We extend our state of the art information sharing algorithm to a high-fidelity simulation with advanced sensor and communication models. We ensure that a team of agents performing routine surveillance can self-police and identify if any in their midst are venturing off their nominal flight-paths. We demonstrate three different capabilities for maintaining robust information sharing: (1) resilience against a faulty sensor; (2) reliability in a communication-denied environment; and (3) performance against adversarial hacking.
Authors: Bo Wu, Steven Carr, Suda Bharadwaj, Zhe Xu, and Ufuk Topcu
Link to Paper: https://arxiv.org/pdf/2008.00164.pdf