Bayesian Adaptive Robotic Control System (BARCS)

Created for the DARPA Subterranean (SubT) Challenge Virtual Track Competitions by MTRI Researchers, BARCS is an autonomous controller integrated into both UAVs and UGVs for rapid subterranean exploration, mapping, and object identification.

Technical challenges of the Virtual Track include: Austere navigation, degraded sensing, severe comms, dynamic terrain, obstacles, endurance limits, and full autonomy

Key innovations of the BARCS approach include:

A space-filling sphere method to identify interior space adaptively in a wide variety of environments
- Enables frontier generation, path branching, and obstacle avoidance
Integrated graph models for traversed and frontier space that are shared and merged between agents
- Enables comms-constrained map sharing between agents and collaborative search methods
Dynamic updates to traversed space through node and edge blacklisting
- Enhances global path planning in dynamic environments
Fully autonomous collaborative exploration