1204 電動車いすによる自律移動を促進する開発(福祉・バリアフリー,OS7 福祉・バリアフリー)

概要

論文の詳細を見る
This paper presents an approach to guide an electric wheelchair to run straight-ahead adaptively in order to reduce user's burden of operating an input device during driving on a slope road and a rugged space. The approach controls input voltages into two DC motors using the reinforcement learning method with a state space filter. The approach to have the following features; it is not required to design beforehand the proper control rule for each user, to redesign in the cases when the condition of a road is changed and a characteristics of the wheelchair included the user is changed, or to design beforehand the proper inner state space of a reinforcement learning agent. Moreover, the basic dynamics of an electric wheelchair has been modeled for confirming effectiveness of the proposed approach through simulations. The above features, the effectiveness of the approach, and the applicability to an actual wheelchair have been confirmed through a series of simulations.
社団法人日本機械学会の論文
2007-12-12