유전자프로그래밍과 진화하드웨어를 이용한 자율이동로봇 제어 = Autonomous Robot Control using Genetic Programming and Evolvable Hardware

다국어 초록 (Multilingual Abstract)

This paper presents a method for evolving a target program using genetic programming. One of the advantage of genetic programming is that it can incorporate background knowledge in problem domains and evolve solutions of structured representation. But in the case of hardware implemented genetic programming, it is difficult to use genetic operators freely and implement the evolved chromosome properly on the given hardware. To overcome these problems, we propose a method that automatically decomposes a complex process into a number of sub-processes to improve flexibility and adaptability of the hardware. We apply our method to autonomous robot control problem. Using our method, the genetic tree found the optimal interpretaion logic for sensory input and evolved to the form which was capable of accomplishing the given task.