DeepMNavigate: Deep Reinforced Multi-Robot Navigation Unifying Local & Global Collision Avoidance
This addresses navigation challenges for multi-robot systems in crowded settings, representing an incremental advancement by combining local and global information.
The paper tackles multi-robot navigation in dense environments by introducing DeepMNavigate, a deep reinforcement learning algorithm that unifies local and global collision avoidance, achieving improved performance over prior methods in benchmarks with tens of agents and narrow passages.
We present a novel algorithm (DeepMNavigate) for global multi-agent navigation in dense scenarios using deep reinforcement learning (DRL). Our approach uses local and global information for each robot from motion information maps. We use a three-layer CNN that takes these maps as input to generate a suitable action to drive each robot to its goal position. Our approach is general, learns an optimal policy using a multi-scenario, multi-state training algorithm, and can directly handle raw sensor measurements for local observations. We demonstrate the performance on dense, complex benchmarks with narrow passages and environments with tens of agents. We highlight the algorithm's benefits over prior learning methods and geometric decentralized algorithms in complex scenarios.