Real-Time Coordination of Multiple Robotic Arms With Reactive Trajectory Modulation

Abstract

Efficiently coordinating multiple robotic arms is vital for secure and optimal operation in a shared workspace. This requires not only successful task completion but also minimizing collision risks from overlapping movements. Introducing real-time motion modulation adds an extra layer of challenge to this coordination task. In this article, we introduce a novel framework for real-time multiarm coordination, offering two main contributions: First, based on fuzzy model-based movement primitives, we propose a method for real-time trajectory modulation by learning from single demonstrations. This capability allows robots to modulate their motions online to reach arbitrary new desired places smoothly without necessitating extra demonstrations from users. Second, our framework incorporates a real-time multiarm coordination strategy that seamlessly integrates the trajectory modulation method with an extended reactive approach. This strategy empowers multiple robotic arms operating within a shared workspace to dynamically regulate their movements and execute tasks simultaneously in a human-desired manner while reactively avoiding mutual collisions. In the experiments, we utilize a group of robotic arms working in a shared workspace to validate the effectiveness of our framework and to make comparisons with state-of-the-art methods.