Mobile robot multi-dimensional path planning method based on AROA improved optimization algorithm

Abstract

The invention relates to a mobile robot multi-dimensional path planning method based on an AROA improved optimization algorithm, and the method comprises the steps: obtaining a task scene of a mobile robot, building a grid map model through the task scene, and marking a path starting point, a target point and obstacle information; constructing a multi-objective cost function, and performing path planning on the grid map model by adopting an improved optimization algorithm based on AROA; the improved optimization algorithm based on AROA is obtained by fusing an AROA optimization algorithm with a BKA optimization algorithm; in the path planning process, the multi-target cost function is used for calculating the path adaptability of population individuals, and the individual with the minimum path adaptability serves as the optimal path individual to be used for obtaining the optimal path of the mobile robot. The method can effectively solve the problem that the path planning of the mobile robot is easy to fall into local optimum and slow in convergence speed in a complex multi-dimensional environment, and the precision and engineering practicability of the path planning are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of autonomous navigation technology for mobile robots, and in particular to a multi-dimensional path planning method for mobile robots based on an improved AROA optimization algorithm. Background Technology

[0002] Mobile robots, with their high flexibility and strong environmental adaptability, have been widely used in many fields such as industry and civilian use. One of the core technologies for their autonomous operation is path planning, which is to design the optimal path in an environment with obstacles, satisfying the requirements of "shortest path, safe obstacle avoidance, and smooth movement". This directly affects the robot's operating efficiency and task reliability.

[0003] In existing path planning techniques, traditional algorithms such as While Dijkstra's algorithm is simple in principle, its computational complexity increases exponentially with the complexity of the environment. It has poor real-time performance in large-scale multi-obstacle scenarios and is difficult to adapt to dynamically changing task requirements. Sampling-based algorithms (such as the fast expanding random tree algorithm) can cope with high-dimensional environments, but the path quality depends on subsequent optimization, and redundant transitions are prone to occur, increasing the robot's motion energy consumption and control difficulty.

[0004] In recent years, metaheuristic optimization algorithms have become a research hotspot in the field of path planning: Attraction-Repulsion Optimization (AROA) maintains population diversity through the attraction-repulsion interaction between individuals, has outstanding global exploration capabilities, and can effectively cover the search space of complex environments, but it has the drawbacks of insufficient local development accuracy and low search efficiency in the later stages of convergence; Black-winged Kite Algorithm (BKA) simulates the attack and migration behavior of black-winged kites, has excellent local optimization performance, and can quickly fine-tune path details, but its global search range is limited and it is easy to get trapped in local optima, resulting in problems such as "inflexible obstacle avoidance" or "path redundancy" in the planned path.

[0005] Furthermore, existing methods often focus on a single constraint in their path cost function (such as considering only path length or obstacle avoidance safety), neglecting key engineering requirements such as potential energy cost and path smoothness, resulting in insufficient practicality of the planned path. At the same time, the population initialization often adopts a random distribution method, which lacks diversity and further limits the search efficiency and optimization accuracy of the algorithm, failing to meet the engineering application needs of mobile robots in complex scenarios. Summary of the Invention

[0006] To address the shortcomings of existing single optimization algorithms in mobile robot path planning scenarios, such as slow convergence speed, susceptibility to local optima, and insufficient practicality of planned paths, this invention provides a multi-dimensional path planning method for mobile robots based on an improved AROA optimization algorithm, enabling efficient, safe, and smooth path planning for mobile robots in complex environments.

[0007] To achieve the above objectives, the present invention provides the following solution: A multi-dimensional path planning method for mobile robots based on an improved AROA optimization algorithm, comprising: Obtain the task scenario of the mobile robot, use the task scenario to build a grid map model, and mark the path start point, target point and obstacle information; A multi-objective cost function is constructed, and a path planning algorithm based on AROA is used to perform path planning on the grid map model; the AROA-based improved optimization algorithm is obtained by fusing the AROA optimization algorithm with the BKA optimization algorithm. During the path planning process, the path suitability of individuals in the population is calculated using the multi-objective cost function. The individual with the lowest path suitability is selected as the optimal path individual to obtain the optimal path for the mobile robot.

[0008] Optionally, the obstacle information includes: threat radius, device size, and danger spread distance.

[0009] Optionally, constructing the multi-objective cost function includes: ; in, These are the weighting coefficients. For a multi-objective cost function, For path length cost, The cost of avoiding obstacles As a cost of potential energy, This is the cost of path smoothing.

[0010] Optionally, the path length cost includes: ; in, This represents the total number of path nodes. For the first The coordinates of each path node. For the first The coordinates of each path node; The cost of obstacle avoidance includes: ; in, The total number of obstacles. The shortest distance from the obstacle to the path segment. Threat radius, For equipment dimensions, Extend the distance to the danger zone; The potential energy cost includes: ; in, The total potential energy of the path nodes is obtained by superimposing the repulsive potential energy and the gravitational potential energy; The path smoothing cost includes: ; in, These are vectors representing adjacent path segments.

[0011] Optionally, path planning for the grid map model using an improved optimization algorithm based on AROA includes: Step 1: Set the population size and maximum number of iterations, and generate an initial population through a hybrid chaotic mapping seed model, and select the initial optimal individuals in the initial population that satisfy the grid boundary constraints; Step 2: Perform population iterative update on the initial optimal individual, i.e. calculate the energy factor. When the energy factor is greater than the target threshold, perform global exploration. When the energy factor is not greater than the target threshold, perform local development. Expand the search boundary with an improved FADs behavior simulation mechanism and retain the historical optimal individual in combination with the memory mechanism. Step 3: Repeat step 2 until the maximum number of iterations is reached, and use the multi-objective cost function to calculate the path suitability of the individuals in the population. The individual with the lowest path suitability is selected as the optimal path individual to obtain the optimal path for the mobile robot.

[0012] Optionally, performing the global exploration includes: generating a direction vector based on the AROA attraction-repulsion operator and combining it with Lévy flight to generate a search step size, which is used to expand the population search range. ; Where L(s) is the Levy flight search step size, which controls the span of the population's global exploration; β is the gamma function used to generate a random step size that conforms to the Lévy distribution; β is the step size adjustment parameter, with a value range of 1.0-1.8, balancing the randomness and stability of the search; s is the distance factor between the current individual in the population and the historical best individual.

[0013] Optionally, performing the local development includes: simulating BKA attacks and migration behavior to generate candidate locations, and introducing Cauchy distribution for local fine-tuning; The probability density function of the Cauchy distribution is: ; in, For scale parameters, For position parameters.

[0014] Optionally, the improved FADs behavior simulation mechanism extends the search boundary by including: ; Where J is a random matrix with elements ranging from 0 to 1, used to generate random search directions. The logical matrix elements are either 0 or 1, used to filter valid search positions; 1 indicates that the position is retained, and 0 indicates that it is discarded. A random number between 0 and 1, used to switch between the two search modes. Two individuals were randomly selected to introduce population diversity. These are the coordinates of the individual's position in the population after the (i+1)th iteration. Let be the coordinates of the individual's position in the population at the i-th iteration. Here, t is the decay coefficient, t is the current iteration number, and iter is the maximum iteration number. This is used to reduce search fluctuations in later stages. The minimum coordinate boundary of an individual's location within the population is determined by the effective range of the grid map. The maximum coordinate boundary of an individual's location within the population is determined by the effective range of the grid map.

[0015] Optionally, the population iterative update of the initial optimal individual further includes: Calculate the AROA decay parameter to adjust the global exploration intensity: ; Calculate the BKA inertia weight to balance the proportion of local development: ; in, This represents the current iteration number. This represents the maximum number of iterations.

[0016] The beneficial effects of this invention are as follows: Faster convergence speed: This invention integrates AROA and BKA, combining Levy flight global exploration and Cauchy distribution local fine-tuning, which improves the convergence speed by more than 25% compared to the single AROA algorithm, and can reach stability after 35 iterations in a 20×20 grid map; Superior path quality: The multi-objective cost function of this invention integrates path length, obstacle avoidance safety, potential energy cost and smoothness. The planned path length is shortened by 12.3% and the smoothing cost is reduced by 26.8% compared with the single BKA algorithm, thereby reducing robot motion energy consumption and control difficulty. Enhanced environmental adaptability: The hybrid chaotic mapping seed model of this invention improves the diversity of the initial population, and the improved FADs behavior simulation mechanism expands the search boundary, achieving a 100% obstacle avoidance success rate in multi-obstacle scenarios, and improving real-time performance by 40% compared to the traditional A* algorithm in large-scale environments; More practical for engineering: This invention is adaptable to multiple scenarios such as indoor services and industrial inspections. It can meet different needs by adjusting the weight coefficients (such as increasing the path length weight in industrial scenarios and increasing the obstacle avoidance weight in indoor scenarios). It has low deployment costs and strong compatibility. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a multi-dimensional path planning method for a mobile robot based on an improved AROA optimization algorithm according to an embodiment of the present invention. Figure 2 The improved algorithm of this invention, compared with the original algorithm, is used in grid-based applications. Figure 1 Convergence comparison chart below; Figure 3 The improved algorithm of this invention, compared with the original algorithm, is used in grid-based applications. Figure 1 Path comparison chart; Figure 4 The improved algorithm of this invention, compared with the original algorithm, is used in grid-based applications. Figure 2 Convergence comparison chart below; Figure 5 The improved algorithm of this invention, compared with the original algorithm, is used in grid-based applications. Figure 2 Path comparison chart; Figure 6 This is a comparison graph showing the convergence of the improved algorithm and the original algorithm under the same test function in this embodiment of the invention. Figure 7 The improved algorithm of this invention, compared with other commonly used single algorithms, is used in grid-based applications. Figure 2 The path diagram. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] like Figure 1 As shown in the figure, this embodiment discloses a multi-dimensional path planning method for mobile robots based on an improved AROA optimization algorithm, including: acquiring the task scenario of the mobile robot, establishing a grid map model using the task scenario, and marking the path start point, target point, and obstacle information; constructing a multi-objective cost function, and using an improved AROA optimization algorithm to perform path planning on the grid map model; the improved AROA optimization algorithm is obtained by fusing the AROA optimization algorithm with the BKA optimization algorithm; during the path planning process, the multi-objective cost function is used to calculate the path suitability of individuals in the population, and the individual with the lowest path suitability is taken as the optimal path individual to obtain the optimal path for the mobile robot.

[0022] Specifically, this embodiment discloses a multi-dimensional path planning method for mobile robots based on an improved AROA optimization algorithm, the method comprising the following steps: Step 1: Construct a grid map model: Based on the mobile robot task scenario, generate a 20×20 grid map, define its effective coordinate range as [1,20]×[1,20], and mark the path start point, target point, and obstacle information; Furthermore, the obstacle includes parameters such as threat radius R, device size D, and danger extension distance S, which are integrated into a map structure to provide basic environmental data.

[0023] Step 2: Construct a multi-objective path cost function: Integrating four types of constraints—path length, obstacle avoidance, potential energy cost, and path smoothness—a weighted summation multi-objective cost function is established: ; in, These are the weighting coefficients; Furthermore, the mathematical expressions for each cost component are: path length cost : ; in, This represents the total number of path nodes. For the first The coordinates of each path node; Obstacle avoidance costs : ; in, The total number of obstacles. The shortest distance from the obstacle to the path segment; Potential energy cost : ; in, (The total potential energy of the path nodes is obtained by superimposing the repulsive potential energy and the gravitational potential energy). Path smoothing cost : ; in, These are vectors representing adjacent path segments.

[0024] Step 3: Population Initialization: Set the population size Maximum number of iterations Parameters such as these are used to generate an initial population through a hybrid chaotic mapping seed model; Furthermore, the hybrid chaotic mapping seed model is as follows: ; in, For random modulation parameters, For the coordinates of an individual's location within the population, The coordinates of the individual population positions at the nth iteration (n is the iteration number, and the initial coordinates are randomly initialized when n=0). The optimal initial individual that satisfies the grid boundary constraints is selected as the "basic reference solution" for subsequent iterations, avoiding extremely poor position coordinates during population initialization (such as falling directly into obstacles), thereby accelerating the convergence speed of the algorithm. The selection logic for subsequent iterations: Subsequent iterations combine the initial optimal individual with other initial population individuals that satisfy the constraints to form an iterative population. Through competition and updates among individuals within the population, a better path solution is gradually optimized.

[0025] Step 4: Dynamic parameter update: During the iteration process, the AROA attenuation parameter is calculated: ; Calculate the BKA inertia weight: ; in, This represents the current iteration number; Furthermore, θ decreases linearly with each iteration (θ = 2 × (1 − t / iter)). (where iter is the current iteration number and iter is the maximum iteration number). In the early stages of iteration, the θ value is large, resulting in high global exploration intensity and an expanded search range; in the later stages of iteration, the θ value is small, leading to a decrease in global exploration intensity.

[0026] w decays cosine as iteration progresses (w=0.5×(1+cos(πt / iter))): In the early stage of iteration, w is large and the proportion of local development is low; in the later stage of iteration, w is small and the proportion of local development gradually increases. This parameter change is related to the "local development stage" in step 5: when w decays to the threshold (0.3 in this invention), the algorithm will prioritize the local development operation, use the Cauchy distribution to fine-tune the current better individual, and thus accurately optimize the path details after "globally exploring the better region".

[0027] Step 5: Population Iterative Update: a. Calculate the energy factor: ,according to The numerical switching search phase; b. When At that time, a global exploration is performed: the direction vector is generated based on the AROA attraction-repulsion operator, and the search step size is generated by combining the Levy flight; Furthermore, the Levy flight step size formula is as follows: ; Where L(s) is the Levy flight search step size, which controls the span of the population's global exploration; β is the gamma function used to generate a random step size that conforms to the Lévy distribution; β is the step size adjustment parameter, with a value ranging from 1.0 to 1.8, preferably 1.5, to balance the randomness and stability of the search; s is the distance factor between the current individual in the population and the historical best individual. c. When At the same time, perform local development: simulate BKA attack / migration behavior to generate candidate locations, and introduce Cauchy distribution for local fine-tuning; Furthermore, the probability density function of the Cauchy distribution is: ; in, For position parameters, For scale parameters; d. An improved FADs behavior simulation mechanism expands the search boundary and, combined with a memory mechanism, retains historically optimal individuals to avoid losing high-quality path solutions found during iteration. This prevents the algorithm from deviating from the optimal region due to random updates of individuals in the population and provides a "high-quality reference benchmark" for subsequent iterations. In each iteration, the improved FADs behavior simulation mechanism first expands the search boundary to generate new individuals, then compares the fitness of the new individuals with the "historically optimal individuals." If the new individuals have better fitness, the historically optimal individuals are updated. The current historically optimal individuals are then directly applied to the next iteration. This process is repeated until the preset number of iterations is reached, and finally, the path corresponding to the individual with the lowest fitness is output.

[0028] Furthermore, the mathematical model for the search behavior is as follows: ; in, J is a random matrix. For logical matrices, A random number between 0 and 1. Two individuals were randomly selected.

[0029] Step 6: Output the optimal path: If the number of iterations reaches... Output the path corresponding to the individual with the lowest fitness; otherwise, return to step 4 to continue iterating.

[0030] In one embodiment, the following describes the implementation of the present invention in detail using indoor service robot path planning as an example: S1: Grid map model construction: Generate a 20×20 grid map with coordinate range [1,20]×[1,20], set the starting point to (0,0) and the target point to (20,20); Furthermore, obstacles are set up, each with a threat radius R=5, device size D=8, and danger extension distance S=4, and integrated into a map structure.

[0031] S2: Path cost function settings: Weight coefficients are set to... , , , ; Furthermore, priority should be given to increasing the weight of obstacle avoidance costs to ensure obstacle avoidance safety in indoor environments.

[0032] S3: Algorithm parameter settings: Population size Maximum number of iterations probability of attack behavior ; Furthermore, the random modulation parameters of the hybrid chaotic mapping The probability of FADs occurring is 0.4.

[0033] S4: Algorithm Execution and Results: Execute the improved AROA optimization algorithm according to the above steps, and output the optimal path after 100 iterations; Furthermore, experimental results show that the optimal path length is 113.6, which is 12.3% shorter than the single AROA algorithm; the path smoothing cost is 1.27 rad, which is 26.8% lower than the single BKA algorithm; the algorithm tends to stabilize after the 35th iteration, and the convergence speed is more than 25% faster than the single algorithm. Figure 2-7 As shown.

[0034] Furthermore, this invention can also be adapted to industrial inspection robot scenarios: setting a total of 12 obstacles with weighting coefficients. Population size Its planned path length is shortened by 10.5% compared to a single algorithm, and its stability is improved by 35%, meeting the real-time and reliability requirements of industrial scenarios.

[0035] In summary, this invention achieves efficient path planning for mobile robots in complex environments by combining the AROA-BKA fusion optimization algorithm with a multi-objective cost function, and has broad engineering application value.

[0036] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A multi-dimensional path planning method for mobile robots based on an improved AROA optimization algorithm, characterized in that, include: Obtain the task scenario of the mobile robot, use the task scenario to build a grid map model, and mark the path start point, target point and obstacle information; A multi-objective cost function is constructed, and a path planning is performed on the grid map model using an improved optimization algorithm based on AROA. The AROA-based improved optimization algorithm is obtained by fusing the AROA optimization algorithm with the BKA optimization algorithm. During the path planning process, the path suitability of individuals in the population is calculated using the multi-objective cost function. The individual with the lowest path suitability is selected as the optimal path individual to obtain the optimal path for the mobile robot.

2. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 1, characterized in that, The obstacle information includes: threat radius, equipment size, and danger spread distance.

3. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 1, characterized in that, Constructing the multi-objective cost function includes: ； in, These are the weighting coefficients. For a multi-objective cost function, For path length cost, The cost of avoiding obstacles As a cost of potential energy, This is the cost of path smoothing.

4. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 3, characterized in that, The path length cost includes: ； in, This represents the total number of path nodes. For the first The coordinates of each path node. For the first The coordinates of each path node; The cost of obstacle avoidance includes: ； in, The total number of obstacles. The shortest distance from the obstacle to the path segment. Threat radius, For equipment dimensions, Extend the distance to the danger zone; The potential energy cost includes: ； in, The total potential energy of the path nodes is obtained by superimposing the repulsive potential energy and the gravitational potential energy; The path smoothing cost includes: ； in, These are vectors representing adjacent path segments.

5. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 1, characterized in that, The path planning for the grid map model using the improved AROA-based optimization algorithm includes: Step 1: Set the population size and maximum number of iterations, and generate an initial population through a hybrid chaotic mapping seed model, and select the initial optimal individuals in the initial population that satisfy the grid boundary constraints; Step 2: Perform population iterative update on the initial optimal individual, i.e. calculate the energy factor. When the energy factor is greater than the target threshold, perform global exploration. When the energy factor is not greater than the target threshold, perform local development. Expand the search boundary with an improved FADs behavior simulation mechanism and retain the historical optimal individual in combination with the memory mechanism. Step 3: Repeat step 2 until the maximum number of iterations is reached, and use the multi-objective cost function to calculate the path suitability of the individuals in the population. The individual with the lowest path suitability is selected as the optimal path individual to obtain the optimal path for the mobile robot.

6. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 5, characterized in that, Performing the global exploration includes: generating a direction vector based on the AROA attraction-repulsion operator, and combining this with Lévy flight to generate a search step size, which is used to expand the population search range. ； Where L(s) is the Levy flight search step size, which controls the span of the population's global exploration; β is the gamma function used to generate a random step size that conforms to the Lévy distribution; β is the step size adjustment parameter, with a value range of 1.0-1.8, balancing the randomness and stability of the search; s is the distance factor between the current individual in the population and the historical best individual.

7. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 5, characterized in that, The local development process includes: simulating BKA attacks and migration behavior to generate candidate locations, and introducing Cauchy distribution for local fine-tuning; The probability density function of the Cauchy distribution is: ； in, For scale parameters, For position parameters.

8. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 5, characterized in that, The improved FADs behavior simulation mechanism extends the search boundary by including: ； Where J is a random matrix with elements ranging from 0 to 1, used to generate random search directions. The logical matrix elements are either 0 or 1, used to filter valid search positions; 1 indicates that the position is retained, and 0 indicates that it is discarded. A random number between 0 and 1, used to switch between the two search modes. Two individuals were randomly selected to introduce population diversity. These are the coordinates of the individual's position in the population after the (i+1)th iteration. Let be the coordinates of the individual's position in the population at the i-th iteration. Here, t is the decay coefficient, t is the current iteration number, and iter is the maximum iteration number. This is used to reduce search fluctuations in later stages. The minimum coordinate boundary of an individual's location within the population is determined by the effective range of the grid map. The maximum coordinate boundary of an individual's location within the population is determined by the effective range of the grid map.

9. The multi-dimensional path planning method for mobile robots based on the improved AROA optimization algorithm according to claim 5, characterized in that, The population iterative update of the initial optimal individual also includes: Calculate the AROA decay parameter to adjust the global exploration intensity: ； Calculate the BKA inertia weight to balance the proportion of local development: ； in, This represents the current iteration number. This represents the maximum number of iterations.

Images