日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

週刊ニュースレター購読
arXiv:1306.0785

Robust multirobot coordination using priority encoded homotopic constraints

Robust multirobot coordination using priority encoded homotopic constraints

シェア:XThreadsFacebookLINEはてブBluesky

著者: Jean Gregoire, Silvère Bonnabel, Arnaud de La Fortelle

分類: cs.RO

原文アブストラクト

We study the problem of coordinating multiple robots along fixed geometric paths. Our contribution is threefold. First we formalize the intuitive concept of priorities as a binary relation induced by a feasible coordination solution, without excluding the case of robots following each other on the same geometric path. Then we prove that two paths in the coordination space are continuously deformable into each other if and only if they induce the \emph{same priority graph}, that is, the priority graph uniquely encodes homotopy classes of coordination solutions. Finally, we give a simple control law allowing to safely navigate into homotopy classes \emph{under kinodynamic constraints} even in the presence of unexpected events, such as a sudden robot deceleration without notice. It appears the freedom within homotopy classes allows to much deviate from any pre-planned trajectory without ever colliding nor having to re-plan the assigned priorities.