抄録
F-016
暫定的優先順位増加による複数エージェント経路計画アルゴリズムの拡張
奥村圭祐・田村康将・Xavier Défago(東工大)
The problem that makes multiple agents move to their destinations without collisions is called Multi-agent Path Finding (MAPF), which is receiving a lot of attention due to its high practically, e.g., warehouse applications with autonomous vehicles. Priority Inheritance with Backtracking (PIBT) is a decoupled approach to solve the iterative MAPF problem. PIBT ensures that all agents reach their destinations in finite time when the environment is given as a biconnected graph. In this paper, we extend target graphs of PIBT by introducing temporary inflation of agent priorities and discuss the effectiveness of the approach.