自律機械システムの構造形態と知能の進化的設計法に関する研究

概要

論文の詳細を見る
This paper deals with the evolutionary design method of morphology and intelligence in autonomous mechanical/robotic systems. A robot performs a given task in a realistic virtual world including physics such as gravity, collision and friction; and receives fitness according to its performance. The robot which gained the higher fitness through genetic programming (GP) operators such as crossover or mutation evolves to a reasonably optimal morphology and intelligence. Thus the robot which is generated by this method will be a highly flexible robotic system because of its survival from the change environment including the uncertainty of real world. In this work, the robot is given three kinds of task which are specified by the number of object to be picked up and the task time. Various morphologies and interesting intelligence are emeged according to the difference in the task imposed. We examine the relationship between its morphology and intelligence and show the capability and flexibility of these evolutionary robotic systems.
一般社団法人日本機械学会の論文
1999-03-25