Localization-Driven Multi-Robot Exploration for Indoor Search and Rescue

Search and Rescue (SAR) operations in complex indoor environments require efficient robotic coordination to ensure timely victim detection while optimizing area coverage and operation time. However, relying solely on exploration may be insufficient, as early victim localization can enhance path planning and prioritize victim access over the exploration of empty areas. In this work, we integrate Radio Frequency (RF)-based victim localization into State-of-Art multi-robot exploration approaches to guide decision-making and prioritize areas with higher likelihood of victim presence. Specifically, we perform probabilistic victim localization using Angle of Arrival (AoA) triangulation of RF signals, enhanced by Received Signal Strength Indicator (RSSI) filtering. This information is then used to dynamically update exploration priorities and replan robot paths toward potential victim locations. We evaluate the proposed method using the Cramér-Rao Lower Bound (CRLB) and Root Mean Square Error (RMSE). Extensive simulations confirm that the proposed approach significantly increases SAR efficiency, allowing up to ~34% of victims to be reached earlier in typical indoor environments and reducing the total mission time by up to 50%.