Neuro-Fuzzy Mobile Robot Navigation

Abstract

In mobile robot navigation, it is fundamental to have goal seeking and obstacle avoidance in the navigation algorithm. In this paper neural networks are used to learn both goal seeking and obstacle avoidance from a dataset generated by two different fuzzy logic navigation algorithms. The neural network was able to learn from both algorithms and produce a smoother path than the two. Additionally, a neural network was able to learn how to escape a concave obstacle.