Authors:
Matouš Kulhan
and
Pavel Surynek
Affiliation:
Faculty of Information Technology, Czech Technical University in Prague, Thákurova 9, 160 00 Praha 6, Czech Republic
Keyword(s):
Path Finding, Planning, Acting, Multiple Agents, Indoor Quadcopters, Crazyflie, MAPF, Localization, Testing Platform.
Abstract:
We study the planning-acting loop for multi-agent path finding with continuous time (MAPF R ). The standard MAPF is a problem of navigating agents from their start positions to specified individual goal positions so that agents do not collide with each other. The standard MAPF takes place in a discrete graph with agents located in its vertices and instantaneous moves of agents across edges. MAPFR adds continuous elements to MAPF via allowing agents to wait in a vertex for arbitrary length of time to avoid the collision. We focus in this paper on executing MAPFR plans with a group of Crazyflies, small indoor quadcopters. We show how to modify the existing continuous-time conflict-based search algorithm (CCBS) for MAPF R to produce plans that are suitable for execution with the quadcopters. Our platform can be used for testing suitability of variants of MAPF for execution with real agents. Our finding is that the MAPF variant with continuous time and the related CCBS algorithm allows f
or extensions that can produce safe plans for quadcopters, namely cylindrical protection zone around each quadcopter can be introduced at the planning level.
(More)