Multi-Robot Frontier-Based Exploration under Uncertainty with SLAM and Temporal Map Fusion

Nandhini Sriram, Shubhankit Tewari, Narayani Adane, Sudharsan, Ajan Muthuraj, Atriya Biswas

Keywords: Swarm, Frontier Exploration, SLAM, Autonomous System

TL;DR: A unified multi-drone system enabling robust exploration and rescue in GPS-denied environments via hybrid SLAM, cooperative mapping, and dynamic hazard avoidance.

Abstract: Autonomous multi-drone rescue in GPS-denied, obstacle-rich, and time-constrained environments demands tight integration of robust localization, collaborative mapping, and safe exploration. We present a fully autonomous swarm system of ten drones designed for the Swarm Rescue Challenge 2026, addressing these challenges through four tightly coupled contributions. First, a hybrid localization module combines FastSLAM with Iterative Closest Point (ICP) scan matching maintaining reliable pose estimation through both GPS available and unavailable zones. Second, an adaptive frontier-based exploration strategy clusters occupancy grid boundaries into scored regions and assigns them across the swarm using a utility to enable scalable cooperative coverage without redundant exploration. Third, an age-aware temporal map fusion scheme propagates each drone's occupancy grid to its peers every timestep yielding a collaboratively maintained, denoised global map. Fourth, a dynamic kill zone detection and avoidance system identifies destroyed drones via semantic observation and communication loss, permanently marking hazardous regions and broadcasting them swarm-wide for immediate avoidance. A five-state finite state machine governs overall mission execution, while multi-tiered stuck recovery and exponential potential field repulsion handle local minima and inter-drone collisions.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 11