Representation-Optimal Multi-Robot Motion Planning Using Conflict-Based Search
Representation-Optimal Multi-Robot Motion Planning Using Conflict-Based Search
复制标题
使用基于冲突的搜索进行表示优化多机器人运动规划
DOI：
10.1109/lra.2021.3068910
复制
发表时间：
2019
影响因子：
5.2
通讯作者：
N. Amato
中科院分区：
文献类型：
--
作者：
Irving Solis;James Motes;R. Sandström;N. Amato
多智能体运动规划 (MAMP) 是计算一组智能体的可行路径的问题，每个智能体在连续状态空间内都有单独的起始状态和目标状态。现有方法可以分为耦合方法和解耦方法，耦合方法提供最优解决方案，但难以实现可扩展性；解耦方法提供可扩展解决方案，但不提供最优性保证。最近的工作探索了混合方法，在更简单的离散子问题多智能体寻路（MAPF）中利用耦合和解耦方法的优点。在这项工作中，我们将混合 MAPF 的最新发展应用于 MAMP 的连续域。我们展示了我们的方法的可扩展性，可以管理多达 32 个代理的组，展示了处理多达 8 个高自由度操纵器的能力，并为异构团队进行规划。在所有情况下，我们的方法都能显着加快计划速度，同时提供更高质量的解决方案。
Multi-Agent Motion Planning (MAMP) is the problem of computing feasible paths for a set of agents each with individual start and goal states within a continuous state space. Existing approaches can be split into coupled methods which provide optimal solutions but struggle with scalability or decoupled methods which provide scalable solutions but offer no optimality guarantees. Recent work has explored hybrid approaches that leverage the advantages of both coupled and decoupled approaches in an easier discrete subproblem, Multi-Agent Pathfinding (MAPF). In this work, we adapt recent developments in hybrid MAPF to the continuous domain of MAMP. We demonstrate the scalability of our method to manage groups of up to 32 agents, demonstrate the ability to handle up to 8 high-DOF manipulators, and plan for heterogeneous teams. In all scenarios, our approach plans significantly faster while providing higher quality solutions.