Mobile robot path planning method based on improved ant colony algorithm

Abstract

The invention provides a mobile robot path planning method based on an improved ant colony algorithm, and relates to the technical field of robot path planning, and the method is technically characterized in that the method comprises the following steps: carrying out environment modeling by adopting a grid method, and constructing a multi-target performance evaluation function; an improved ant colony algorithm is adopted to carry out optimal path search, path planning is carried out on a grid map from a starting node to a target node, and the improvement of the ant colony algorithm comprises the steps of distributing pheromone concentration by adopting a non-uniform initialization method; improving a state transition rule; differentiating a pheromone updating strategy; and smoothing the path. An initial pheromone distribution optimization strategy is introduced, heuristic information based on a triangular inequality, a steering angle suppression function and a state transition rule of artificial potential field force are combined, a differential pheromone updating strategy based on a multi-target performance evaluation index is adopted, and finally a planned path is smoothed. Therefore, the convergence speed, the path quality and the environment adaptability of the algorithm are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of robot path planning technology, and in particular to a mobile robot path planning method based on an improved ant colony algorithm. Background Technology

[0002] With the rapid development of mobile robot technology, path planning, as a core component of autonomous navigation, directly impacts the robot's operational efficiency through its optimization efficiency and path quality. Ant colony optimization, a biomimetic optimization algorithm that simulates the foraging behavior of ants in nature, is widely used in robot path planning problems due to its advantages such as positive feedback, strong robustness, and good parallelism.

[0003] However, existing ant colony-based mobile robot path planning methods still suffer from the following technical bottlenecks:

[0004] Specifically, the problems are as follows: First, traditional ant colony optimization (ACO) algorithms use uniformly distributed pheromones in the initial stage, leading to a lack of effective guidance for the ants in the early stages of the search, resulting in significant blindness and prolonged convergence time. Second, traditional state transition rules and pheromone update methods easily cause the algorithm to converge prematurely to local optima, lacking targeted guidance for the search direction, resulting in significant limitations in both efficiency and quality. Furthermore, the paths generated by traditional methods contain too many unnecessary turns, leading to disjointed robot movement, increased mechanical wear, and energy consumption. Finally, in obstacle-rich environments, traditional ACO algorithms are prone to deadlock, lacking an effective escape mechanism. Summary of the Invention

[0005] The purpose of this invention is to solve the technical bottleneck problem of existing mobile robot path planning methods based on ant colony algorithms.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A mobile robot path planning method based on an improved ant colony algorithm includes the following steps:

[0008] S1: The working environment of the robot is modeled using the grid method, the positions of the start and end points are defined, and a multi-objective performance evaluation function that integrates path length, smoothness and safety is constructed.

[0009] S2: An improved ant colony algorithm is used for optimal path search. Path planning is performed on a grid map from the starting node to the target node. The improvements to the ant colony algorithm include:

[0010] The pheromone concentration is allocated using a non-uniform initialization method based on the global cost of the path.

[0011] Improvements to the state transition rule: fusion of pheromone concentration, heuristic information based on the triangle inequality, turning angle suppression function, and artificial potential force;

[0012] Differentiated pheromone update strategies include local updates, global updates, and penalty mechanisms for the worst path;

[0013] S3: Path smoothing: Generates a continuous smooth trajectory by removing redundant nodes and parametric curve fitting.

[0014] Preferably, the modeling steps in S1 are as follows:

[0015] Divide the given two-dimensional plane space uniformly into An array of grid cells, where each grid cell is assigned a state value to characterize its physical properties:

[0016] “0”: This indicates that the grid is a free area, and the robot can pass through safely;

[0017] “1”: This indicates that the grid is an obstacle area that the robot must avoid;

[0018] Through the above mapping, a two-dimensional binary raster map is generated, in which the coordinates of the starting point and the target point of the path planning are clearly defined.

[0019] Preferably, the method for constructing the multi-objective performance evaluation function in S1 is as follows: First, determine three sub-indicators: path length, smoothness, and safety. After determining the three sub-indicators, use a weighted summation method to construct a comprehensive multi-objective performance evaluation function. This serves as the basis for the algorithm's decision to search for the optimal path:

[0020] (1)

[0021] In the above formula, , and These are the weighting coefficients for path length, smoothness, and safety, respectively. , , These represent the path length index, smoothness index, and safety index, respectively.

[0022] Preferably, the three sub-indicators are calculated as follows:

[0023] Path length metric calculation method:

[0024] (2)

[0025] (3)

[0026] in, This represents the total number of grid cells along the path. For the first on the path Each grid cell has the following coordinates: ; Indicates adjacent grid cells and The Euclidean distance between them;

[0027] The formula for calculating the smoothness index is as follows:

[0028] (4)

[0029] in, Representing vectors and The included angle; This represents the total number of all turning points in the path. and These are the weighting factors for the steering angle and the number of turns, respectively, and satisfy the following conditions: ;

[0030] The formula for calculating the safety index is as follows:

[0031] (5)

[0032] (6)

[0033] In the above formula, Indicates the first The Euclidean distance from each grid cell to its nearest obstacle, and This represents the coordinates of the nearest obstacle.

[0034] Preferably, the calculation formula for the non-uniform initialization method in S2 is as follows:

[0035] (7)

[0036] In the above formula, Based on the pheromone intensity constant, Let S represent the Euclidean distance between points A and B, and let S and E be the start and end points of the path, respectively.

[0037] Preferably, the state transition rule in S2 incorporates the following information:

[0038] Heuristic information based on the triangle inequality: , Indicates the current node With candidate nodes The Euclidean distance between them Indicates candidate nodes The Euclidean distance between the endpoint E and the destination E. Indicates the current node The Euclidean distance between the endpoint E and the destination E; It is a very small positive number, used to prevent the denominator from being zero;

[0039] Steering angle suppression function: ,in, This refers to the steering angle formed when the robot shifts from its current direction of motion to a candidate direction;

[0040] Artificial potential force: including gravitational factor, repulsive factor, and combined artificial potential force.

[0041] Among them, the gravitational factor: , This is the gravitational gain coefficient. This represents the Euclidean distance from the current node to the target point.

[0042] Repulsive factor: , The repulsive force gain coefficient, The distance from the current node to the nearest obstacle. The radius of influence of the repulsive force field is denoted as .

[0043] Combined artificial potential field force: , This is the adjustment constant for the potential field sensitivity; This represents the effective component of gravity in the candidate direction; This represents the effective component of the repulsive force in the candidate direction.

[0044] Preferably, the state transition rule in S2 is expressed as follows:

[0045] (8)

[0046] yes Uniformly distributed random numbers in an interval, when Time: Ants rely on probability The probability of making a choice is calculated using the following formula:

[0047] (9)

[0048] in, Ants At the current node The set of feasible neighborhood nodes consists of the grid cells within its eight neighborhoods that are not occupied by obstacles and have not been visited by the ant in this round; These are the weighting coefficients for pheromones, heuristic information, turning angle suppression function, and artificial potential force, respectively.

[0049] Preferably, the differential pheromone update strategy in S2 specifically includes:

[0050] Local pheromone update: , The local pheromone evaporation coefficient; Set as the aforementioned non-uniform initial pheromone ;

[0051] Global pheromone update: First, calculate the multi-objective performance evaluation index value of the path constructed by each ant based on the multi-objective performance evaluation function formula. And determine the optimal path for this iteration. and worst path Then, perform the following operations:

[0052] Global Enhancement: For the Contemporary Optimal Path Pheromones are enhanced on the upper edge:

[0053] (10)

[0054] in, The global pheromone evaporation coefficient. , Where is the pheromone intensity constant. It is the multi-objective performance evaluation index value of the contemporary optimal path;

[0055] Penalty mechanism: for the worst contemporary path Further weaken the pheromones on the surface:

[0056] (11)

[0057] in, The penalty coefficient is... , It is the multi-objective performance evaluation index value of the worst path in the contemporary era.

[0058] Preferably, in step S3, redundant nodes are first optimized to simplify the path structure, and then a parametric curve fitting method is used to smooth the key path points.

[0059] This application also provides a mobile robot path planning device based on an improved ant colony algorithm, comprising:

[0060] Environment Modeling Module: Used to model the robot's working environment using a grid method, defining the start and end points;

[0061] Multi-objective evaluation module: used to construct a multi-objective performance evaluation function that integrates path length, smoothness, and safety.

[0062] An improved ant colony algorithm module is used for optimal path search on a grid map. This module includes an initial pheromone optimization unit, a multi-factor state transition unit, and a differential update unit. The initial pheromone optimization unit employs a non-uniform initialization method based on global path cost. The multi-factor state transition unit integrates pheromone concentration, heuristic information, turning angle suppression, and artificial potential force. The differential update unit implements local updates, global updates, and a worst-case path penalty mechanism.

[0063] Path smoothing module: used to generate continuous smooth trajectories through redundant node removal and curve fitting;

[0064] Path output module: Used to output the final planned executable path.

[0065] Compared with the prior art, this application has the following beneficial effects:

[0066] The mobile robot path planning method based on the improved ant colony algorithm provided in this application introduces an initial pheromone allocation optimization strategy, combines heuristic information based on the triangle inequality, the turning angle suppression function and the state transition rule of the artificial potential field force, and adopts a differentiated pheromone update strategy based on multi-objective performance evaluation index. Finally, the planned path is smoothed, thereby significantly improving the convergence speed, path quality and environmental adaptability of the algorithm. Attached Figure Description

[0067] Figure 1 This is a flowchart of a mobile robot path planning method based on an improved ant colony algorithm according to one embodiment of the present invention.

[0068] Figure 2 An environmental grid map is provided in one embodiment of the present invention.

[0069] Figure 3 This is an eight-neighborhood movement model in one embodiment of the present invention;

[0070] Figure 4 This is a schematic diagram of the path turning angle in one embodiment of the present invention;

[0071] Figure 5 This is a schematic diagram of the principle of the triangle inequality in one embodiment of the present invention;

[0072] Figure 6 This is a schematic diagram of the gravity factor in one embodiment of the present invention;

[0073] Figure 7 This is a schematic diagram of the repulsion factor in one embodiment of the present invention;

[0074] Figure 8This is a flowchart of the improved ant colony algorithm in one embodiment of this application. Detailed Implementation

[0075] The present invention will be further described in detail below with reference to specific embodiments.

[0076] Please see Figure 1 A mobile robot path planning method based on an improved ant colony algorithm includes the following steps:

[0077] S1: Environmental Modeling and Construction of Multi-Objective Performance Evaluation Metrics

[0078] Specifically, in one embodiment, the steps are as follows:

[0079] S11: Environment Modeling: The working environment of the robot is modeled using the grid method.

[0080] Specifically, in one embodiment, the given two-dimensional planar space is uniformly divided into An array of grid cells, where each grid cell is assigned a state value to characterize its physical properties:

[0081] "0": This indicates that the grid is a free area, and the robot can pass through safely.

[0082] “1”: This indicates that the grid is an obstacle area that the robot must avoid.

[0083] The above mapping generates a two-dimensional binary raster map, in which the coordinates of the starting point and the target point of the path planning are clearly defined. Figure 2 The map is a 20x20 grid map, where black grids represent obstacles, S is the starting point, and E is the ending point.

[0084] like Figure 3 As shown, this application adopts an eight-neighborhood movement model, which allows the robot to move from its current grid to its eight neighboring grids in one step (i.e., up, down, left, right, upper left, upper right, lower left, and lower right directions).

[0085] S12: Construction of Multi-Objective Performance Evaluation Metrics:

[0086] This step aims to construct a comprehensive multi-objective evaluation function to overcome the limitations of a single path length index and ensure that the planned path achieves comprehensive optimization in multiple performance indicators such as length, smoothness, and safety.

[0087] Specifically, in one implementation, three core performance indicators are first defined: path length, smoothness, and safety.

[0088] 1. Path length metric:

[0089] Path length is a fundamental metric for measuring path planning efficiency, directly impacting a robot's running time and energy consumption. This metric is defined as the total path length from the starting point to the ending point, calculated by summing the Euclidean distances between adjacent nodes along the path.

[0090] The formula for calculating the path length is as follows:

[0091] (1)

[0092] (2)

[0093] in, This represents the total number of grid cells along the path. For the first on the path Each grid cell has the following coordinates: ; Indicates adjacent grid cells and The Euclidean distance between them.

[0094] 2. Smoothness index

[0095] The smoothness metric is used to optimize the smoothness of the path, reduce unnecessary turns, thereby improving the stability of robot motion and reducing mechanical wear. This metric is determined by the sum of the turning angles in the path and the number of turns.

[0096] The smoothness calculation formula is as follows:

[0097] (3)

[0098] in, Indicates at node The turning angle at that point, i.e., the vector and The included angle; This represents the total number of all turning points in the path. and These are the weighting factors for the steering angle and the number of turns, respectively, and satisfy the following conditions: .

[0099] Steering angle The specific geometric definition is as follows Figure 4 As shown. Suppose there are three consecutive path points. , , Then the node Turning angle The specific calculation method is as follows:

[0100] (4)

[0101] 3. Safety indicators

[0102] The safety metric aims to improve the obstacle avoidance capability of a path, ensuring that the robot maintains a safe distance from obstacles. This metric is defined as the sum of the Euclidean distances from each node on the path to its nearest obstacle.

[0103] The formula for calculating the safety index is as follows:

[0104] (5)

[0105] (6)

[0106] In the above formula, Indicates the first The Euclidean distance from each grid cell to its nearest obstacle, and This represents the coordinates of the nearest obstacle.

[0107] After defining the three sub-indicators mentioned above, a comprehensive multi-objective performance evaluation function is constructed using the weighted summation method. This serves as the basis for the algorithm's decision to search for the optimal path. The function's expression is as follows:

[0108] (7)

[0109] in, , and These are the weighting coefficients for path length, smoothness, and safety, respectively. Due to the safety index... A larger value indicates a safer path, while the evaluation function... A smaller value indicates better overall path performance, therefore, it is taken as [value] in the function. reciprocal Calculations are performed. The specific values ​​of each weighting coefficient can be set according to the robot's actual working environment and the emphasis placed on different performance characteristics.

[0110] S2: Use an improved ant colony algorithm to search for the optimal path.

[0111] In one embodiment, to overcome the shortcomings of traditional ant colony algorithms, this application implements the following three key improvements:

[0112] S21: Optimization of initial pheromone allocation strategy:

[0113] To overcome the blindness of the initial search, this application abandons the traditional uniform initialization method and adopts a non-uniform initialization method based on the global cost of the path.

[0114] This method involves each feasible edge on the map. Assign initial pheromone concentration The calculation formula for the non-uniform initialization method based on path global cost is as follows:

[0115] (8)

[0116] In the above formula, Based on the pheromone intensity constant, Let S represent the Euclidean distance between points A and B, and let S and E be the start and end points of the path, respectively.

[0117] S22: State transition rules incorporating multiple factors

[0118] The state transition rules were improved, and a multi-factor collaborative decision-making mechanism was constructed that integrates pheromone concentration, heuristic information, turning angle inhibition function, and artificial potential field guidance.

[0119] 1. Heuristic information based on the triangle inequality

[0120] This factor is used to enhance the goal orientation of ant search. Its core idea is to use the geometric principle of triangle inequality to provide macro-guidance for path selection toward the destination.

[0121] Specifically, the consistency between the candidate direction and the target direction is evaluated by calculating the difference between the "path length through candidate nodes" and the "straight-line distance from the current node to the destination". The smaller the difference, the closer the candidate node is to the ideal path, and the greater its attractiveness.

[0122] like Figure 5 As shown, the specific calculation formula for the heuristic information is as follows:

[0123] (9)

[0124] in, Indicates the current node With candidate nodes The Euclidean distance between them Indicates candidate nodes The Euclidean distance between the endpoint E and the destination E. Indicates the current node The Euclidean distance between the endpoint E and the destination E; It is a very small positive number, used to prevent the denominator from being zero.

[0125] 2. Steering angle suppression function

[0126] This factor aims to improve path smoothness by penalizing large steering angles. Its value varies with the steering angle. As it increases, it monotonically decreases; specifically, in exponential function form:

[0127] (10)

[0128] in, This is the turning angle formed when the robot changes from its current direction of motion to a candidate direction. The smaller the value, the smoother the path. The closer the value is to 1, the higher the tendency for that path to be chosen.

[0129] 3. Artificial potential force

[0130] The core innovation of artificial potential field force lies in the use of projection addition method, which integrates the projected components of the attraction of the target point and the repulsion of the obstacle in the candidate movement direction to provide ants with real-time obstacle avoidance and target approach guidance.

[0131] (1) Gravitational factor

[0132] Gravity attracts the ant towards the target point. The target point is relative to the current position. The magnitude of gravity Defined as:

[0133] (11)

[0134] in, This is the gravitational gain coefficient. This represents the Euclidean distance from the current node to the target point. Gravity in the candidate directions... The effectiveness depends on how close the candidate direction is to the target direction.

[0135] like Figure 6 As shown, angle For vectors (From the current node to the target point) and vector The angle between (the current node and the candidate node). When the candidate direction is directly facing the target. When the direction deviates, the gravitational force is at its maximum; when the direction deviates... At that time, the gravitational effect weakens.

[0136] (2) Repulsive factor

[0137] Repulsive force drives the ant away from the obstacle. The obstacle's position relative to the ant's current location. The magnitude of the repulsive force Defined as:

[0138] (12)

[0139] in, The repulsive force gain coefficient, The distance from the current node to the nearest obstacle. This represents the maximum radius of influence of the repulsive field. The repulsive force is located in the candidate direction. The inhibition effect is closely related to whether the candidate direction is directly facing the obstacle.

[0140] like Figure 7 As shown, angle For vectors (From the obstacle to the current node) and vector The angle between them. When the ant is moving in a direction directly towards the obstacle. When the direction is opposite to the obstacle, the repulsive force has the strongest inhibitory effect, and the path is effectively suppressed; When the repulsive force decreases, its influence diminishes.

[0141] (3) Comprehensive artificial potential field force

[0142] The combined potential force is expressed by fusing the projected components of the above-mentioned gravitational and repulsive forces through an exponential function, as follows:

[0143] (13)

[0144] in, This is the adjustment constant for the potential field sensitivity; This represents the effective component of gravity in the candidate direction; This represents the effective component of the repulsive force in the candidate direction. The candidate direction is considered effective when it simultaneously satisfies both the conditions of "pointing towards the target" and "moving away from the obstacle." A larger value significantly increases the probability that the path will be selected.

[0145] After defining the above core factors, ants at nodes Select the next node The state transition rule is as follows. This rule uses a preset decision threshold. To balance “utilizing” known good paths with “exploring” new paths.

[0146] (14)

[0147] when At time: Ants directly select the neighboring node with the largest comprehensive heuristic function value, where yes Uniformly distributed random numbers within an interval. When Time: Ants rely on probability The probability of making a choice is calculated using the following formula:

[0148] (15)

[0149] in, Ants At the current node The set of feasible neighborhood nodes consists of the grid cells within its eight neighborhoods that are not occupied by obstacles and have not been visited by the ant in this round; These are the weighting coefficients for pheromones, heuristic information, turning angle inhibition function, and artificial potential force, respectively, used to adjust the relative importance of different factors in decision-making.

[0150] S23: Differentiated pheromone update strategy based on multi-objective performance evaluation indicators

[0151] 1. Local pheromone update

[0152] This process is executed immediately after each ant moves a step, aiming to increase the diversity of path exploration within the current iteration. When an ant moves from node... Move to node After that, it will affect the edge. The pheromones on the surface undergo localized evaporation and renewal, following the rules as follows:

[0153] (16)

[0154] in, The local pheromone evaporation coefficient; Set as the aforementioned non-uniform initial pheromone This operation prevents pheromones from accumulating excessively on a single path, prompting other ants of the same generation to explore different paths.

[0155] 2. Global pheromone update

[0156] This process is executed after all ants have completed path construction for this iteration, aiming to strengthen excellent solutions and weaken inferior solutions based on path quality. Before updating, the multi-objective performance evaluation index value of the path constructed by each ant is first calculated according to formula (7). And determine the optimal path for this iteration. and worst path Then, perform the following operations:

[0157] Global Enhancement: For the Contemporary Optimal Path Pheromones are enhanced on the upper edge:

[0158] (17)

[0159] in, The global pheromone evaporation coefficient. , Where is the pheromone intensity constant. It is the multi-objective performance evaluation index value of the contemporary optimal path.

[0160] Penalty mechanism: for the worst contemporary path Further weaken the pheromones on the surface:

[0161] (18)

[0162] in, The penalty coefficient is... , It is the multi-objective performance evaluation index value of the worst path in the contemporary era.

[0163] S24: Improve the overall process of the ant colony algorithm

[0164] like Figure 8 As shown, the overall process of the improved ant colony algorithm is as follows:

[0165] (1) Initialization

[0166] Environment and parameter settings: Load the raster map and define the start point S and end point E. Set the number of ants. Maximum number of iterations Pheromones and heuristic factor weights Parameters such as these.

[0167] Improved initial pheromone allocation: A non-uniform initialization method based on path global cost (Equation 8) is adopted for the edges on the map. Assign initial pheromone concentration .

[0168] (2) Iterative loop

[0169] Set an iteration counter and repeatedly execute the following steps until the maximum number of iterations or the convergence condition is met.

[0170] (2.1) Path construction

[0171] Each ant starts from the starting point and gradually constructs a complete path according to the multi-factor state transition rules (Equations 14 and 15). After each step, the ant immediately performs a local pheromone update (Equation 16) to dynamically adjust the current pheromone distribution.

[0172] (2.2) Global pheromone update

[0173] Once all ants have completed path construction, a global pheromone update is performed.

[0174] a. Path evaluation: Calculate the comprehensive evaluation value of each path using the multi-objective evaluation function (Equation 7).

[0175] b. Differentiated update: Pheromones are enhanced on the current best path (Equation 17), and pheromone is penalized on the current worst path (Equation 18).

[0176] (3) Result output

[0177] After the iteration terminates, the globally optimal path obtained from the search is output as the planning result.

[0178] S3: Smooth the planned path:

[0179] The initial path planned by the improved ant colony algorithm is a polyline path formed by connecting a series of grid center points. While this path can achieve collision-free passage, it suffers from numerous turning points and abrupt turns, failing to meet the continuous and smooth kinematic constraints of a mobile robot. To address this issue, this invention further processes the planned path. First, a pruning strategy is used to optimize the path geometry by eliminating redundant nodes that do not contribute to connectivity, thereby simplifying the path structure and shortening its physical length. Next, Bézier curves are used to smoothly fit the retained critical path points, transforming the discrete polyline into a continuously differentiable optimized trajectory. Finally, a high-quality executable path is generated that maintains collision-free characteristics while conforming to the robot's kinematic constraints.

[0180] Specifically, in one embodiment, step S3 is as follows:

[0181] S31: Redundant node removal based on pruning strategy

[0182] This step aims to eliminate redundant nodes in the initial path that do not contribute to connectivity. This not only effectively shortens the physical length of the path but also reduces path points and computational complexity for subsequent smoothing processes. Its core principle is the collinearity criterion in path geometry: if three consecutive nodes on the path are collinear... , , satisfy arrive If the line connecting the nodes does not cross any obstacles, then the intermediate node... It is a redundant node and can be safely removed.

[0183] The specific steps are as follows:

[0184] (1) Initialization: Record the original path as a sequence of nodes. ,in Starting from point S, Create an empty list for endpoint E. and will Join in.

[0185] (2) Cyclic pruning: Assume the current The last node is .from middle Starting from the next node, check subsequent nodes in turn. For each pair Determine if the straight line segment connecting the two passes through an obstacle (i.e., check if the state of all grid cells the line segment passes through is "0"). Record the last one that can be connected to the obstacle. Nodes that achieve collision-free straight-line connections .

[0186] (3) Node update: join in and will Set as new .

[0187] (4) Termination condition: Repeat steps (2) and (3) until (The finish line). At this moment, This is the simplified path after removing redundant nodes.

[0188] S32: Path smoothing based on Bézier curves

[0189] A simplified path sequence is obtained by using a pruning strategy to remove redundant nodes. Then, a piecewise second-order Bézier curve was used for fitting. Specifically, for three consecutive nodes... , , , path segment The midpoint is set as the starting point of the curve, and the node is set as the starting point. Set as control point, path segment The midpoint is set as the endpoint of the curve. Then, the smooth path segment is generated using the formula for a second-order Bézier curve:

[0190] (19)

[0191] in, For the range The curve parameters are determined. All nodes are traversed to generate piecewise curves, which are then connected to form a smooth and differentiable robot trajectory.

[0192] In summary, the mobile robot path planning method based on the improved ant colony algorithm provided in this application introduces an initial pheromone allocation optimization strategy, combines heuristic information based on the triangle inequality, the turning angle suppression function, and the state transition rules of the artificial potential field, and adopts a differentiated pheromone update strategy based on multi-objective performance evaluation indicators. Finally, the planned path is smoothed, thereby significantly improving the convergence speed, path quality, and environmental adaptability of the algorithm.

Claims

1. A mobile robot path planning method based on an improved ant colony algorithm, characterized in that: Comprise the following steps: S1: adopt grid method to model the working environment of the robot, define the position of the starting point and the end point, and construct a multi-objective performance evaluation function of comprehensive path length, smoothness and safety; S2: adopt improved ant colony algorithm for optimal path search, path planning from the starting node to the target node on the grid map, the improvement of the ant colony algorithm includes: Non-uniform initialization method based on path global cost is used to distribute pheromone concentration; Improvement of state transition rule: fusion of pheromone concentration, heuristic information based on triangular inequality, turning angle suppression function and artificial potential field force; Differential pheromone update strategy, including local update, global update and punishment mechanism for the worst path; S3: path smoothing processing: through redundant node removal and parameterized curve fitting to generate continuous smooth trajectory.

2. The mobile robot path planning method based on the improved ant colony algorithm according to claim 1, characterized in that: The modeling step in S1 is as follows: The given two-dimensional planar space is evenly divided into a grid cell array, each grid cell is endowed with a state value to represent its physical properties: "0": indicates that the grid is a free area, and the robot can safely pass through; "1": indicates that the grid is an obstacle area, and the robot needs to avoid; Through the above mapping, a two-dimensional binary grid map is generated, in which the coordinates of the starting point and the target point of path planning are clearly set.

3. The mobile robot path planning method based on the improved ant colony algorithm according to claim 2, characterized in that: The multi-objective performance evaluation function construction method in the S1 is shown as follows: firstly, three sub-indexes, path length, smoothness and safety, are determined, then the comprehensive multi-objective performance evaluation function is constructed by using the weighted summation method which is used as the decision basis for searching the optimal path by the algorithm. （1） In the above equation, , , are the weight coefficients of the path length, smoothness and safety indicators, respectively, , , represent the path length, smoothness and safety indicators, respectively.

4. The mobile robot path planning method based on the improved ant colony algorithm according to claim 3, characterized in that: The calculation methods of the three sub-indexes are as follows: Path length index calculation method: （2） （3） wherein, is the total number of grids on the path; is the thgrid on the path whose coordinates are ; denotes the Euclidean distance between adjacent grids and ; The smoothness index calculation formula is as follows: （4） wherein denotes the angle between the vectors and ; is the total number of turning points in the path; and are the turning angle and the weight factor for the number of turns, respectively, and satisfy ; The safety index calculation formula is as follows: （5） （6） In the above equation, denotes the Euclidean distance of the th grid cell to its nearest obstacle, while denotes the coordinates of the nearest obstacle.

5. The mobile robot path planning method based on the improved ant colony algorithm according to claim 1, characterized in that: The calculation formula of the non-uniform initialization method in S2 is as follows: （7） In the above equation, is a pheromone intensity constant, denotes the Euclidean distance between points A and B, S and E are the start and end points of the path, respectively.

6. The mobile robot path planning method based on the improved ant colony algorithm according to claim 5, characterized in that: The information introduced in the state transition rule in S2 is as follows: Heuristic information based on the triangle inequality: , denotes the Euclidean distance between the current node and the candidate node , denotes the Euclidean distance between the candidate node and the end point E, denotes the Euclidean distance between the current node and the end point E; is a very small positive number to prevent division by zero; turning angle suppression function: wherein, is the turning angle formed when the robot moves from the current motion direction to the candidate direction; Artificial potential field force: including attractive factor, repulsive factor, and comprehensive artificial potential field force, where the gravitational factor is: , is the gravitational gain coefficient, is the Euclidean distance from the current node to the target point; repulsive force factor: , is a repulsive force gain coefficient, is the distance from the current node to the nearest obstacle, is the maximum influence radius of the repulsive force field; a synthetic artificial potential field force: , is a constant of adjustment for the sensitivity of the potential field; is an effective component of the attractive force in the candidate direction; is an effective component of the repulsive force in the candidate direction.

7. The mobile robot path planning method based on the improved ant colony algorithm according to claim 6, characterized in that: The state transition rule in S2 is represented as: (8） is a random number uniformly distributed in the interval [0, 1], when : ants select according to a probability calculated by （9） wherein, represents an ant at the current node a feasible neighborhood node set of the current node, consisting of the grids within its eight-neighborhood that are not occupied by obstacles and have not been visited by the ant in the current round; are weight coefficients of pheromone, heuristic information, turning angle inhibition function and artificial potential field, respectively.

8. The mobile robot path planning method based on the improved ant colony algorithm according to claim 1, characterized in that: The differential pheromone update strategy in S2 specifically includes: Local pheromone update: , is set to the local pheromone evaporation coefficient; is set to the aforementioned non-uniform initial pheromone ; Global pheromone update: first, calculate the multi-objective performance evaluation index value of the path constructed by each ant based on the multi-objective performance evaluation function formula , and determine the optimal path of this iteration and the worst path ; subsequently, the following operations are performed: Global enhancement: pheromone enhancement on edges of the contemporary optimal path ​ （10） wherein, is a global pheromone volatility coefficient, , is a pheromone intensity constant, is a multi-objective performance evaluation index value of the current optimal path. Penalty mechanism: additional weakening of pheromone on the worst path of the current generation ​ （11） wherein is a penalty coefficient, , is the multi-objective performance evaluation index value of the worst path of the current generation.

9. The mobile robot path planning method based on the improved ant colony algorithm according to claim 1, characterized in that: In S3, first, redundant node optimization is performed to simplify the path structure, and then parameterized curve fitting method is used for smooth processing of key path points.

10. A mobile robot path planning device based on an improved ant colony algorithm, characterized by, It includes: Environment modeling module: used to model the working environment of the robot by using grid method, define the position of the starting point and the end point; Multi-objective evaluation module: used to construct a multi-objective performance evaluation function of comprehensive path length, smoothness and safety; Improved ant colony algorithm module: used for optimal path search on the grid map, the improved ant colony algorithm module includes initial pheromone optimization unit, multi-factor state transition unit and differential update unit, wherein the initial pheromone optimization unit uses non-uniform initialization method based on path global cost, the multi-factor state transition unit fuses pheromone concentration, heuristic information, turning angle suppression and artificial potential field force; the differential update unit is used to realize local update, global update and worst path punishment mechanism; Path smoothing processing module: used to generate continuous smooth trajectory through redundant node removal and curve fitting; Path output module: used to output the final planned executable path.

Images