Citation Context

...of lmoniotion patterns is applied to a niulti-legged format.ioii and a snakelike formation, as a result effective locomotion hss been obtained. Keyword:Hyper-redundant systems, Q-learning: Reinforcement learning, Genetic Algorithm, QDSEGA 1 ' Introduction Recently. flexible autonomous system that. can accoinplih xwious t.asks automatically has been much attention. Howewr mat. of conventional researches of autonomous system are restricted to some simple rohuts like a mobile robot [11[2]. 50 it cannot, arconiplish various tasks. On the other hand, in the field of researches of redundant robot[3][4][5], the mbut rw accomplish YBTious tasks using its redundancy hut it requires control nicthods for each tasks. So it cannot be autonomous system. Considering these points, we propose new approach to realize autonomous system. To realize the system, we employ a hyper-redundaut robot and a reinforcement learnillg controller *QDSEGA[6]". The hyper-redundant robot is composs of many uuiform units aiid separation and recombination is possible. Usiiig its redundancy the hyper-redundant robot ran accomplish various tasks. On the other hand, QDSEGA (Q-learning with dynamic st.ruct.uring exploration s...