Task and Domain Adaptive Reinforcement Learning for Robot Control
This addresses the problem of single-task orientation and environmental inflexibility in robot control for robotics researchers, though it is incremental as it builds on existing transfer learning techniques.
The paper tackles the limited adaptability of deep reinforcement learning in robot control by introducing an adaptive agent that uses transfer learning to adjust policies for different tasks and environmental conditions, achieving zero-shot transfer to real-world blimp control for various tasks.
Deep reinforcement learning (DRL) has shown remarkable success in simulation domains, yet its application in designing robot controllers remains limited, due to its single-task orientation and insufficient adaptability to environmental changes. To overcome these limitations, we present a novel adaptive agent that leverages transfer learning techniques to dynamically adapt policy in response to different tasks and environmental conditions. The approach is validated through the blimp control challenge, where multitasking capabilities and environmental adaptability are essential. The agent is trained using a custom, highly parallelized simulator built on IsaacGym. We perform zero-shot transfer to fly the blimp in the real world to solve various tasks. We share our code at https://github.com/robot-perception-group/adaptive_agent.