When to Ask for Help: Proactive Interventions in Autonomous Reinforcement Learning
This addresses the challenge of reducing constant monitoring for irreversible states in autonomous agents, which is incremental as it builds on existing methods for intervention efficiency.
The paper tackles the problem of irreversible states in autonomous reinforcement learning by proposing an algorithm that learns to detect and avoid such states and proactively requests human intervention when needed, resulting in better sample- and intervention-efficiency compared to existing methods in continuous control environments.
A long-term goal of reinforcement learning is to design agents that can autonomously interact and learn in the world. A critical challenge to such autonomy is the presence of irreversible states which require external assistance to recover from, such as when a robot arm has pushed an object off of a table. While standard agents require constant monitoring to decide when to intervene, we aim to design proactive agents that can request human intervention only when needed. To this end, we propose an algorithm that efficiently learns to detect and avoid states that are irreversible, and proactively asks for help in case the agent does enter them. On a suite of continuous control environments with unknown irreversible states, we find that our algorithm exhibits better sample- and intervention-efficiency compared to existing methods. Our code is publicly available at https://sites.google.com/view/proactive-interventions