An Adaptive Path Planning Based on Improved Fuzzy Neural Network for Multi-robot Systems

Introduction

Intelligent mobile robot is an integrated system that can achieve functions of environments apperceiving, dynamic decision-making, motion control and path planning (Fikes, Hart, & Nilsson, 1972; Holland, 1975). It has been widely used in the field of dangerous environments, e.g., planetary exploration, military reconnaissance, medical service, earthquake rescue, deep sea and nuclear industry because of its high flexibility, intelligence and environmental adaptability. The collision detection is the important problem should be concerned in the multi-robot systems (Asama, Matsumoto, & Ishida, 1989).

Fuzzy neural network (FNN) is a local approximation network that combines the knowledge structure of fuzzy logic reasoning with the self-learning ability of neural network (Liu, 2011). The traditional neural network is provided with fuzzy input signals and fuzzy weights, which has been widely applied. A* algorithm is the best first search method in the heuristic search algorithm proposed by the Nilsson in 1980, which is a global planning algorithm based on area and searching optimum paths between nodes based on valuation function in figure (Brooks, 1986). A* algorithm has been applied in the fields of artificial intelligence and is used for path searching because of its high efficiency and fast speed (Farooq, Arif, Amar, & Tahir, 2010). The heuristic function h(n) in A* guarantees the minimum cost evaluation value from any node n to the target node, and continuously provides the “best” direction to search by using the heuristic information in the evaluation function. Based on the different terrain, the different heuristic function is selected to improve the efficiency.

Normally, robots often don’t walk along the feasible paths in the A* algorithm planning, but the edge and corner of the obstacles. The path planning using A* is very easy to collide with obstacles and lead to the failure in a real environment. The path planning using FNN is a relatively smooth path, but with the shortcoming of that the path is not the shortest, resulting in that the robot will be in the motion mutation and the dead-zone formation (Hofner, & Schmidt, 1995). The combination of the two algorithms is proposed to the improved FNN, providing smooth feasible optimal path for the robots. The grid map is divided into two stages by the two-stage fusion algorithm in the path planning (Azouaoui, Ouadah, Mansour, & Semani, 2011). The first stage is only an experience setting of the probability, and the obstacles and the feasible regions are distinguished in detail in the second stage. The number of grids in the first stage can be divided into the grids in the second stage, and the grids in the first stage is composed of N×N in the second stage. Therefore, the shortest path is briefly planned by A* algorithm in the first stage, and then a more reliable smooth path is planned by FNN in the second stage.