CN115097814A - Path planning method, system and application of mobile robot based on improved PSO algorithm - Google Patents

Abstract

The invention belongs to the field of mobile robot path planning, and discloses a mobile robot path planning method, system and application based on an improved PSO algorithm, and constructing a fitness function synthesizes two evaluation functions respectively considering path length and obstacle risk; In the improved PSO algorithm, a bidirectional learning strategy is introduced to expand the search range of particles and enrich the diversity of the population. In the bidirectional learning strategy, in order to overcome the problem that the optimization process cannot be well adjusted when optimizing complex functions, a dual adaptive strategy to better balance the search behavior of particles in the swarm. Then, through the attraction-repulsion strategy, the particles can be guided by the global optimal particle and the global worst particle respectively and evolve towards a better direction, which improves the local optimization performance and convergence ability of the algorithm. The invention completes the path planning method framework based on the improved PSO algorithm, and realizes the optimal path planning of the robot in a static environment.

Description

Path planning method, system and application of mobile robot based on improved PSO algorithm

technical field

The invention belongs to the field of mobile robot path planning, in particular to a mobile robot path planning method based on an improved PSO algorithm.

Background technique

With the development of science and technology, mobile robots have been widely used in different fields, such as robot rescue, robot patrol, robot monitoring, etc. Among them, path planning is one of the important skills of mobile robots. Reasonable path planning can make the robot plan the shortest and collision-free smooth path from the starting point to the target point when performing the task. Due to the complex working environment of the robot, it is difficult to find the shortest and smooth path. Therefore, how to plan the path under the optimal indicators (time, distance, energy consumption, etc.) under the premise of satisfying various constraints has important theoretical value and practical significance.

At present, the classical path planning methods can be divided into two categories.

A path planning method is a variant of some general methods, such as artificial potential field method and roadmap planning method. In the artificial potential field method, when the gravitational force and repulsion force at a certain point are equal and opposite in direction, the robot will think that it has reached the target point and stop moving or wandering in a certain area. Roadmap Planning Methods When obstacles increase, the number of connections between their vertices also increases, resulting in an increase in planning complexity and planning time.

Another path planning method is heuristic method. The main meta-heuristic methods used in robot path planning include simulated annealing, genetic algorithm, and particle swarm optimization. With heuristics, there is no guarantee that a solution will be found, but if a solution is found, it will be much faster than deterministic methods.

As an intelligent optimization algorithm, PSO algorithm has the advantages of fast search speed and easy implementation. It is widely used in solving problems such as path planning and target search. In order to prevent particle swarm optimization from falling into local optimum due to premature convergence in the search process, scholars at home and abroad have proposed many improved particle swarm optimization. However, due to the existence of adjustable parameters such as inertia weight w and learning factor c of the algorithm, if the settings are improper, the algorithm will have problems of premature convergence and slow convergence in the optimization process, resulting in a suboptimal planning path.

In order to overcome the above shortcomings, it is hoped to develop a new PSO optimization method to realize the shortest and collision-free path planning of mobile robots.

The difficulty of solving the problems existing in the above-mentioned prior art: due to the adjustability of parameters such as the population size, inertia weight, and learning factor of the PSO algorithm, how to reasonably set the value of such parameter settings, and avoid the PSO algorithm in the local optimal problem The phenomenon of premature convergence or even lack of population diversity improves the performance of the algorithm, making it difficult for the mobile robot to plan the optimal path and reduce waste of resources.

The significance of solving the problems existing in the above-mentioned prior art: a mobile robot path planning method based on the improved PSO algorithm of the present disclosure can significantly improve the optimization effect of the basic PSO algorithm, and effectively overcome the path that occurs when the basic PSO algorithm is applied to path planning. It has the advantages of low energy consumption and low wear and tear.

SUMMARY OF THE INVENTION

Aiming at the problems existing in the prior art, the present invention provides a path planning method for a mobile robot based on an improved PSO algorithm.

The present invention is implemented in this way, finding a low-energy-consumption barrier-free path for a mobile robot working in a complex environment.

A path planning method for mobile robots based on improved PSO algorithm, the fitness function constructed by this method combines two evaluation functions considering path length and obstacle risk respectively;

The two-way learning strategy is introduced into the improved PSO algorithm to expand the search range of particles and enrich the diversity of the population;

In the bidirectional learning strategy, the dual adaptive strategy is used to balance the search behavior of particles in the group; through the attraction-repulsion strategy, the particles can be guided by the global optimal particle and the global worst particle respectively and evolve towards a better direction;

Based on the path planning method framework of the improved PSO algorithm, the optimal path planning of the robot in the static environment is realized.

Further, include the following steps:

Step 1: Construct a path planning optimization model f according to the path length cost L and the robot avoiding obstacle cost Ls;

Step 2: Initialize the relevant parameters of the particle swarm algorithm;

Step 3: Evaluate the path planning optimization model f;

Step 4: Sort f from good to bad, each individual randomly selects a better individual from the group as the learning object xk, and updates the global optimal individual gbesti and the global worst individual gworsti;

Step 5: adaptively update the learning factor c, the inertia weight w and the decision factor F through the relevant parameters;

Step 6: If R<F adopts the bidirectional learning strategy, update the position xi and velocity vi of the particle; otherwise, adopt the attraction and repulsion strategy to update the particle position xi;

Step 7: Update the fitness function value f;

Step 8: Judging the number of iterations, if the number of iterations t reaches the maximum number T, output the optimal result and stop the operation; otherwise, t=t+1, and return to step 4.

Further, the specific calculation formula of the path length L is:

Among them, (xi,yi) is the path node, there are a total of n path nodes, L is the sum of the lengths of adjacent path nodes at time t, indicating the length of the path at time t;

The obstacle threat cost Ls of the mth path is defined as:

为路径到障碍物的最短距离；Among them, assuming that it is a circular obstacle, rk is the radius of the circular obstacle, k is the kth (k=1, 2,..., g) obstacle; lk is the distance from the center of the circle to each path, then

is the shortest distance from the path to the obstacle;

Calculate the mobile robot path planning optimization model f according to formula (1) and formula (2):

f=u ₁ L+u ₂ L _s (3)

where u ₁ and u ₂ are inertia factors in [0,1].

Further, in the step 2, the initialization related parameters include: population size M, particle dimension D, maximum number of iterations T, adaptive learning factor c, learning factor maximum value c _max , minimum value c _min ; inertia weight w, Inertia weight maximum value w _max , minimum value w _min ; particle initial position xi and velocity parameter vi; adaptive decision factor F.

Further, the specific formula for updating the adaptive learning factor c through the relevant parameters is:

Among them, fit is the fitness value of the current particle individual, and fit _max is the largest fitness value in the contemporary particle.

Further, the specific formula for updating the adaptive inertia weight w through the relevant parameters is:

Among them, A is the parameter that controls the curvature of the curve, t is the current iteration number, and T is the maximum iteration number.

Further, the specific formula for updating the adaptive decision factor F through the relevant parameters is:

Among them, d ₁ and d ₂ are the upper and lower limits of the decision factor, a is the parameter controlling the curvature of the curve, t is the current iteration number, and T is the maximum iteration number.

Further, the specific formula for calculating particle velocity vi and particle position xi is:

1) If R<F adopts the bidirectional learning strategy, the calculation formula of particle velocity vi and particle position xi is:

xi(t+1)=xi(t)+vi(t+1) (8)

Among them, i is the ith particle, i=1,2,...N; xk is the learning object, and r1 is a uniform random number.

2) If R≥F adopts the attraction and repulsion strategy, the calculation formula of particle velocity vi and particle position xi is:

x _i (t+1)=r ₂ x _i (t)+r ₃ (gbest _i (t)-x _i (t))-r ₃ (gworst _i (t)-x _i (t)) (9)

Among them, r ₂ is a uniform random number, r ₃ =1/2(1-r ₂ ), gbest _i global best individual, gworst _i global worst individual.

Another object of the present invention is to provide a mobile robot path planning system based on the improved PSO algorithm using the mobile robot path planning method based on the improved PSO algorithm, including: an environment modeling unit, an initialization parameter unit, a path finding unit and a maximum optimal path output unit;

The environment modeling unit uses the robot's own sensor group to collect working environment information, and constructs a path planning optimization model f according to the path length cost L and the robot avoiding obstacle cost Ls;

The initialization parameter unit is used to initialize the relevant parameters of the particle swarm algorithm;

The path finding unit adaptively updates the learning factor c, the inertia weight w and the decision factor F through the relevant parameters;

If R<F adopts the bidirectional learning strategy, update the position xi and velocity vi of the particle; otherwise, adopt the attraction and repulsion strategy, update the particle position xi; update the fitness function value f;

The optimal path output unit is used to evaluate the path planning optimization model f, sort f from good to bad, each individual randomly selects a better individual from the group as the learning object xk, and updates the global optimal individual gbesti and the global optimal individual. Poor individual gworsti.

Another object of the present invention is to provide a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, causes the processor to perform the following steps:

Use the robot's own sensor group to collect the working environment information, and construct the path planning optimization model f according to the path length cost L and the robot avoiding obstacle cost Ls;

Initialize the parameters related to particle swarm optimization;

The path finding unit adaptively updates the learning factor c, the inertia weight w and the decision factor F through the relevant parameters;

If R<F adopts the bidirectional learning strategy, update the position xi and velocity vi of the particle; otherwise, adopt the attraction and repulsion strategy, update the particle position xi; update the fitness function value f;

Evaluate the path planning optimization model f, sort f from the best to the worst, each individual randomly selects a better individual from the group as the learning object xk, and updates the global optimal individual gbesti and the global worst individual gworsti.

Another object of the present invention is to provide an information data processing terminal for implementing the above-mentioned mobile robot path planning system based on the improved PSO algorithm.

The advantages and positive effects of the present invention are:

The mobile robot path planning method based on the improved PSO algorithm of the present invention constructs a path planning objective function including the path length and the collision penalty between the mobile robot and the obstacle. A method combining two-way learning strategy and attraction-repulsion strategy is proposed, so that the improved PSO algorithm can improve the local search ability as well as the global optimization ability. In addition, a dual adaptive optimization strategy is adopted to adaptively adjust the inertia weight and learning factor in the bidirectional learning strategy, so that the global and local optimization capabilities of the algorithm can achieve a better balance. Finally, the present invention uses the improved PSO algorithm to complete the framework of the path planning of the mobile robot. The invention can not only realize the balance between the global search ability and the local search ability, but also realize the path planning and solution of the mobile robot with low consumption and no collision.

Description of drawings

Fig. 1 is the principle schematic diagram of the mobile robot path planning method based on the improved PSO algorithm;

Fig. 2 is the structural block diagram of the mobile robot path planning system based on the improved PSO algorithm;

Figure 3 Particle fitness ranking diagram;

Figure 4 is a comparison diagram of the traditional PSO algorithm and the improved PSO algorithm two-dimensional path mobile robot motion trajectory;

Figure 5 is the optimal fitness curve of the traditional PSO algorithm and the improved PSO algorithm.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

The application principle of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

Example 1:

In order to solve the problem of premature convergence and slow convergence in the optimization process due to the existence of adjustable parameters such as inertia weight w and learning factor c, if they are not set properly, the planned path will not be optimal, which will affect the mobile robot. Operation efficiency and quality in a two-dimensional environment, as shown in FIG. 1 , this embodiment provides a schematic flowchart of a path planning method for a mobile robot based on an improved PSO algorithm, including the following steps:

Step 1: Construct a path planning optimization model f according to the path length cost L and the robot avoiding obstacle cost Ls;

Step 2: Initialize the relevant parameters of the particle swarm algorithm;

Step 3: Evaluate the path planning optimization model f;

Step 4: Sort f from good to bad, each individual randomly selects a better individual from the group as the learning object xk, and updates the global optimal individual gbesti and the global worst individual gworsti;

Step 5: adaptively update the learning factor c, the inertia weight w and the decision factor F through the relevant parameters;

Step 6: If R<F adopts the bidirectional learning strategy, update the position xi and velocity vi of the particle; otherwise, adopt the attraction and repulsion strategy to update the particle position xi;

Step 7: Update the fitness function value f;

Step 8: Judging the number of iterations, if the number of iterations t reaches the maximum number T, output the optimal result and stop the operation; otherwise, t=t+1, and return to step 4.

Further, the specific calculation formula of the path length L is:

Among them, (xi,yi) is the path node, there are a total of n path nodes, L is the sum of the lengths of adjacent path nodes at time t, indicating the length of the path at time t;

The obstacle threat cost Ls of the mth path is defined as:

为路径到障碍物的最短距离；Among them, assuming that it is a circular obstacle, rk is the radius of the circular obstacle, k is the kth (k=1, 2,..., g) obstacle; lk is the distance from the center of the circle to each path, then

is the shortest distance from the path to the obstacle;

Calculate the mobile robot path planning optimization model f according to formula (1) and formula (2):

f=u ₁ L+u ₂ L _s (3)

where u ₁ and u ₂ are inertia factors in [0,1].

In the step 2, the initialization related parameters include: population size M, particle dimension D, maximum number of iterations T, adaptive learning factor c, learning factor maximum value c _max , minimum value c _min ; inertia weight w, inertia weight Maximum value w _max , minimum value w _min ; particle initial position xi and velocity parameter vi; adaptive decision factor F.

The specific formula for updating the adaptive learning factor c through the relevant parameters is:

Among them, fit is the fitness value of the current particle individual, and fit _max is the largest fitness value in the contemporary particle.

The specific formula for updating the adaptive inertia weight w through the relevant parameters is:

Among them, A is the parameter that controls the curvature of the curve, t is the current iteration number, and T is the maximum iteration number.

The specific formula for updating the adaptive decision factor F through the relevant parameters is:

Among them, d ₁ and d ₂ are the upper and lower limits of the decision factor, a is the parameter controlling the curvature of the curve, t is the current iteration number, and T is the maximum iteration number.

The specific formula for calculating particle velocity vi and particle position xi is:

1) If R<F adopts the bidirectional learning strategy, the calculation formula of particle velocity vi and particle position xi is:

xi(t+1)=xi(t)+vi(t+1) (8)

Among them, i is the ith particle, i=1,2,...N; xk is the learning object, and r1 is a uniform random number.

2) If R≥F adopts the attraction and repulsion strategy, the calculation formula of particle velocity vi and particle position xi is:

x _i (t+1)=r ₂ x _i (t)+r ₃ (gbest _i (t)-x _i (t))-r ₃ (gworst _i (t)-x _i (t)) (9)

Among them, r ₂ is a uniform random number, r ₃ =1/2(1-r ₂ ), gbest _i global best individual, gworst _i global worst individual.

As shown in FIG. 2 , an embodiment of the present invention provides a mobile robot path planning system based on the improved PSO algorithm using the mobile robot path planning method based on the improved PSO algorithm, including: an environment modeling unit, an initialization parameter unit, and a path finding unit. unit and optimal path output unit;

The environment modeling unit uses the robot's own sensor group to collect working environment information, and constructs a path planning optimization model f according to the path length cost L and the robot avoiding obstacle cost Ls;

The initialization parameter unit is used to initialize the relevant parameters of the particle swarm algorithm;

The path finding unit adaptively updates the learning factor c, the inertia weight w and the decision factor F through the relevant parameters;

If R<F adopts the bidirectional learning strategy, update the position xi and velocity vi of the particle; otherwise, adopt the attraction and repulsion strategy, update the particle position xi; update the fitness function value f;

The optimal path output unit is used to evaluate the path planning optimization model f, sort f from good to bad, each individual randomly selects a better individual from the group as the learning object xk, and updates the global optimal individual gbesti and the global optimal individual. Poor individual gworsti.

An embodiment of the present invention provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, causes the processor to perform the following steps:

Use the robot's own sensor group to collect the working environment information, and construct the path planning optimization model f according to the path length cost L and the robot avoiding obstacle cost Ls;

Initialize the parameters related to particle swarm optimization;

The path finding unit adaptively updates the learning factor c, the inertia weight w and the decision factor F through the relevant parameters;

If R<F adopts the bidirectional learning strategy, update the position xi and velocity vi of the particle; otherwise, adopt the attraction and repulsion strategy, update the particle position xi; update the fitness function value f;

Evaluate the path planning optimization model f, sort f from the best to the worst, each individual randomly selects a better individual from the group as the learning object xk, and updates the global optimal individual gbesti and the global worst individual gworsti.

An embodiment of the present invention provides an information data processing terminal, and the information data processing terminal is used to implement the mobile robot path planning system based on the improved PSO algorithm.

Example 2:

In order to verify the feasibility and effectiveness of the mobile robot path planning method based on the improved PSO algorithm proposed in Embodiment 1, this embodiment is verified by a specific numerical simulation comparison experiment with the traditional PSO algorithm:

Specifically, it is assumed that the path planning model is established as follows: Suppose the robot avoids 5 static obstacles, and the robot moves from the starting point (0, 0) to the target point (6, 8), represented by circles with different radii, and the static obstacle position information As shown in Table 1:

Table 1 Obstacle location information and radius size

According to the mobile robot path planning model f and the above-mentioned related data, the improved PSO algorithm is used to optimize and select the optimal path motion trajectory, and the specific parameters are set as follows:

The population size M is 100, the maximum number of iterations T is 400, _cmax =2, _cmin =0.5; _wmax =0.8, _wmin =0.1A=0.4; d1 ₌ 0.4, _d2 =0.6, a=10.

It can be seen from FIG. 4 that the traditional PSO algorithm and the improved PSO algorithm designed by the present invention can all realize the path planning task of the mobile robot. Comparing with the traditional PSO algorithm, it is not difficult to see that the path planned by the improved PSO algorithm designed by the present invention enables the robot to move along the edge of the obstacle when an obstacle is detected, and the trajectory of the robot is smoother. , reducing the shaft wear and energy consumption of the robot wheel in the real situation; in addition, it can be seen from Figure 5 that the path of this method is shorter, which can achieve the purpose of reducing energy consumption.

The above descriptions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included in the protection of the present invention. within the range.

Claims (9)

1. A mobile robot path planning method based on an improved PSO algorithm is characterized by comprising the following steps:

step 1: constructing a path planning optimization model f according to the path length cost L and the obstacle avoidance cost Ls of the robot;

step 2: initializing relevant parameters of a particle swarm algorithm;

and step 3: evaluating a path planning optimization model f;

and 4, step 4: sorting f from the best to the worst, randomly selecting a better individual from the group as a learning object xk by each individual, and updating the global optimal individual gbesti and the global worst individual gwesti;

and 5: adaptively updating a learning factor c, an inertia weight w and a choice factor F through the related parameters;

step 6: if R is less than F, adopting a bidirectional learning strategy to update the position xi and the speed vi of the particle; otherwise, updating the particle position xi by adopting an attraction and repulsion strategy;

and 7: updating the fitness function value f;

and 8: judging the iteration times, if the iteration times T reach the maximum times T, outputting an optimal result, and stopping operation; otherwise, t is t +1, and the step 4 is returned.

2. The mobile robot path planning method based on the improved PSO algorithm according to claim 1, wherein the specific calculation formula of the path length L is:

wherein, (xi, yi) is a path node, n path nodes are total, and L is the sum of the lengths of the adjacent path nodes at the time t and represents the length of the path at the time t;

the barrier threat cost Ls for the mth path is defined as:

wherein, it is assumed that it is a circular obstacle, rk is the radius of the circular obstacle, and k is the k (k is 1,2, …, g) th obstacle; lk is the distance from the center of a circle to each path, then

The shortest distance from the path to the obstacle;

calculating a mobile robot path planning optimization model f according to the formula (1) and the formula (2):

f＝u ₁ L+u ₂ L _s (3)

wherein u is ₁ 、u ₂ Is [0,1 ]]An internal inertial weight factor;

in step 2, the initialized relevant parameters include: population size M, particle dimension D, maximum iteration number T, adaptive learning factor c and maximum learning factor c _max Minimum value c _min (ii) a Inertia weight w, maximum value of inertia weight w _max Minimum value w _min (ii) a The initial particle position xi and the velocity parameter vi; an adaptive choice factor F.

3. The PSO algorithm-based mobile robot path planning method according to claim 2, wherein the specific formula for updating the adaptive learning factor c through the relevant parameters is as follows:

wherein fit is the fitness value of the current particle individual, and fit is _max Is the largest fitness value among the current generation of particles.

4. The improved PSO algorithm-based mobile robot path planning method according to claim 3, wherein the specific formula for updating the adaptive inertial weight w through the relevant parameters is as follows:

wherein A is a parameter for controlling the curvature of the curve, and t is the current iteration number; and T is the maximum iteration number.

5. The improved PSO algorithm based mobile robot path planning method according to claim 2, wherein the specific formula of the adaptive decision factor F updated by the related parameters is:

wherein, d ₁ 、d ₂ Deciding the upper and lower limits of a factor, wherein a is a parameter for controlling the curvature of the curve, and t is the current iteration number; and T is the maximum iteration number.

6. The improved PSO algorithm-based mobile robot path planning method according to claim 2, characterized in that the particle velocity v is calculated _i And the particle position x _i The concrete formula of (1) is as follows:

1) if R < F adopts a bidirectional learning strategy, the calculation formula of the particle speed vi and the particle position xi is as follows:

xi(t+1)＝xi(t)+vi(t+1) (8)

wherein i is the ith particle, i is 1,2, … N; xk is a learning object, and r1 is a uniform random number.

2) If R is larger than or equal to F and an attraction and repulsion strategy is adopted, the calculation formula of the particle speed vi and the particle position xi is as follows:

x _i (t+1)＝r ₂ x _i (t)+r ₃ (gbest _i (t)-x _i (t))-r ₃ (gworst _i (t)-x _i (t)) (9)

wherein r is ₂ Is a uniform random number, r ₃ ＝1/2(1-r ₂ )，gbest _i Global optimal individual, gworst _i Global worst individual.

7. An improved PSO algorithm-based mobile robot path planning system applying the improved PSO algorithm-based mobile robot path planning method according to any one of claims 1-6, characterized in that the improved PSO algorithm-based mobile robot path planning system comprises: the system comprises an environment modeling unit, an initialization parameter unit, a path searching unit and an optimal path output unit;

the environment modeling unit acquires working environment information by using a sensor group of the robot and constructs a path planning optimization model f according to the path length cost L and the obstacle avoidance cost Ls of the robot;

the initialization parameter unit is used for initializing the related parameters of the particle swarm algorithm;

the path searching unit adaptively updates a learning factor c, an inertia weight w and a decision factor F through the relevant parameters;

if R is less than F, adopting a bidirectional learning strategy to update the position xi and the speed vi of the particle; otherwise, updating the particle position xi by adopting an attraction and repulsion strategy; updating the fitness function value f;

the optimal path output unit is used for evaluating a path planning optimization model f, sorting the f from good to bad, randomly selecting a better individual from a group as a learning object xk by each individual, and updating the global optimal individual gbesti and the global worst individual gwesti.

8. A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:

collecting working environment information by using a self-contained sensor group of the robot, and constructing a path planning optimization model f according to the path length cost L and the obstacle avoidance cost Ls of the robot;

initializing relevant parameters of a particle swarm algorithm;

the path searching unit adaptively updates a learning factor c, an inertia weight w and a choice factor F through the relevant parameters;

if R is less than F, adopting a bidirectional learning strategy to update the position xi and the speed vi of the particle; otherwise, updating the particle position xi by adopting an attraction and repulsion strategy; updating the fitness function value f;

and evaluating a path planning optimization model f, sequencing f from good to bad, randomly selecting a better individual from a group as a learning object xk by each individual, and updating the global optimal individual gbesti and the global worst individual gwesti.

9. An information data processing terminal, characterized in that the information data processing terminal is used for implementing the mobile robot path planning system based on the improved PSO algorithm according to claim 7.

Images