Nesting, safety, layering, and autonomy: a coordinated computational intelligence (CCI) approach to folding legged robot locomotion and gymnastic training

Abstract

A coordinated computational intelligence (CCI) approach is presented for folding-legged robot locomotion and gymnastics. The new approach is based on the hypotheses that (1) the human cerebellum consists of a school of semiautonomous neural-fuzzy cerebellar agents coordinated by common sense cerebellar motion laws; and (2) agents can learn individually and learned agents can discover new agents and cerebellar laws. Based on the CCI theory, a multiagent cerebellar architecture, MAC-J, is extended from 3-link uniped locomotion control to 4-link uniped locomotion control and from locomotion control to gymnastic training. Basic ideas are illustrated with a 4-link uniped simulation The principle of "increasing intelligence with decreasing precision" in hierarchical control is extended to four principles for the nesting, safety, layering and autonomy of cerebellar agents.