CN117434950A

CN117434950A - Mobile robot dynamic path planning method based on Harris eagle heuristic hybrid algorithm

Info

Publication number: CN117434950A
Application number: CN202311457420.8A
Authority: CN
Inventors: 袁杰; 赵瑛瑛; 闫学勤; 梁荣光; 常乾坤; 李辉; 袁昊; 段成龙; 曹学伟
Original assignee: Xinjiang University
Current assignee: Xinjiang University
Priority date: 2023-11-02
Filing date: 2023-11-02
Publication date: 2024-01-23

Abstract

The invention provides a path planning method based on a heuristic hybrid algorithm, which mainly solves the problems of unsmooth and low efficiency of a planned path. Comprising the following steps: setting the shortest distance between starting and ending points as an objective function, adding a penalty term as a fitness value, and then sequencing the fitness values of the Harris eagle population based on a Harris eagle optimization algorithm. And then selecting haris eagle individuals with larger and smaller fitness values to form a temporary haris eagle population, iteratively updating the fitness values by adopting a simplex local search strategy, and increasing the diversity of a search algorithm by adopting self-adaptive dynamic escape energy. Aiming at the problem of local minima in the path planning process, an artificial potential field method is introduced to dynamically avoid the obstacle, and when the path is in the local optimum, a normal vector guiding strategy is adopted to escape from the local minima point. Finally, through the fusion of the two algorithms, dynamic obstacle avoidance is realized while the global performance of path planning is ensured, and the smoothness of the path is realized in practical application, so that the safety performance of the mobile robot is improved.

Description

Mobile robot dynamic path planning method based on Harris eagle heuristic hybrid algorithm

Technical Field

The invention belongs to the technical field of dynamic path planning of mobile robots, and particularly relates to a dynamic path planning method of a mobile robot based on a Harris eagle heuristic hybrid algorithm in a complex environment.

Background

Mobile robots are increasingly being used in autonomous navigation, automated manufacturing, warehouse management, and unmanned applications. In these applications, dynamic path planning is a critical technique that allows robots to efficiently plan paths to reach targets based on changing environments and task demands. Path planning becomes more challenging, especially in complex environments, such as factory workshops, warehouses, or urban roads. Conventional path planning methods are typically planned based on static maps or environmental models, but in dynamic environments, these methods may not be effective in dealing with changes in obstacles and real-time decision-making requirements. Accordingly, researchers have been seeking to develop more advanced dynamic path planning methods to meet the ever-increasing demands of robotic applications.

The Harris eagle optimization algorithm is derived from hunting strategies of Harris eagles in nature. This algorithm simulates the behavior of the harris eagle during search and hunting. It combines the strategies of depth search and fine detection, making it a balance between global and local search. The harris eagle optimization algorithm is excellent in searching an optimal solution, avoiding a local minimum problem and adapting to a complex environment.

The dynamic path planning method of the mobile robot based on the Harisk heuristic hybrid algorithm represents an important progress of path planning technology, and can be applied to the fields of autonomous navigation, automatic manufacturing, logistics, unmanned driving and the like so as to improve the motion capability and adaptability of the robot in a complex environment.

Disclosure of Invention

The invention relates to a mobile robot dynamic path planning method based on a Harris eagle heuristic hybrid algorithm. The method fully utilizes the global search characteristic of the Harris eagle optimization algorithm and the capability of fine detection, so that the robot can conduct path planning in complex and dynamic environments.

The technical scheme of the invention is as follows: a mobile robot dynamic path planning method based on Harris eagle heuristic hybrid algorithm. The method is characterized by comprising the following steps of:

(1) The harris eagle population is randomly distributed in the search space, each harris eagle represents an independent candidate solution, namely, the harris eagle is regarded as a possible path node, and the more optimal solution is determined by continuously and iteratively updating the size of the adaptive value sequence to order.

And selecting the shortest path with the smallest fitness value, namely the shortest distance, as the planned candidate path according to the searched fitness value sorting information.

In the planned path, the objective function is set as the shortest distance between the starting point and the ending point, the obstacle cannot collide in the search space, and the minimum distance d between the current node and the obstacle _min And taking > 0 as a constraint condition, adding an objective function of the constraint condition as an fitness function of the Harris eagle population, and carrying out iterative updating on the fitness value of the whole Harris eagle population.

When a candidate solution that satisfies the objective function occurs, the position of the harris eagle individual is converted to a node on the path to form the path. First, the optimal solution of the problem is converted into a set of numerical coordinates, each corresponding to the location of a node. The set of digital coordinates is then mapped to an actual coordinate space to generate an actual path node. These path nodes are processed in a normalized manner. Finally, a path between the start point and the end point is formed by linear interpolation.

However, in the process, due to the uniform distribution of the harris eagles in the initial search space, the early convergence speed of the algorithm is low, the algorithm is easy to fall into local optimum, according to the problem, the randomness of the adaptive dynamic escape energy factor to the algorithm is increased, more effective search is realized, and meanwhile, the convergence speed of the algorithm is improved by introducing a simplex local search strategy and the local optimum is escaped.

(2) The introduction of dynamic adaptive escape energy. Escape energy E is a conversion index from global exploration to local development and is also an index for measuring the optimization capacity of the HHO algorithm. However, as the number of iterations increases, the algorithm tends to search locally in advance.

By introducing random numbers with normal distribution and random numbers ranging between [0,1], the method is respectively suitable for optimizing tasks of different stages. I.e., earlier in the algorithm, is more focused on global searches and later in the algorithm is more prone to local searches.

The design of simplex local search strategy. The random search of the early Harris eagle population diffuseness reduces the convergence rate of the algorithm, and as the iteration number increases, the Harris eagle optimization algorithm attracts other Harris eagles with lower fitness values, reduces the randomness of the algorithm and falls into local optimum.

Aiming at the problems, the simplex is designed to approximately solve the linear path planning optimization problem. The planned candidate paths are discretized firstly, the continuous path planning problem is converted into a discrete linear problem between every two feasible solutions, and in the linear problem, a more optimized solution is found by using a simple local search strategy scientific research.

And selecting the temporary sub-population which is determined to be the simplex local search strategy from the Harris eagle population by the first 10% and the last 10% according to the size of the fitness value sequence. The temporary sub-population is a basic feasible solution to the optimization problem. The simplex local search strategy can replace worse candidate solutions and be randomly executed, and searching between the global and local is better balanced.

By adapting the maximum point, i.e. the worst point X, of the fitness value of the temporary sub-population _w Four operation modes of internal contraction, external contraction, reflection and expansion are performed to iteratively replace the worst point X _w And (5) completing iterative updating of the simplex local search.

Figure 1 shows the geometry of all the replacement points in one iteration. First, during each simplex local search, the worst points are replaced in various ways, resulting in a new geometry for each iteration. Second, a new worst case point will appear in the new geometry. The process continues.

(3) Aiming at complex environments, the Harris eagle optimization algorithm can only aim at global search, but has the defect of real-time obstacle avoidance. The APF algorithm is used to overcome the limitation and realize dynamic obstacle avoidance, but the traditional APF has the problem of being easy to fall into local optimum. Therefore, for the problem of local optimization in the APF method, normal vector guidance is introduced, and an endpoint stabilization strategy is set near the endpoint to avoid meaningless oscillations.

The degree of oscillation between the end points of the mobile robot and the target is determined by comparing the distance and the step size between the end points and the mobile robot and the target point. And judging whether the mobile robot directly reaches the end point through comparing the distance with the step length.

Aiming at the problem that the target is unreachable in the dynamic environment caused by the algorithm falling into the local optimum, the normal vector is introduced to guide the mobile robot to change the direction of the resultant force, and the mobile robot escapes from the local minimum value. Fig. 2 shows a schematic view of a mobile robot free from local oscillation escape.

(4) The Harisk algorithm and the artificial potential field method are fused, whether the artificial potential field method is adopted for dynamic obstacle avoidance is judged according to whether an obstacle suddenly appears or not, and when an unknown obstacle appears in a search space and moves in an unknown direction and uncertain speed mode, the artificial potential field method is called to realize dynamic obstacle avoidance in a complex environment.

The invention provides a mobile robot dynamic path planning method based on a Harisk heuristic hybrid algorithm, which fully exerts the global searching characteristic and the fine detection capability of the Harisk algorithm and enables the robot to carry out path planning in complex and dynamic environments. By introducing dynamic self-adaptive escape energy and simplex local search strategy, the performance of the algorithm is improved, the convergence speed is increased, and the problem of trapping in local optimum is avoided. The method also combines the artificial potential field method, so that the robot can realize dynamic obstacle avoidance when facing the suddenly appearing unknown obstacle, and the safety and reliability of path planning are ensured. By means of normal vector guidance, the trouble of oscillation and local minima is avoided, and the robot is ensured to effectively reach the target point. In a word, the innovative path planning method integrates various strategies, provides a smoother and reliable solution for navigation of the mobile robot in a complex environment, and contributes to important progress in the development of intelligent robot technology.

Drawings

FIG. 1 is a frame diagram of a Harisk heuristic hybrid algorithm

FIG. 2 simplex local search geometry strategy diagram

FIG. 3 schematic diagram of endpoint oscillation

FIG. 4 schematic of normal vector steering

FIG. 5 schematic diagram of a mobile robot escape local oscillation

FIG. 6 flow chart of Harisk heuristic hybrid algorithm

Detailed Description

The invention will be described in further detail with reference to the accompanying drawings and examples, in which: the specific embodiments described herein are to be considered in an illustrative sense only and are not intended to limit the invention. Furthermore, for convenience of description, only some, but not all, structures related to the present invention are shown in the drawings.

The invention considers the inherent challenges of dynamic obstacle avoidance in path planning, and is important to keep balance between adapting to complex environments and maintaining algorithm performance. Therefore, HHO is combined with simplex local search strategy and normal vector guidance APF, and a hybrid dynamic path planning algorithm is provided. The algorithm ensures the smoothness of the planned path while effectively avoiding dynamic obstacles, and fig. 1 shows the framework of the hybrid algorithm.

The invention relates to an algorithm of a mobile robot dynamic path planning method based on a Harris eagle heuristic hybrid algorithm, which is characterized by comprising the following steps:

(1) The invention relates to a method for planning a path by using a Harris eagle optimization algorithm. The HHO algorithm is a heuristic optimization algorithm, simulates hunting behavior of Harris eagle and is used for searching an optimal path planning solution. The HHO algorithm adopts a two-stage optimization method and is divided into an exploration stage and a development stage. These two phases help to find the best solution in path planning. In the exploration phase, the algorithm explores the search space extensively through global search methods, while in the development phase it focuses more on fine tuning and optimizing paths to find the best paths.

(2) Setting an initial population. The position vectors of the initial harris eagle population represent possible path nodes. Wherein the size N of the population is selected according to the needs of the particular application and the complexity of the problem. Each halis eagle in the population represents one possible path planning solution (where the initial halis eagle population N is 90 in this example). The maximum number of iterations T is determined, i.e. how many iterations the algorithm will perform to find the best path. The value of T is typically set according to the complexity and computational resources of the problem (where in this implementation, the maximum number of iterations T is 300). The initial set up of the harris eagle population is the starting point for path planning. Each halisil's position vector represents a set of possible path nodes, where each node corresponds to a candidate solution. These initial solutions will be continually optimized in subsequent iterations to find the best path planning solution.

(3) Objective function of path planning. The objective function is used to measure the quality of the path. It is based on the sum of the euclidean distances between all neighboring nodes on the path. The current position vector of the harris eagle entity is treated as a set of high-dimensional vector solutions with a fixed number of nodes m (N > m). Some poor quality candidate solutions are discarded in the iterative update. The objective function is calculated by the following equation, which includes a start point, an end point, and a path node.

Where m represents the number of nodes, p _i Representing the coordinate position of the ith node, p _s Represents the starting point, p _a Representing the target point. The position of the population is continuously adjusted by iteration of the HHO algorithm to minimize the objective function f _min . This will result in a continuously increasing quality of the path, such that the planned path gradually tends towards the optimal solution.

(4) Constraint conditions and penalty terms. The HHO algorithm finds the optimal solution by finding the shortest path and avoiding collisions with obstacles. The path planning problem imposes a minimum distance constraint to avoid collisions with obstacles. Minimum distance d _min Calculated by the following formula.

Where n is the number of obstacles, route _i Representing two-dimensional coordinates of each node on the path, obstacles _j Is a horizontal and vertical coordinate containing the center of the obstacle circle, radius _j Subtracting 2 for the radius of the obstacle means that a sufficient safety distance between the waypoint and the obstacle is ensured.

No collision with obstacles should occur during path planning using HHO algorithm. Constraint d _min >0. However, when the distance d between the path node and the obstacle _min When negative, a constraint term should be added and set to res to indicate the overall degree of violation between the path and the obstacle. In order to incorporate the constraint into the objective function, so that the HHO algorithm does not generate an invalid solution, a penalty factor P is added to balance the relationship between the objective function and the constraint variables, and the generated solution is ensured to meet constraint conditions as much as possible. The magnitude obtained by introducing the individual position vector of the harris eagle into the objective function is a fitness value as shown in the following formula. The fitness value corresponds to the quality of the path and is used to determine the quality of the candidate solution.

f _t ＝f _t-1 +P×res

Wherein f _t-1 Represents the t-1 thObjective function value of the multiple iterations.

When a candidate solution that satisfies the objective function occurs, the position of the harris eagle individual is converted to a node on the path to form the path. First, the optimal solution of the problem is converted into a set of numerical coordinates, each corresponding to the location of a node. The set of digital coordinates is then mapped to an actual coordinate space to generate an actual path node. These path nodes are handled in a normalized manner. Finally, a path between the start point and the end point is formed by linear interpolation.

The objective function is calculated in each iteration taking into account the constraint penalty term. For individuals outside the search range, the sigmoid function is used to limit them to a particular search range.

(5) Search of harris eagles in the search space. According to escape energy E and random number r, the search behavior of Harris eagle individuals is divided into four modes: soft-enclosed, hard-enclosed, progressive rapid-dive soft-enclosed, and progressive rapid-dive hard-enclosed.

The exploration phase is the initial phase of the harris eagle optimization algorithm, the purpose of which is to explore the entire search space by means of a global search method to find the best possible path planning solution. In the exploration phase, the position of the harris eagle is updated according to the following formula. This formula below illustrates the location update process:

wherein t represents the current iteration number, X _i (t) represents the position of the ith Harriscrat eagle during the t-th iteration, X _rand (t) represents an individual position randomly selected from the population during the t-th iteration, r _i (i=1, 2,3, 4) are all [0,1]The random number in between, the optimal position is considered the position of the hunting object. r is (r) ₁ ,r ₂ ,r ₃ And r ₄ Is four independent random numbers following normal distribution, the values of which are 0,1]Is a kind of medium. X is X ^t _prey Representing the optimal individual position in the population in the t-th iteration process; u (U)B and LB represent the upper and lower limits of the search space. X is X _m (t) represents the average position of the harris eagle at the t-th iteration, as shown in the following formula.

Where N represents the total number of hawks, i.e., the mean vector of all candidate solution vectors is calculated.

The transition to the development phase is explored. The hunting is energy-consuming gradually in the escape process, the Harris eagle optimization algorithm designs the escape energy to simulate the conversion between different development behaviors, and the escape energy of the hunting is modeled as:

E ₀ the initial energy of the prey during running is randomly valued in the (-1, 1) range in each iteration process; t is the maximum iteration number, dynamic escape energy E is in a descending trend along with the iteration number, when escape energy |E| is more than or equal to 1, the HHO algorithm is in an exploration stage, when |E| is less than 1, the HHO algorithm is developed, and a better solution is searched in the field of the current solution.

In the development stage, after the hunting is found in the exploration stage, the HHO algorithm presents four different modes to simulate the development stage according to the escape behavior of the hunting and the chasing strategy of the harris eagle. Defining a random number r ₅ According to the defined random number size and escape energy, 4 position updating modes, namely soft wrapping, hard wrapping, asymptotic fast dive soft wrapping and asymptotic fast dive hard wrapping, are designed.

Soft surrounding: when |E| is more than or equal to 0.5 and r ₅ And when the energy is more than or equal to 0.5, the prey has enough energy to escape, and the Harris eagle optimization algorithm updates the position by adopting the following formula:

X(t+1)＝ΔX(t)-E|JX _rabbit (t)-X(t)|

ΔX(t)＝X _rabbit (t)-X(t)

wherein J represents the whole escape of the prey, i.e. the rabbitJump intensity in running process, the value of J changes randomly in each iteration process; Δx ^t Is the difference between the rabbit's position vector and the current position.

Hard surrounding: when |E| < 0.5 and r ₅ More than or equal to 0.5, the escape energy is lower, the prey is exhausted, and the position is updated by adopting a hard surrounding strategy:

X(t+1)＝X _rabbit (t)-E|ΔX(t)|

asymptotic rapid dive soft surrounding: when E is more than or equal to 0.5 and less than 1 and r ₅ <0.5, the prey had sufficient escape energy but did not escape the enclosure, and the next step was evaluated using the Levy flight strategy using the following equation:

Y＝X _rabbit (t)-E|JX _rabbit (t)-X(t)|

if the fitness value is not improved at this time, executing a flight strategy:

Z＝Y+S×LF(D)

where D is the spatial dimension, S is a 1 XD random vector, and LF is the Levy flight function.

Wherein u and v are random numbers uniformly distributed in [0,1], and β=1.5, so the secondary tapping mode can be summarized as follows:

asymptotically rapid dive hard surrounds: when r is less than 0.5 and |E| < 0.5, the hunting objects have the opportunity to escape, but escape energy is insufficient, the hunting objects do not escape from the surrounding rings, the Harris hawk reduces the surrounding rings among the hunting objects by continuously reducing the average position, and the mathematical model is as follows:

wherein y=x _rabbit (t)-E|JX _rabbit (t)-X(t)|,Z＝Y+S×LF(D)

These modes allow the algorithm to be more focused in the development phase, searching for domains with better solutions, thus improving the quality of path planning, and by combining the exploration and development phases and the variation of escape energy and random numbers, the algorithm can find high-quality solutions in path planning, avoiding sinking into locally optimal solutions. The combination of these policies and behaviors facilitates global searching, ensuring that the generated paths are optimal.

(6) Escape energy E is a conversion index from global exploration to local development and is also an index for measuring the optimization capacity of the HHO algorithm. However, as the number of iterations increases, the algorithm tends to search locally in advance. Based on this, in HHO algorithms, escape factors are introduced into the dynamic adaptive escape energies α and β, aimed at balancing global exploration and local development. This enhancement helps to increase the efficiency of the algorithm, enabling more efficient random exploration. The factors α and β related to the number of iterations are shown in the following equation.

Wherein r is _i Is a normal random number, is consistent with biological habit, r _j Is [0,1]]The random number accords with the characteristics of natural and social phenomena, and the diversity of the algorithm is increased.

The escape energy E is dynamically adjusted based on the current iteration count. This adjustment allows the algorithm to adapt to different optimization tasks by taking into account the complexity and convergence status of the problem. In the initial phase, the current iteration number t is small, (1-t/t) is a large factor, which will keep the dynamic escape energy E at a high value. Higher escape energy helps to explore the solution space more widely. As the iteration progresses, the current iteration number t becomes progressively larger, the value of the factor (1-t/t) decreases, and the dynamic escape energy E decreases progressively, allowing a more focused search in the local solution space, enhancing convergence to the local optimum.

(7) In the HHO algorithm, the Harris hawk performs non-target random search at an early stage, so that the convergence rate of the algorithm is reduced. With the increase of the iteration times, the algorithm attracts other harris hawks with lower fitness values, reduces the randomness of the algorithm and falls into local optimum.

The invention introduces a simplex local search strategy based on a simplex idea. Simplex is commonly used in mathematical algorithms to solve the linear programming problem. Thus, the continuous path is discretized and the continuous problem is translated into a discrete linear problem between every two possible solutions. In the linear problem, simplex is used to optimize local search strategies to find more optimal solutions.

The first 10% and the last 10% of fitness ordering in the initial harris eagle population are determined as temporary sub-populations of simplex local search strategy according to HHO algorithm. The temporary sub-population is a basic feasible solution to the optimization problem. The simplex local search strategy can replace worse candidate solutions and be randomly executed, and searching between the global and local is better balanced.

First, if the random number c is less than 0.5, the simplex local search strategy is executed, otherwise it is not executed. Then, the point with the highest fitness value is selected as the worst point X _w The point with the lowest fitness value is selected as the best point X _g Having suboptimal fitness value selected as suboptimal advantage X _s And is based on point X _g And point X _s To determine the center point X _c As shown in the following formula.

Optimum point X _g Secondary advantage X _s And the worst point X _w An initial simplex geometry is formed. As shown in fig. 2. FIG. 2 shows a schematic process diagram of a simplex local search strategy in two-dimensional space. Worst point X _w Is replaced by internal contraction, external contraction, reflection and expansion.

Point X _w 、X _g And X _s The objective function values of (2) are denoted as f respectively _w 、f _g And f _s . There are several ways to replace the worst point X _w . To the central point X _c Performing a reflection operation to obtain a reflection point X by _r ：

X _r ＝X _c +ε(X _c -X _w )

If f _r <f _g Meaning that the search direction is close to the optimal solution, the expansion operation is performed. Expansion point X _e Obtained by the formula:

X _e ＝X _c +μ(X _r -X _c )

if fe is<fg, then replace the worst point X with the point Xe _w As shown in FIG. 2 (b), otherwise, point X is used _r To replace worst point X _w 。

If f _r >f _g Meaning that the search direction is far from the optimal solution, then an internal contraction is performed. Internal contraction point X _a Obtained by equation (14):

X _a ＝X _c +μ(X _w -X _c )

if f _a <f _w Then use point X _a Replace worst point X _w As shown in fig. 2 (d).

If f _w >f _r >f _g Then an external contraction is performed. External contraction point X _b Obtained by the formula:

X _b ＝X _c -μ(X _w -X _c )

if f _b <f _w The worst point X _w Is marked by X _b Alternatively, as shown in FIG. 2 (e), otherwise, the worst point X _w Is marked by X _r The substitutions are shown in Table 2 (c).

In the above equation, ε is the reflection factor and μ is the scale factor. The simplex local search strategy iteratively performs the reflection, expansion, constraint, and substitution operations. The convergence speed and performance of the algorithm in complex problems are improved.

Figure 2 shows the geometry of all the replacement points in one iteration. First, during each simplex local search, the worst points are replaced in various ways, resulting in a new geometry for each iteration. Second, a new worst case point will appear in the new geometry. The process continues by replacing the new worst pixel as shown in fig. 2. The iterative process is continued until the maximum iterative limit of the simplex local search strategy is reached, so that the simplex local search strategy can be continuously close to the minimum objective function. Finally, the recalculated fitness values of the updated temporary sub-populations are combined into fitness values of the harris-hopk populations, and they are reordered.

The input of the simplex local search strategy is the fitness value of the temporary sub-population, and the output is the candidate solution updated through the simplex local search strategy iteration. The strategy replaces worse candidate solutions by comparing their fitness values. The candidate solution obtained by the simplex local search strategy ensures that it is not worse than the worst solution, and the resulting solution typically brings the candidate solution closer to the optimal solution. The complex local search strategy is added in the HHO algorithm, so that the path is smoother, the inflection points are fewer, and the exploration capability of the algorithm is enhanced.

(8) And the artificial potential field method is used for dynamically avoiding barriers. The artificial potential field method (Artificial Potential Field Method) is a technique commonly used in robotic control and path planning. The robot simulation system simulates an interaction force field between objects to help the robot plan a motion path, avoid collision and realize specific task targets. The core idea of this method is to guide the robot to the target location while avoiding obstacles, as if the object were moving in a potential energy field.

The following is the basic principle and components of the artificial potential field method:

target potential energy: potential energy representing movement of the robot toward the target location. This potential energy is an attractive field that attracts the robot to the target location. Typically, the target potential energy assumes a minimum value at the target location. Obstacle potential energy: representing the potential for the obstacle to repel the robot. The obstacle potential energy is a repulsive field that prevents the robot from approaching the obstacle. The magnitude of the potential energy is inversely proportional to the distance of the obstacle, and the closer the potential energy is to the obstacle, the greater the potential energy is. Total potential energy field: and superposing the target potential energy and the barrier potential energy together to obtain a total potential energy field. The robot moves in the total potential energy field and moves the gradient of the potential energy field according to the current position. And (3) movement control: the robot decides the moving direction and speed according to the gradient information of the total potential energy field. Typically, the robot moves in a gradient descent direction to reduce the total potential energy until the target position is reached or it is unable to proceed further (e.g., because it is blocked by an obstacle).

(9) Endpoint stabilization strategy. Near the end point, the mobile robot may experience oscillations caused by the attractive force of the target point and the repulsive force of the obstacle. As shown in fig. 3.

Fig. 3 shows a schematic diagram of the path around the front and rear end points of oscillation and set positioning accuracy. It can be seen that after the end point is reached, a redundant path is created due to the repulsive force of the obstacle. A path without positioning accuracy will cross the target point.

The degree of oscillation between the end points of the robot and the target is determined by comparing the distance and the step size between the end points and the mobile robot and the target point. A value slightly greater than 0 is set called eps, which is present to ensure that the distance between the mobile robot and the end point is not less than zero. If the distance between the robot and the target is smaller than the step size but larger than eps, it is indicated that the robot is very close or has reached the vicinity of the target, the position of the robot is updated directly to the target point. Then, the target point is added to the route list to record the movement locus of the robot.

(10) Normal vector steering strategy. Aiming at the problem that tar cannot reach due to the fact that an APF algorithm is prone to being trapped in local optimization, normal vector guiding is introduced to avoid oscillation. Guided by the normal vector, the mobile robot can more effectively adjust its navigation path, thereby avoiding getting stuck or colliding with obstacles. A schematic of the normal vectors and angles introduced is shown in fig. 4.

In the APF algorithm, the angular difference between the repulsive force and the repulsive force vector is represented by θ, and decreases as the directions of the attractive force and the repulsive force are closer. The magnitude of the angle θ is shown in the following equation.

Wherein F is _att And F _rep Respectively, as attractive and repulsive forces.

The normal vector guidance steps are as follows: the resultant force is first calculated and then the inner product and the modulo length between the two vectors are determined. Next, the angle is converted using an inverse trigonometric function, and the angle difference is calculated. Finally, it is determined whether the angle difference is less than θ. If so, normal vector steering is performed to adjust the direction. At this time, both components of the resultant force are rotated to obtain a new movement direction of the resultant force of the mobile robot. If the included angle difference is larger than or equal to theta, the normal vector guiding operation is skipped, and the resultant force directly guides the movement of the mobile robot. A schematic diagram of the mobile robot getting rid of the oscillations is shown in fig. 5.

(11) A mobile robot dynamic path planning algorithm method based on a Harris eagle heuristic hybrid algorithm. The hybrid algorithm combines the global characteristic of IHHO and the real-time obstacle avoidance advantage of IAPF method, and finally realizes more efficient dynamic obstacle avoidance path planning. The algorithm is divided into two steps: first, an unknown obstacle is detected using IHHO algorithm. Subsequently, the IAPF algorithm dynamically adjusts the position of the robot to avoid potential collisions with obstacles during the search. If no collision is detected, the algorithm selects the nearest unexplored path node as the next target and continuously explores it as the new current position. If an obstacle is detected and the angle between attraction and repulsion is less than θ, a normal vector guided APF is used to avoid the obstacle. These steps are repeated until a maximum iteration of fitness values or optimal solution is reached. A flow chart of the HHI blending algorithm is shown in fig. 6.

The HHI hybrid algorithm takes advantage of IHHO and IAPF, improves path smoothness in global searches, and solves the unreachable problem by escaping local oscillations in local searches. The adaptability of the algorithm in the dynamic environment is enhanced.

The invention relates to a path planning algorithm method research of a mobile robot in a complex environment, and provides a mobile robot dynamic path planning method based on a Harris eagle heuristic hybrid algorithm, which fully exerts global searching characteristics and fine detection capability of the Harris eagle algorithm and enables the robot to perform path planning in the complex and dynamic environment. By introducing dynamic self-adaptive escape energy and simplex local search strategy, the performance of the algorithm is improved, the convergence speed is increased, and the problem of trapping in local optimum is avoided. The method also combines APF, so that the robot can realize dynamic obstacle avoidance when facing the suddenly appearing unknown obstacle, and the safety and reliability of path planning are ensured. By means of normal vector guidance, the trouble of oscillation and local minima is avoided, and the robot is ensured to effectively reach the target point. In a word, the innovative path planning method integrates various strategies, provides a smoother and reliable solution for navigation of the mobile robot in a complex environment, and contributes to important progress in the development of intelligent robot technology.

The above examples are only for illustrating the technical solution of the present application, and are not limiting.

Claims

1. A mobile robot dynamic path planning method based on a Harris eagle heuristic hybrid algorithm is characterized by comprising the following steps:

(1) A population of harris eagles is randomly distributed within the search space, wherein each harris eagle represents an independent candidate solution, each candidate solution representing a possible path node. These harris eagles update the adaptation values through successive iterations and determine a more optimal solution through the size of the ordering. And designing a single-objective optimization problem under a constraint condition based on a Harris eagle optimization algorithm, so as to optimize the planned path.

(2) And selecting a path with the minimum fitness value, namely the shortest path, as a planned candidate path based on the searched fitness value sorting information. And updating dynamic self-adaptive escape energy and updating partial Harris eagle populations by using a simplex local search strategy for the Harris eagle populations at different stages.

(3) In the planned candidate path, the objective function is set as the shortest distance between the start point and the end point, while the constraint must be satisfied: the path cannot intersect the obstacle and the minimum distance dmin of the current node from the obstacle must be greater than 0. These constraints are added to the objective function as a fitness function of the harris eagle population to iteratively update the fitness value of the entire harris eagle population.

(4) When a candidate solution that satisfies the objective function occurs, the position of the harris eagle individual is converted to a node on the path to form the path. First, the optimal solution to the problem is converted into a set of numerical coordinates, where each coordinate corresponds to the location of a path node. These digital coordinates are then mapped to the actual coordinate space to generate the actual path nodes. These path nodes are processed in a normalized manner and finally a path between the start point and the end point is generated by linear interpolation.

2. According to the mobile robot dynamic path planning method, the self-adaptive dynamic escape energy factors are introduced according to the requirements, so that the randomness of the algorithm is improved. At the early stages of the algorithm, global searches are more focused, while at the later stages local searches are more favored.

3. The method introduces a simplex local search strategy for approximately solving the linear path planning optimization problem. First, the planned candidate paths are discretized, converting the continuous path planning problem into a discrete linear problem. Then, by simplex local search strategy, a better solution is found between these discretized solutions.

4. According to the size of the fitness value sequence, the method selects the harris eagles of the first 10% and the last 10% from harris eagle populations, and determines the harris eagles as temporary sub-populations of the simplex local search strategy. This temporary sub-population contains a basic feasible solution to the optimization problem. Simplex local search strategies may replace poor candidate solutions and be performed randomly to better balance the trade-off between global and local searches.

5. The simplex local search strategy is implemented by applying the simplex local search strategy to the highest point (worst point X _w ) Performing four operation modes of internal contraction, external contraction, reflection and expansion to iteratively replace the worst point X _w To complete iterative updating of the simplex local search.

6. Aiming at the problem of dynamic obstacle avoidance in a complex environment, an artificial potential field method (APF) is introduced. When unknown obstacles appear in the search space and the obstacles move in an unknown direction and at an uncertain speed, an artificial potential field method is called to realize dynamic obstacle avoidance. This ensures the safety and reliability of the path planning of the robot when facing unknown obstacles.

7. To avoid the problems of oscillations and local minima, a normal vector is introduced to guide the mobile robot to change the direction of the resultant force. This ensures that the robot can effectively avoid unwanted oscillations, reaching the target point more quickly.

8. The mobile robot dynamic path planning method described in the claim, wherein the method provides a smooth, efficient and reliable path planning solution under complex and dynamic environments through the combination of a harris eagle heuristic hybrid algorithm, a dynamic adaptive escape energy factor, a simplex local search strategy, normal vector guidance and an artificial potential field method. This innovative approach ensures that the robot can reach the target point efficiently and safely avoid obstacles in various challenging scenarios.