Robust Multiple-Path Orienteering Problem: Securing Against Adversarial Attacks

The multiple-path orienteering problem asks for paths a team of robots that maximize the total reward collected while satisfying budget constraints on path length. This models many multirobot routing tasks, such as exploring unknown environments and information gathering environmental monitoring. In this article, we focus how to make robot robust failures when operating in adversarial environments. We introduce (RMOP), where seek worst case guarantees against an adversary is capable attacking at most $\alpha$ robots. consider two versions problem: RMOP offline online. version, there no communication or replanning execute their plans, our main contribution general approximation scheme with bounded guarantee depends factor single-robot orienteering. particular, show algorithm yields a: 1) constant-factor cost function modular; 2) notation="LaTeX">$\log$ submodular; 3) submodular, but are allowed exceed budgets by amount. online modeled two-player sequential game solved adaptively receding horizon fashion based Monte Carlo tree search. addition theoretical analysis, perform simulation studies ocean monitoring tunnel information-gathering applications demonstrate efficacy approach.