유전자 알고리즘을 이용한 자율주행 로봇의 진화에 관한 연구

초록

In this paper, the concept of evolvable hardware and evolutionary robotics are introduced and constructing the mobile robot controller without human operator is suggested. The robot controller is evolved to avoid obstacles by genetic learning which determines the weights between sensor inputs and motor outputs. Genetic algorithms which is executed in a computer(PC) searches the best weights by interacting with robot performanc under it's environment. The experiment is done by real mobile robot Khepera and a simple GA.