An algorithm combined the fuzzy theory with the neural network, which has the advantages of the self-learning, recognition and information processing.
Published in Chapter:
An Adaptive Path Planning Based on Improved Fuzzy Neural Network for Multi-robot Systems
Zhiguo Shi (University of Science and Technology, China), Huan Zhang (University of Science and Technology, China), Jingyun Zhou (University of Science and Technology, China), and Junming Wei (Australian National University, Australia)
Copyright: © 2016
|Pages: 25
DOI: 10.4018/978-1-4666-9572-6.ch012
Abstract
The fuzzy neural network (FNN) is the combination of fuzzy theory with neural network, which has advantages of validity and adaptability in robot path planning. However, the path planning based on the FNN is not optimal because of the limitations of the subjective experience and motion mutation and the dead-zone. In this paper, FNN is improved by using A* graph search algorithm to guarantee an optimal path, providing the rationality and the feasibility, in which the grid map is divided into two stages, including the A* algorithm in the first stage and FNN in the second stage. In addition, a neural network based on adaptive control strategy is introduced to compensate the sensor failure and ensures the stability, which is caused by the loss of data and information uncertainty. The simulation results show that the approach is with effective performance in the robot path planning.