Path planning is a very important and critical research topic in mobile robot navigation planning.
Mobile robots always work in dynamic, complex and uncontrollable environments, such as inspection of nuclear power plants and inspection of substations. However, complex and dynamic working environments require mobile robots to be able to adjust their movement routes to complete tasks autonomously. One of the most critical factors that enables robots to move autonomously is path planning.
Mobile robot path planning technology means that the robot plans the optimal path from the starting point to the target point without collision in the operating environment under the premise of following some optimization indicators (such as the shortest time, optimal distance and lowest energy consumption, etc.).
Currently, the three major problems in path planning for mobile robots are environment modeling, algorithm convergence speed, and the tendency to fall into local optimal solutions. Environmental modeling can be solved through 2D grid maps; while convergence speed and avoiding local optimality involve the design of optimization algorithms or the integration of other algorithms.
Path planning methods can be divided into the following two types:
Classic methods (such as A*, Dijkstra algorithm), this type of method solves faster, but if there are multiple minimum values, it cannot guarantee the optimal solution.
Metaheuristic methods (i.e., relying on various new optimization algorithms) are simple and flexible, can jump out of local optima, and are favored by many scholars.
The author of this article will solve the robot path planning through classic and meta-heuristic algorithms, and introduce the author's original improved algorithm for comparative analysis.
00 Article Directory
1 Environment modeling
2 Optimization algorithm
3 Experimental results
4 Source code acquisition
01 Environment Modeling
This article sets the environment of the mobile robot as a two-dimensional plane space and uses the grid method to establish an environment model. The raster method is a classic method in map modeling, which can express environmental spatial information intuitively and the information data is easy to store. In a raster map, the location and size of obstacles are known, and the size and location of obstacles do not change during the movement of the robot.
The figure below shows two 2D raster environments
In the grid map, which represents simple and complex working environments respectively, the obstacle grid is a dark area, the feasible grid is a light area, and the 8 directly adjacent grids of each grid are directly passable grids. , the robot can select the next forward grid among the feasible grids according to certain rules, and the distance between adjacent grids is represented by the Euclidean distance between grid midpoints.
[Picture: Zhang Jun, Zhang Ting, Yu Shikun. Improvement of global path planning using ant colony algorithm [J]. Journal of Xinxiang University, 2023, 40(06): 25-28+51.]
Assuming that the robot reaches the target point through a total of M steps from the starting point, the objective function is the collision-free path length L of the mobile robot, and its length is the shortest:
In the formula, d(m) is the Euclidean distance.
02 Optimization algorithm
In this article, the author adopts
Classic algorithm: A*
Heuristic algorithms: WOA, PSO, GWO and the author's original improved algorithm AAMCWOA.
The specific principle will not be elaborated here.
03 Experimental results
In a simple raster map, the results of running the program are as follows:
Its path length and running time are as follows:
In a complex raster map, the results of running the program are as follows:
Its path length and running time are as follows:
It can be seen that path planning through heuristic algorithms can effectively jump out of the local optimum compared with classic methods, but its computational complexity is larger.
04 Source code acquisition
1. The classic A* algorithm is applied to grid path planning to obtain:
See GZH: KAU’s Cloud Experiment Bench
Its files are as follows:
2.A*, GWO, WOA, and PSO algorithms are applied to raster (simple/complex) path planning to obtain:
See GZH: KAU’s Cloud Experiment Bench
Supports one-click running of the main program, and can generate excel tables of each algorithm's running results (path length, running time)
3.A*, GWO, WOA, PSO, and AAMCWOA algorithms are applied to grid (simple/complex) path planning to obtain:
See GZH: KAU’s Cloud Experiment Bench
Supports one-click running of the main program, and can generate excel tables of each algorithm's running results (path length, running time)
Another note: If anyone has optimization problems to be solved (in any field), you can send them to me, and I will selectively update articles that use optimization algorithms to solve these problems.
If this article is helpful or inspiring to you, you can click Like/Reading (ง•̀_•́)ง in the lower right corner (you don’t have to click)