Obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method

By combining simulated annealing genetic algorithm and improved artificial potential field method, the efficiency and safety problems of robot path planning in complex environments are solved, and faster and safer target point arrival is achieved.

CN118915741APending Publication Date: 2024-11-08CHINA SHIP DEV & DESIGN CENT
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202410966989.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

It is difficult to find the optimal solution in complex environments for existing robot obstacle avoidance planning, and traditional path planning algorithms are slow to converge and have high computational costs, making it difficult to quickly and safely reach the target point.

Method used

Combining simulated annealing genetic algorithm and improved artificial potential field method, by constructing a grid map, using genetic algorithm to perform global path planning, and introducing simulated annealing algorithm and fill potential field, optimizing path selection and variation operations, reducing repulsive obstacles to the robot The impact of

Benefits of technology

It improves the efficiency and safety of robot path planning, reduces the probability of local minimum point traps, and allows robots to reach target points faster and safer, and adapt to changes in complex environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118915741A_ABST
    Figure CN118915741A_ABST
Patent Text Reader

Abstract

The invention discloses an obstacle avoidance method combining a simulated annealing genetic algorithm and an improved artificial potential field method, and the method comprises the steps: constructing a grid map of an environment, and determining an obstacle grid and a passable grid; performing global path planning based on a genetic simulated annealing algorithm to obtain a global planning path; and according to the global planning path, moving in a resultant potential field of an improved repulsive potential energy function. The method further comprises the following steps: detecting whether a local minimum point is trapped; if yes, introducing a filling potential field; and if not, continuing to move along the global planning path until the target point is reached. According to the method, the traditional genetic algorithm is introduced into the simulated annealing thought, the optimal path planning can be faster, the global optimal solution can be obtained more easily, the improved artificial potential field method is fused on the basis of the genetic simulated annealing algorithm, and the two methods make the best of each other, so that the probability that the mobile robot falls into the local minimum point is effectively reduced, and the optimal path planning speed is improved. And the robot can reach the target point more safely.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of obstacle avoidance, and in particular relates to an obstacle avoidance method combining a simulated annealing genetic algorithm with an improved artificial potential field method. Background Art

[0002] During the robot's motion process, if the environmental information is known, an appropriate collision avoidance path can be acquired and planned in advance. However, in a real environment, the mobile robot must obtain local environmental information and process this information comprehensively in order to obtain a feasible collision-free path. Although this approach has its advantages, the entire planning process still needs to be addressed and improved to ensure that the robot can respond to various environmental changes quickly and accurately and perform its tasks.

[0003] When environmental information cannot be obtained in advance, the robot must rely on its own sensors to scan in real time to obtain environmental information and perform obstacle avoidance operations. This type of robot obstacle avoidance planning has very high requirements, and there are still many problems that need to be solved. Despite this, this type of robot obstacle avoidance planning is the most widely used and has attracted a lot of attention and research. Researchers are working to solve the problems encountered, improve the obstacle avoidance performance of this robot, and provide support and guarantee for its wide application in future industries, services, and medical fields.

[0004] At present, robot path planning mainly faces the following problems: First, as the scale of the environment increases, the number of obstacles and the complexity of geometric shapes increase, making it difficult for the robot to reach the target point safely and quickly. Second, a single path planning algorithm can only run correctly in a relatively simple environment. When the environmental information increases, a single path planning algorithm cannot find the optimal solution in a large-scale search space, and its performance will also be affected, and the environmental adaptability is still insufficient. In addition, traditional path planning algorithms face bottlenecks such as slow convergence speed, long running time, and high computational cost. Summary of the invention

[0005] The purpose of the present invention is to provide an obstacle avoidance method combining a simulated annealing genetic algorithm with an improved artificial potential field method, so that a robot can reach a target point more quickly and safely.

[0006] The technical solution adopted by the present invention is:

[0007] According to a first aspect of the present invention, there is provided an obstacle avoidance method combining a simulated annealing genetic algorithm with an improved artificial potential field method, the method comprising:

[0008] Construct a grid map of the environment and determine the obstacle grid and the traversable grid;

[0009] Perform global path planning based on genetic simulated annealing algorithm to obtain the global planning path;

[0010] According to the global planning path, the resultant potential field of the improved repulsive potential energy function moves.

[0011] Furthermore, the method also includes:

[0012] Detect whether it falls into a local minimum point; if so, introduce a filling potential field; if not, continue to move along the global planning path until reaching the target point.

[0013] Furthermore, global path planning is performed based on the genetic simulated annealing algorithm to obtain a global planning path, including:

[0014] Number each grid and determine the starting and ending points;

[0015] The initial population is constructed using a genetic algorithm. Each chromosome individual in the population represents an obstacle-free path from the starting point to the end point, corresponding to a set of grid numbers from the starting point to the end point.

[0016] The idea of ​​​​simulated annealing algorithm is introduced in the selection operation of genetic algorithm, as follows:

[0017] Use the fitness function to calculate the fitness value of each individual in the current population, that is, the fitness value corresponding to each path;

[0018] Select individuals whose fitness values ​​are higher than the preset threshold or the top a% of the highest fitness values ​​as elite individuals;

[0019] Randomly select two chromosome individuals i and j from the remaining non-elite individuals and calculate the corresponding fitness values;

[0020] Based on the principle of simulated annealing algorithm, one of chromosome individuals i and j is selected; if fit(i)>fit(j), chromosome individual j is directly selected; otherwise, chromosome individual j is selected with probability p;

[0021]

[0022] Where T is the current temperature, which gradually decays; the expression of the current temperature T is T = T0 × v g , T0 represents the initial temperature, v represents the decay rate, g represents the number of iterations; fit(i) and fit(j) are the fitness values ​​of chromosome individuals i and j respectively;

[0023] Perform crossover and mutation operations on the elite individuals and the selected individuals, and recalculate their fitness values ​​until the optimal global planning path is obtained.

[0024] Furthermore, the method for constructing the initial population is:

[0025] Randomly select a series of traversable grids as path nodes and connect them with the starting point and the end point to form an obstacle-free path from the starting point to the end point;

[0026] There cannot be any obstacles or duplicate nodes on the path, and the grid corresponding to each node must be passable.

[0027] Furthermore, path distance and path smoothness are used as evaluation indicators for path planning, and their weighted summation is used to obtain the fitness function, as follows:

[0028] fit=ω1fit1+ω2fit2

[0029]

[0030]

[0031]

[0032]

[0033] Where fit is the fitness value, ω1 and ω2 are the weighting factors of fitness fit1 and fit2 respectively; i represents the number of path nodes, end represents the number of path nodes, and x and y represent the coordinates of path nodes;

[0034] The shorter and smoother the path, the higher the fitness value.

[0035] Furthermore, the crossover includes:

[0036] Randomly select two parent chromosomes from the population to form a group, and find the same node between the two chromosomes; perform a crossover operation on the node; if there are multiple identical nodes, randomly select any one of them for crossover;

[0037] If a pair of parent chromosomes do not have a common gene sequence, two nodes are randomly selected from the pair of chromosomes for crossover; if the path obtained by crossover is discontinuous, the method of constructing the initial population is used to make it a continuous feasible path, and the duplicate nodes contained in the daughter chromosomes are deleted.

[0038] Further, the variations include:

[0039] First, two nodes are randomly selected from the parent individuals as mutation points, excluding the starting point and the end point; the parent chromosomes are divided into three paths, namely the first path, the middle path and the tail path;

[0040] The middle path is retained; the new first path and tail path are regenerated using the method of constructing the initial population;

[0041] The new first path, the middle path and the original tail path are recombined, and duplicate nodes are deleted; similarly, the original first path, the middle path and the new tail path are recombined, and duplicate nodes are deleted.

[0042] Furthermore, the improved repulsive potential energy function expression is:

[0043]

[0044] Where U rep is the repulsive potential energy, k d is the adaptive repulsive field adjustment coefficient, k rep is the repulsive force scale factor, ||ww o || is the distance between the object and the obstacle, ρ0 is the influence radius of each obstacle, ||ww g || n is the distance between the object and the end point to the power of n.

[0045] Furthermore, the calculation formula of the adaptive repulsive field adjustment coefficient is as follows:

[0046]

[0047] In the formula, k d is the adaptive repulsive field adjustment coefficient; d is the minimum distance from the current position of the object to the global planning path; L is the total length of the global planning path; k0 is the repulsive field constant factor.

[0048] According to a second aspect of the present invention, there is provided a mobile robot, which adopts the obstacle avoidance method combining the simulated annealing genetic algorithm described in any one of the above with the improved artificial potential field method.

[0049] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0050] The present invention proposes an algorithm that integrates a genetic algorithm and a simulated annealing algorithm. Specifically, a simulated annealing algorithm is introduced in the selection operation, which can find the global optimal solution in the genetic algorithm in a random manner and provide a powerful local search capability. Therefore, the two algorithms complement and combine each other to effectively solve the premature convergence problem. In addition, the present invention also improves the repulsive potential energy function to reduce the influence of repulsive obstacles on the robot. By multiplying the repulsive potential energy function by the nth power of the distance between the mobile robot and the end point, it can be ensured that the target point has the minimum global potential energy. On the other hand, the filling potential field method is used to solve the local minimum point problem. When the robot is stuck in a local minimum point, the filling potential field is introduced to fill the potential energy into the gap to help the robot escape and continue to move along the correct path to reach the target point. By applying the optimized algorithm to the robot path planning, a hybrid algorithm based on the genetic algorithm and the simulated annealing algorithm is used to avoid obstacles. When modeling the environment, the grid method is selected, and the obstacle information is transmitted in combination with the robot perception system, so as to design an intelligent robot obstacle avoidance algorithm in a complex static obstacle environment. The working principle of the algorithm is as follows: First, the robot uses its perception system to obtain environmental information, and then converts the information into subdivided grids with different costs to build a grid map of the environment. Next, the optimized hybrid algorithm is introduced into the planning system, using the efficient global exploration ability of the genetic algorithm and the local exploration ability of the simulated annealing algorithm to find the optimal path for the robot. Finally, the robot follows the path and continuously perceives the surrounding environment in order to make timely adjustments and replanning. This technology can not only improve the navigation performance and planning efficiency of the robot, but also expand its application scope, so that intelligent robots can better adapt to various real-world scenarios.

[0051] Introducing the traditional genetic algorithm into the simulated annealing idea can make the optimal path planning faster and easier to obtain the global optimal solution. As the scale of the environment expands and the complexity of the environment increases geometrically, we should integrate the genetic simulated annealing algorithm with the improved artificial potential field method to complement each other, thereby effectively reducing the probability of the mobile robot falling into the local minimum point and allowing the robot to reach the target point more safely.

[0052] In the obstacle avoidance algorithm designed by the present invention, the grid method can make the intelligent mobile robot understand the complexity of the environment in advance, so as to further think and analyze. The genetic algorithm is the important core of the present invention. It is equivalent to the "brain thinking" in the robot, which can make the mobile robot intelligent and quickly plan the optimal path after understanding the environmental information. The genetic algorithm has the advantages of being able to find the optimal solution in a large-scale search space and maintaining strong robustness, strong adaptability and high efficiency. This is very suitable for the path planning of intelligent mobile robots in complex environments. While the genetic algorithm gives the mobile robot the ability to perform theoretical analysis, it should also consider the problems that arise during the actual movement process, and the improved artificial potential field method can do this. The artificial potential field method allows the robot to move in a virtual force field, so that the robot can effectively avoid encountering obstacles. The organic combination of the two enables the robot to have the ability of "theoretical understanding" and "practical activities" like humans at the same time, so that the robot can reach the target point more quickly and safely. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 To optimize the traditional artificial potential field method algorithm flow chart;

[0054] Figure 2 Design a flow chart for the robot's obstacle avoidance algorithm. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0056] This paper analyzes and studies the currently widely used intelligent mobile robot obstacle avoidance planning algorithms, focusing on the genetic algorithm and the artificial potential field method, compares their respective advantages and disadvantages, proposes an optimization scheme and verifies its effectiveness. By using an obstacle avoidance method based on a hybrid algorithm of a genetic algorithm and a simulated annealing algorithm, the space is modeled using the grid method, and an intelligent robot obstacle avoidance algorithm is designed in a complex static obstacle environment. The main contents of this paper are as follows:

[0057] (1) Focus on two local path planning algorithms: artificial potential field method and genetic algorithm. Analyze and compare their advantages and disadvantages.

[0058] The artificial potential field method is a method for path planning by converting the robot's motion into movement in a virtual potential field. This algorithm regards the motion space of the mobile robot as a virtual force field. It has the advantages of simple description, small amount of calculation, and suitability for real-time control, and has been widely used. Although this method is prone to some problems, such as local minimum and inability to reach the target point, this is because it only obtains local part of the environmental information and lacks the ability to analyze global information. Therefore, when using the artificial potential field method, it is necessary to adjust and optimize it according to the specific situation in order to better play its role.

[0059] Genetic algorithm is an algorithm with efficient obstacle avoidance path planning and search capabilities. It can use the objective function value as search information, and does not require the process of deriving or differentiating the objective function value, so the optimal path can be obtained faster. In addition, the search of genetic algorithms is based on probabilistic rules, so it is flexible and has strong adaptability to environmental changes. However, genetic algorithms also have disadvantages. For example, in order to apply genetic algorithms, some populations need to be initialized in advance. Since not all populations are necessary, some unnecessary populations may be involved, which increases the complexity of subsequent calculations. In addition, various operators in genetic operations are difficult to determine, which may affect the accuracy and efficiency of the algorithm. Therefore, when using genetic algorithms, it is necessary to combine the specific situation and select and optimize relevant parameters according to the specific needs of the problem to achieve the best effect.

[0060] The two algorithms are compared. The artificial potential field method is mainly used in mobile robot path planning. Compared with the genetic algorithm, its principle is simpler, the amount of calculation is relatively small, and the path planning speed is faster. Although the genetic algorithm has a strong global search capability, it also has the defects of early convergence and slow convergence speed. In practical applications, its execution efficiency is not high.

[0061] (2) Optimize the design of the above two obstacle avoidance algorithms and propose an optimization scheme to avoid entering the local optimum and enable the intelligent robot to reach the target point quickly and effectively. Compare and analyze the optimized algorithm with the unoptimized algorithm to verify its effectiveness.

[0062] There are two problems in the practical application of the traditional artificial potential field method: the unreachable target problem and the local minimum point problem. The unreachable target point problem means that when the mobile robot approaches the target point, the repulsive force of the obstacle is equal to the gravitational force generated at the target, which causes the mobile robot to oscillate; while the local minimum problem means that the mobile robot is restricted from reaching the target point due to the mutual cooperation of the repulsive forces of various obstacles. In order to solve these problems, the repulsive potential energy function can be improved to reduce the impact of repulsive obstacles on the robot. By multiplying the repulsive potential energy function by the nth power of the distance between the mobile robot and the end point, it can be ensured that the target point has the minimum global potential energy. On the other hand, the filling potential field method is used to solve the local minimum point problem. When the robot is stuck in a local minimum point, the filling potential field is introduced to fill the potential energy into the gap to help the robot escape and continue along the correct path to reach the target point.

[0063] Traditional genetic algorithms mainly face the problem of premature convergence. They may stop searching for the optimal solution when the search space has not been fully explored and fall into a local optimal solution. This shows that the genetic algorithm pays too much attention to the details of the data set when improving adaptability, resulting in unsatisfactory search results. In order to solve this problem, the present invention introduces a simulated annealing algorithm and proposes an algorithm that combines a genetic algorithm and a simulated annealing algorithm. Specifically, the simulated annealing algorithm is introduced in the selection operation, which can find the global optimal solution in a random manner in the genetic algorithm, while providing a powerful local search capability. Therefore, the two algorithms complement and combine each other to effectively solve the problem of premature convergence.

[0064] (3) Design an obstacle avoidance algorithm for intelligent robots in complex environments and conduct simulation experiments.

[0065] By applying the optimized algorithm to robot path planning, a hybrid algorithm based on genetic algorithm and simulated annealing algorithm is used for obstacle avoidance. At the same time, the grid method is selected to model the environment, and the obstacle information is transmitted in combination with the robot perception system, so as to design an intelligent robot obstacle avoidance algorithm in a complex static obstacle environment. The working principle of the algorithm is as follows: First, the robot uses its perception system to obtain environmental information, and then converts the information into subdivided grids with different costs to construct a grid map of the environment. Then, the optimized hybrid algorithm is introduced into the planning system, and the efficient global exploration ability of the genetic algorithm and the local exploration ability of the simulated annealing algorithm are used to find the optimal path of the robot. Finally, the robot follows the path and continuously perceives the surrounding environment for timely adjustment and replanning. This technology can not only improve the navigation performance and planning efficiency of the robot, but also expand its application scope, so that the intelligent robot can better adapt to various real-life scenarios.

[0066] Introducing the traditional genetic algorithm into the simulated annealing idea can make the optimal path planning faster and easier to obtain the global optimal solution. As the scale of the environment expands and the complexity of the environment increases geometrically, we should integrate the genetic simulated annealing algorithm with the improved artificial potential field method to complement each other, thereby effectively reducing the probability of the mobile robot falling into the local minimum point and allowing the robot to reach the target point more safely.

[0067] In the obstacle avoidance algorithm designed by the present invention, the grid method can make the intelligent mobile robot understand the complexity of the environment in advance, so as to further think and analyze. The genetic algorithm is the important core of the present invention. It is equivalent to the "brain thinking" in the robot, which can make the mobile robot intelligent and quickly plan the optimal path after understanding the environmental information. The genetic algorithm has the advantages of being able to find the optimal solution in a large-scale search space and maintaining strong robustness, strong adaptability and high efficiency. This is very suitable for the path planning of intelligent mobile robots in complex environments. While the genetic algorithm gives the mobile robot the ability to perform theoretical analysis, it should also consider the problems that arise during the actual movement process, and the improved artificial potential field method can do this. The artificial potential field method allows the robot to move in a virtual force field, so that the robot can effectively avoid encountering obstacles. The organic combination of the two enables the robot to have the ability of "theoretical understanding" and "practical activities" like humans at the same time, so that the robot can reach the target point more quickly and safely.

[0068] like Figure 1 and Figure 2 As shown, the method specifically includes the following parts:

[0069] (1) Grid-based environmental modeling

[0070] The present invention mainly studies the robot path planning problem in a static environment, that is, it is assumed that the obstacles in the robot working environment are all in a static state. Under this premise, the present invention chooses to use the grid modeling method to construct the environment model.

[0071] Grid modeling is a commonly used environmental modeling method. Its working principle is to replace the binary "0" and "1" of the environment to be modeled. This method can represent the environmental space as a grid in a discretized way, and describe each grid (position) through state coding, thereby effectively communicating the environment with the robot. In addition, compared with other modeling methods such as geometric method and topological method, the grid method is more intuitive and easy to operate, and is more suitable for applications such as efficient simulation and path planning in complex scenes.

[0072] (2) Global path planning using genetic simulated annealing algorithm

[0073] 1. Encoding method

[0074] The present invention adopts a grid number encoding method to encode the path information of the robot. Specifically, in the grid map, each small grid is numbered according to the row priority order (from top to bottom, from left to right), and is used as a representative of the coordinate information. For example, taking the grid map of the present embodiment as an example, the starting point of the robot is numbered 0, and the end point is numbered 99. At some obstacles in the middle, there will also be corresponding numbers to mark the position information.

[0075] By using the grid number encoding method, the environmental coordinate information can be expressed quickly and easily, and a digital matrix can be used to represent the state of the entire grid map, avoiding the complexity and burden of using a spatial coordinate system.

[0076] 2. Population initialization

[0077] In the robot path planning problem, population initialization is one of the keys to the optimization algorithm. Since the obstacle information in the robot's mobile environment is known, and after the environment is modeled by rasterization, the obstacle black grid and the passable white grid are easy to distinguish, so some simple and effective methods can be used when initializing the population. Specifically, between the starting point and the end point of the mobile robot, a series of free grids marked as 0 can be randomly selected as path nodes, and connected to the starting point and the end point to form an obstacle-free path. In this way, some initial populations can be established to provide a seed set for subsequent operations of the optimization algorithm.

[0078] At the same time, in the process of constructing the initial population, it is necessary to ensure that the generated path meets all constraints. For example, there cannot be any obstacles or repeated nodes on the path, and the grid corresponding to each node must be passable. By setting strict constraints, we can ensure the quality of the population and lay a good foundation for subsequent evolutionary optimization operations.

[0079] 3. Fitness function

[0080] In the robot path planning problem, the fitness function of the optimization algorithm plays a key role. It can not only evaluate the quality of individuals in the current population and guide the next evolutionary selection operation, but also directly determine the quality of the solution. Generally speaking, the fitness function is the optimization objective function of the model, which needs to be designed and constructed according to the specific problem scenario. In the present invention, path distance and path smoothness are used as evaluation indicators for robot path planning, and their weighted summation is used to obtain the fitness function, as follows:

[0081] fit=ω1fit1+ω2fit2

[0082]

[0083]

[0084]

[0085]

[0086] In the formula, ω1 and ω2 are weighting factors of fitness fit1 and fit2 respectively. Fitness function fit1 is the reciprocal of the path length. The shorter the path length, the higher the fitness value. Fitness function fit2 is the angle of the path node. Three adjacent points form a triangle, and the angle of the path node can be calculated according to the cosine theorem. The cosine values ​​are calculated in sequence according to the order of the chromosome sequence. The larger the angle and the closer it is to 180°, the smoother the path and the higher the fitness value. By weighting and summing these two factors, the fitness value corresponding to each path can be evaluated, so as to compare and screen out better paths.

[0087] 4. Genetic manipulation

[0088] (1) Improve the selection operator

[0089] The present invention adopts the idea of ​​simulated annealing algorithm in the selection operation. The specific steps are as follows:

[0090] a. Adopt the elite strategy to select the individuals with the highest fitness in the initial population as elite individuals. These individuals do not participate in the subsequent selection operation.

[0091] b. Randomly select two chromosomes i and j from the remaining non-elite individuals and calculate the corresponding fitness values.

[0092] c. Based on the principle of simulated annealing algorithm, select one of chromosomes i and j to evolve to the next generation. If fit(i)>fit(j), directly select chromosome j to evolve to the next generation. Otherwise, accept chromosome j to evolve to the next generation with probability p.

[0093]

[0094] In the formula, T is the current temperature. As the population gradually evolves to the next generation, the current temperature T will gradually decay. The expression of the current temperature T is T = T0 × v g T0 represents the initial temperature, v represents the decay rate, and g represents the number of iterations.

[0095] d. Repeat steps b and c until the population size of the next generation reaches the specified value M.

[0096] Since genetic algorithms can easily select only individuals with high fitness in the population for evolution, they are prone to premature convergence. Therefore, during the selection process, some individuals with low fitness are allowed to evolve to the next generation, but at the same time, the low fitness individuals are prevented from affecting the quality of the population. Therefore, a part of elite individuals are selected first, and then the remaining individuals are annealed. As the number of iterations increases, the temperature T decreases, the probability increases, and the individuals with higher fitness will still evolve to the next generation.

[0097] (2) Improved crossover operator

[0098] After screening, a single-point crossover operation is used. Two parent chromosomes are randomly selected from the population to form a group, and the same nodes between the two chromosomes are found. Then a crossover operation is performed on the node. If there are multiple identical nodes, any one of them will be randomly selected for crossover. If a pair of parent chromosomes does not have a common gene sequence, two nodes will be randomly selected from the pair of chromosomes for crossover. If the path obtained by the crossover is not continuous and the path is incorrect, the method of initializing the population is used to make it a continuous feasible path, and the duplicate nodes contained in the daughter chromosomes are deleted.

[0099] (3) Improved mutation operator

[0100] The traditional single-point random mutation method may cause poor path quality or even infeasible path. The present invention improves the traditional mutation method, and the operation steps are as follows:

[0101] a. First, randomly select two nodes from the parent individuals as mutation points, excluding the starting point and the end point. Divide the parent chromosome into three paths.

[0102] b. Keep the path segment between the two nodes. Use the initial population generation method to generate two new paths from the starting point to node 1 and from node 2 to the target point.

[0103] c. Recombine the newly formed path 1 with the retained path and the original path 2, and delete duplicate nodes. Similarly, recombine the original path 1 with the retained path and the newly formed path 2, and delete duplicate nodes.

[0104] d. Calculate the fitness values ​​of the two newly generated chromosomes, select the one with higher fitness as the offspring, and eliminate the one with lower fitness, and the mutation operation is completed.

[0105] (2) Improved local path planning using the artificial potential field method

[0106] The improved artificial potential field method is a commonly used local path planning method. Different from the traditional artificial potential field method, the improved method constructs a potential field model that is dynamically adjusted according to the distance between the robot and the obstacle to avoid the problem of the robot being "adsorbed" by obstacles during operations such as rotation and deceleration.

[0107] In order to ensure that the robot can continue to move forward on the global path and prevent the robot from deviating from the global path during the actual movement, an adjustment coefficient is set to control the size of the robot's repulsion, and the adjustment coefficient changes as the robot moves. The specific calculation formula of the adaptive repulsion field adjustment coefficient is as follows:

[0108]

[0109] In the formula, k d is the adaptive repulsive field adjustment coefficient; d is the minimum distance from the robot's current position to the global path; L is the total length of the global path; k0 is the repulsive field constant factor. The final improved repulsive potential energy function expression is:

[0110]

[0111] Through the above formula, the adaptive repulsive field adjustment coefficient can be calculated in real time according to the current distance between the robot and the global path. When the robot deviates from the path too far, the value of d / L is relatively small, and the magnitude of the repulsive force is relatively small at this time, so as to prevent the robot from being affected by too large a repulsive force and causing unstable movement, and to avoid causing too many discretized noise parameters to fall off and cause robustness problems; when the robot deviates far from the path, the value of d / L gradually increases, and the adaptive repulsive force also gradually increases, so that the robot is corrected back to the global path.

[0112] The present invention also provides a mobile robot, which adopts the obstacle avoidance method combining the simulated annealing genetic algorithm described in any one of the above items with the improved artificial potential field method.

[0113] In summary, the present invention introduces the traditional genetic algorithm into the simulated annealing idea, which can make the optimal path planning faster and easier to obtain the global optimal solution, and integrates the genetic simulated annealing algorithm with the improved artificial potential field method. The two complement each other, thereby effectively reducing the probability of the mobile robot falling into the local minimum point and allowing the robot to reach the target point more safely.

[0114] It should be noted that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0115] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, and two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0116] It will be easily understood by those skilled in the art that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An obstacle avoidance method combining simulated annealing genetic algorithm with improved artificial potential field method, characterized in that: The method includes: Construct a grid map of the environment and determine the obstacle grid and the traversable grid; Perform global path planning based on genetic simulated annealing algorithm to obtain the global planning path; According to the global planning path, the resultant potential field of the improved repulsive potential energy function moves.

2. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 1, characterized in that: The method further includes: Detect whether it falls into a local minimum point; if so, introduce a filling potential field; if not, continue to move along the global planning path until reaching the target point.

3. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 1, characterized in that: Based on the genetic simulated annealing algorithm, global path planning is performed to obtain the global planning path, including: Number each grid and determine the starting and ending points; The initial population is constructed using a genetic algorithm. Each chromosome individual in the population represents an obstacle-free path from the starting point to the end point, corresponding to a set of grid numbers from the starting point to the end point. The idea of ​​​​simulated annealing algorithm is introduced in the selection operation of genetic algorithm, as follows: Use the fitness function to calculate the fitness value of each individual in the current population, that is, the fitness value corresponding to each path; Select individuals whose fitness values ​​are higher than the preset threshold or the top a% of the highest fitness values ​​as elite individuals; Randomly select two chromosome individuals i and j from the remaining non-elite individuals and calculate the corresponding fitness values; Based on the principle of simulated annealing algorithm, one of chromosome individuals i and j is selected; if fit(i)>fit(j), chromosome individual j is directly selected; otherwise, chromosome individual j is selected with probability p; Where T is the current temperature, which gradually decays; the expression of the current temperature T is T = T0 × v g , T0 represents the initial temperature, v represents the decay rate, g represents the number of iterations; fit(i) and fit(j) are the fitness values ​​of chromosome individuals i and j respectively; Perform crossover and mutation operations on the elite individuals and the selected individuals, and recalculate their fitness values ​​until the optimal global planning path is obtained.

4. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 3 is characterized in that: The method to construct the initial population is: Randomly select a series of traversable grids as path nodes and connect them with the starting point and the end point to form an obstacle-free path from the starting point to the end point; There cannot be any obstacles or duplicate nodes on the path, and the grid corresponding to each node must be passable.

5. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 3, characterized in that: Taking path distance and path smoothness as evaluation indicators of path planning, the fitness function is obtained by weighted summation, as follows: fit=ω1fit1+ω2fit2 Where fit is the fitness value, ω1 and ω2 are the weighting factors of fitness fit1 and fit2 respectively; i represents the number of path nodes, end represents the number of path nodes, and x and y represent the coordinates of path nodes; The shorter and smoother the path, the higher the fitness value.

6. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 4, characterized in that: Crossover includes: Randomly select 2 parent chromosomes from the population to form a group, and find the common nodes between these two chromosomes; Perform a crossover operation on the node; If there are multiple identical nodes, any one of them is randomly selected for crossover; If a pair of parent chromosomes do not have a common gene sequence, two nodes are randomly selected from the pair of chromosomes for crossover; if the path obtained by crossover is discontinuous, the method of constructing the initial population is used to make it a continuous feasible path, and the duplicate nodes contained in the daughter chromosomes are deleted.

7. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 4, characterized in that: Variations include: First, two nodes are randomly selected from the parent individuals as mutation points, excluding the starting point and the end point; the parent chromosomes are divided into three paths, namely the first path, the middle path and the tail path; The middle path is retained; the new first path and tail path are regenerated using the method of constructing the initial population; The new first path, the middle path and the original tail path are recombined, and duplicate nodes are deleted; similarly, the original first path, the middle path and the new tail path are recombined, and duplicate nodes are deleted.

8. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 1, characterized in that: The improved repulsive potential energy function expression is: Where U rep is the repulsive potential energy, k d is the adaptive repulsive field adjustment coefficient, k rep is the repulsive force scale factor, ||ww o || is the distance between the object and the obstacle, ρ0 is the influence radius of each obstacle, ||ww g || n is the distance between the object and the end point to the power of n.

9. The obstacle avoidance method combining simulated annealing genetic algorithm and improved artificial potential field method according to claim 8, characterized in that: The calculation formula of the adaptive repulsion field adjustment coefficient is as follows: In the formula, k d is the adaptive repulsive field adjustment coefficient; d is the minimum distance from the current position of the object to the global planning path; L is the total length of the global planning path; k0 is the repulsive field constant factor.

10. A mobile robot, characterized in that: The mobile robot adopts the obstacle avoidance method combining the simulated annealing genetic algorithm described in any one of claims 1 to 9 with the improved artificial potential field method.

Citation Information

Cited By

  • Multi-area photovoltaic inspection path planning method and system

    CN119180398A

  • CAV dynamic path planning method and system based on improved Dstart-Lite

    CN120521633A

  • Robot obstacle avoidance path planning method and device, electronic equipment and medium

    CN122083956A

  • Robot obstacle avoidance path planning methods, devices, electronic equipment and media

    CN122083956B