IR2: Implicit Rendezvous for Robotic Exploration Teams under Sparse Intermittent Connectivity

Accepted for oral presentation at the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2024).

Abstract

Information sharing is critical in time-sensitive and realistic multi-robot exploration, especially for smaller robotic teams in large-scale environments where connectivity may be sparse and intermittent. Existing methods often overlook such communication constraints by assuming unrealistic global connectivity. Other works consider realistic hardware limitations that restrict communication range, requiring robots to maintain close proximity or line of sight for stable connectivity. For instance, preplanned rendezvous approaches are often inefficient due to unnecessary detours resulting from poorly timed rendezvous, while pursuit-based approaches are often short-sighted due to their greedy nature given the current state. We present $IR^2$, a deep reinforcement learning approach to information sharing for multi-robot exploration. Leveraging attention-based neural networks trained via reinforcement and curriculum learning, $IR^2$ allows robots to reason about the future impact of current decisions. In addition, we propose a hierarchical graph formulation to maintain a sparse yet informative graph, enabling our approach to scale to large-scale environments. We present simulation results in three large-scale Gazebo environments, which show that our approach yields $8.6 - 34.1%$ shorter exploration paths compared to state-of-the-art baselines, and leads to significantly improved map consistency among robots.