Benchmarking Model-Based Reinforcement Learning

TL;DR

This paper provides a comprehensive benchmarking suite for model-based reinforcement learning, comparing various algorithms across standardized environments to identify key challenges and facilitate future research.

Contribution

It introduces a unified benchmarking framework with multiple environments and evaluates existing MBRL algorithms to standardize performance assessment.

Findings

Identified three key research challenges: dynamics bottleneck, planning horizon dilemma, early-termination dilemma.

Provided a comprehensive comparison of MBRL algorithms under unified settings.

Open-sourced the benchmark suite for community use.

Abstract

Model-based reinforcement learning (MBRL) is widely seen as having the potential to be significantly more sample efficient than model-free RL. However, research in model-based RL has not been very standardized. It is fairly common for authors to experiment with self-designed environments, and there are several separate lines of research, which are sometimes closed-sourced or not reproducible. Accordingly, it is an open question how these various existing MBRL algorithms perform relative to each other. To facilitate research in MBRL, in this paper we gather a wide collection of MBRL algorithms and propose over 18 benchmarking environments specially designed for MBRL. We benchmark these algorithms with unified problem settings, including noisy environments. Beyond cataloguing performance, we explore and unify the underlying algorithmic differences across MBRL algorithms. We characterize three key research challenges for future MBRL research: the dynamics bottleneck, the planning horizon dilemma, and the early-termination dilemma. Finally, to maximally facilitate future research on MBRL, we open-source our benchmark in http://www.cs.toronto.edu/~tingwuwang/mbrl.html.