Model-Based Reinforcement Learning via Meta-Policy Optimization
Abstract
Model-based reinforcement learning approaches carry the promise of being data efficient. However, due to challenges in learning dynamics models that sufficiently match the real-world dynamics, they struggle to achieve the same asymptotic performance as model-free methods. We propose Model-Based Meta-Policy-Optimization (MB-MPO), an approach that foregoes the strong reliance on accurate learned dynamics models. Using an ensemble of learned dynamic models, MB-MPO meta-learns a policy that can quickly adapt to any model in the ensemble with one policy gradient step. This steers the meta-policy towards internalizing consistent dynamics predictions among the ensemble while shifting the burden of behaving optimally w.r.t. the model discrepancies towards the adaptation step. Our experiments show that MB-MPO is more robust to model imperfections than previous model-based approaches. Finally, we demonstrate that our approach is able to match the asymptotic performance of model-free methods while requiring significantly less experience.
Cite
@article{arxiv.1809.05214,
title = {Model-Based Reinforcement Learning via Meta-Policy Optimization},
author = {Ignasi Clavera and Jonas Rothfuss and John Schulman and Yasuhiro Fujita and Tamim Asfour and Pieter Abbeel},
journal= {arXiv preprint arXiv:1809.05214},
year = {2018}
}
Comments
First 2 authors contributed equally. Accepted for Conference on Robot Learning (CoRL)