🤖 AI Summary
Multi-robot dynamic formation navigation in complex, confined environments (e.g., corridors) suffers from formation instability, difficulty balancing obstacle avoidance with inter-robot coordination, and poor adaptability to dynamic constraints.
Method: This paper proposes a hierarchical learning-driven framework for adaptive formation control and oscillation suppression. It introduces a novel integration of graph neural networks (GNNs) with hierarchical reinforcement learning (HRL), coupled with spring-damper dynamics modeling, to jointly regulate real-time formation deformation and suppress motion oscillations.
Contribution/Results: Evaluated across ROS, Gazebo, and Unity3D simulations as well as on physical robot platforms, the method enables task-oriented dynamic formation reconfiguration (e.g., wedge formations), achieves smoother and more robust navigation, and significantly outperforms state-of-the-art approaches. It effectively unifies collective coordination objectives with individual reactive obstacle avoidance, enhancing both scalability and adaptability in constrained environments.
📝 Abstract
Coordinated multi-robot navigation is an essential ability for a team of robots operating in diverse environments. Robot teams often need to maintain specific formations, such as wedge formations, to enhance visibility, positioning, and efficiency during fast movement. However, complex environments such as narrow corridors challenge rigid team formations, which makes effective formation control difficult in real-world environments. To address this challenge, we introduce a novel Adaptive Formation with Oscillation Reduction (AFOR) approach to improve coordinated multi-robot navigation. We develop AFOR under the theoretical framework of hierarchical learning and integrate a spring-damper model with hierarchical learning to enable both team coordination and individual robot control. At the upper level, a graph neural network facilitates formation adaptation and information sharing among the robots. At the lower level, reinforcement learning enables each robot to navigate and avoid obstacles while maintaining the formations. We conducted extensive experiments using Gazebo in the Robot Operating System (ROS), a high-fidelity Unity3D simulator with ROS, and real robot teams. Results demonstrate that AFOR enables smooth navigation with formation adaptation in complex scenarios and outperforms previous methods. More details of this work are provided on the project website: https://hcrlab.gitlab.io/project/afor.