Reinforcement Learning Within the Classical Robotics Stack: A Case Study in Robot Soccer
This work addresses robot decision-making in partially observable, real-time, dynamic, and multi-agent settings for robotics applications, representing an incremental advancement by combining RL with existing methods.
The paper tackled the challenge of robot decision-making in complex environments by integrating reinforcement learning within a classical robotics stack, using a multi-fidelity sim2real approach and behavior decomposition, which led to victory in the 2024 RoboCup SPL Challenge Shield Division.
Robot decision-making in partially observable, real-time, dynamic, and multi-agent environments remains a difficult and unsolved challenge. Model-free reinforcement learning (RL) is a promising approach to learning decision-making in such domains, however, end-to-end RL in complex environments is often intractable. To address this challenge in the RoboCup Standard Platform League (SPL) domain, we developed a novel architecture integrating RL within a classical robotics stack, while employing a multi-fidelity sim2real approach and decomposing behavior into learned sub-behaviors with heuristic selection. Our architecture led to victory in the 2024 RoboCup SPL Challenge Shield Division. In this work, we fully describe our system's architecture and empirically analyze key design decisions that contributed to its success. Our approach demonstrates how RL-based behaviors can be integrated into complete robot behavior architectures.