Reset-Free Lifelong Learning with Skill-Space Planning
This work tackles the problem of continuous adaptation for RL agents in dynamic, real-world-like environments, which is a significant challenge for developing robust AI.
This paper addresses the challenge of lifelong reinforcement learning in non-stationary and non-episodic environments. The proposed Lifelong Skill Planning (LiSP) framework enables long-horizon planning and avoids catastrophic failures in such challenging environments, as demonstrated on gridworld and MuJoCo benchmarks.
The objective of lifelong reinforcement learning (RL) is to optimize agents which can continuously adapt and interact in changing environments. However, current RL approaches fail drastically when environments are non-stationary and interactions are non-episodic. We propose Lifelong Skill Planning (LiSP), an algorithmic framework for non-episodic lifelong RL based on planning in an abstract space of higher-order skills. We learn the skills in an unsupervised manner using intrinsic rewards and plan over the learned skills using a learned dynamics model. Moreover, our framework permits skill discovery even from offline data, thereby reducing the need for excessive real-world interactions. We demonstrate empirically that LiSP successfully enables long-horizon planning and learns agents that can avoid catastrophic failures even in challenging non-stationary and non-episodic environments derived from gridworld and MuJoCo benchmarks.