Coordination Under Constraints: A Wireless Signal-Based Framework for Failure-Aware Multi-Robot Exploration

Abstract

This thesis presents the WSR Framework, a decentralized framework for multi-robot exploration in environments with limited communication and no shared global map. The system enables teams of robots to coordinate using only onboard sensing and lightweight relative position estimates derived from wireless signal exchanges from WiFi and UWB sensors. Each robot selects exploration targets independently, guided by a utility function that accounts for local information gain and inferred teammate positions. A key contribution is a confidence-based failure detection mechanism that allows each robot to respond to teammate failures using only motion history. The confidence value adjusts how exploration targets are selected, which helps to redistribute coverage progress. The full system is validated through both simulation and hardware experiments. The results show significant reductions in robot overlap compared to zero-coordination baselines, and performance that approaches that of centralized strategies, without requiring explicit communication or shared maps.