A path planning behavior learning of six legged robot based on human teaching model

This paper deals with a six legged robot path planning behavior learning based on human teaching. The six legged robot requires several capabilities such as perceiving, acting, communicating, and surviving. Furthermore, the six legged robot should learn various behaviors through the interaction with owner (human). This paper focuses on teaching path planning to the robot. Basically, the owner can teach the path planning by simple communication based on trial and error. Human teaches the path planning by using his hand. It will give an input of fuzzy controller used for the input and output movement of legged robot. Generally, legged robot will learn some pattern by a number of learning iteration. In every iterations the robot will have some punishment and reward coefficient of reinforcement learning used to generate better population of Genetic Algorithm until it can reach its minimum level of punishment. Experimental result shows that the legged robot can learn the path planning using the proposed method through the interaction with owner.