Topology-based group routing in partially known environments

Autor: Jory Denny, Benjamin T. Fine
Rok vydání: 2020
Předmět:
Zdroj: SAC
DOI: 10.1145/3341105.3373965
Popis: Multi-agent routing is an important aspect to studying livestock management, traffic patterns, and pedestrian egress. However, much of the literature does not fully account for the underlying dynamics of the agents' group level behavior or fluctuating quality of the paths through an environment. We present an online adaptive learning algorithm for maximizing the throughput of a group of agents operating under uncertain dynamics through a partially known environment. Our approach relies on the novel use of a representation of an environment's topology, specifically, a Reeb Graph. The approach places directing agents at the vertices of this graph and provides an initial graph weighting and update step to progressively determine which paths that subgroups of agents should navigate in order to traverse the space. We validate our algorithm with four complex scenarios showing its flexibility to varying topological structures, uncertain group dynamics, and sudden environmental changes.
Databáze: OpenAIRE