Multi-Goal Multi-Agent Path Finding via Decoupled and Integrated Goal Vertex Ordering

We introduce multi-goal multi agent path finding (MG-MAPF) which generalizes the standard discrete multi-agent (MAPF) problem. While task in MAPF is to navigate agents an undirected graph from their starting vertices one individual goal vertex per agent, MG-MAPF assigns each multiple and visit of them at least once. Solving not only requires collision free paths for but also determining order visiting agent's so that common objectives like sum-of-costs are optimized. suggest two novel algorithms using different paradigms address MG-MAPF: a heuristic search-based algorithm called Hamiltonian-CBS (HCBS) compilation-based built satisfiability modulo theories (SMT), SMT-Hamiltonian-CBS (SMT-HCBS).