Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : May 10, 2017 - May 13, 2017
Service robots need to execute complex tasks in changing situation with actions adapting to the situation. Even if a robot has acquired the action intelligence that gives the optimal task execution manner in the original situation, the robot cannot execute the task in new situations as good as before the situation changes because of the lack of experiences in the new situations. In this research we suppose that multiple robots have acquired their own action intelligence that gives varied task execution manner in the original situation. And we propose a method for the robots to efficiently adapt to new situations by referring the other robots ’task execution manners. The proposed method is applied to a simple numerical example for ball throwing actions to investigate the effectivity.
