unit is used to smooth the initial planned path based on the feasibility check of the straight path, remove redundant intermediate nodes, and output the final planned path. The selection of intermediate nodes that meet the preset evaluation conditions as pseudo-targets specifically involves: The formula for calculating the pseudo-target evaluation value is as follows: ， In the formula, This is a pseudo-target evaluation value. To start from the beginning To the false target Euclidean distance, This represents the actual length of the sub-path the ant takes to reach the false target; like If the value exceeds the preset evaluation threshold, the node is treated as a pseudo-target, and the length of the effective sub-path corresponding to the pseudo-target is recorded.

Images