3929 自律移動体におけるSLAM手法と協調した行動ルール学習法(J18-1 自律型グラウンドビークル,エアビークルの新展開(1),J18 自律型グラウンドビークル,エアビークルの新展開,2005年度年次大会)

概要

論文の詳細を見る
This paper proposes a new learning method of action rule tables for autonomous vehicles combined with SLAM. When autonomous vehicles are operated in uncertainty environments like construction fields, one of the most important functions of those vehicles is to autonomously acquire action rule tables which is combined with mapping data and used for planning the behavior of such vehicles. Our proposed method is aim to improve the reward system in Q-Learning method, which is one of the well-known reinforcement learning methods. The effectiveness of our proposed method is evaluated by numerical simulations.
一般社団法人日本機械学会の論文
2005-09-18