1503 ニューラルネットによる移動ロボットの運動制約を考慮した障害物回避を伴う経路計画(GS-15 移動ロボット(1),研究発表講演)

概要

論文の詳細を見る
This research proposes the introduction of clothoid arcs to the learning path of a neural network in the environment where a mobile robot moves to a destination while avoiding unknown obstacles. Since a mobile robot has nonholomic restraints and its operating performance has a close relation to the curvature of a moving path, it is necessary to consider the motion limitation when the path is planned. The path combining clothoid arcs, which guarantee continuity of curvature, with circular arcs was used to give the learning data. The simulation shows that the mobile robot can reach a destination while avoiding unknown obstacles. The test run of a real robot illustrates the effectiveness of the technique.
2005-03-18