CBSB

CBSB implements CBS-Budget, a bounded-suboptimal variant of Conflict-Based Search for multi-agent path finding. It remains complete while trading a user-specified suboptimality bound for search effort, so a team’s paths can be found within a budget rather than at the cost of an exhaustive optimal search.

The implementation builds on the EECBS codebase. It requires Boost and CMake, and is run from the command line against the standard MAPF benchmark maps and scenario files, with flags for the solver, the suboptimality factor and whether bypassing is enabled.

Reference

J. Lim and P. Tsiotras, “CBS-Budget (CBSB): A Complete and Bounded Suboptimal Search for Multi-Agent Path Finding,” Artificial Intelligence, Vol. 346, 104349, 2025.

Derived from EECBS and distributed under that project’s academic, non-commercial license.

Visit the GitHub repository for the source code.