A method for autonomous action generation based on the easiness criterion

Future robots with complex mechanism will be requested to accomplish various large-scale tasks. When only the task objective is given the control of robot action/motion has a large number of redundant degrees of freedom. For the redundancy resolution a universal criterion should be introduced instead of giving an additional artificial criterion to task by task. This article is concerned with the autonomous action generation problem with such redundancy resolution involved. The policy for making the action generation algorithm is discussed and using the “easiness” as a universal criterion the easiness-based action generation is proposed. As a simple example the muscle-based joint control is investigated.