Vector Quantized Models for Planning
This addresses a key limitation in model-based RL for real-world applications where environments are often stochastic and partially observable, representing an incremental advance over deterministic methods.
The paper tackles planning in stochastic and partially-observable environments by using discrete autoencoders to model multiple action effects, significantly outperforming an offline MuZero variant on stochastic chess and scaling to complex 3D environments like DeepMind Lab.
Recent developments in the field of model-based RL have proven successful in a range of environments, especially ones where planning is essential. However, such successes have been limited to deterministic fully-observed environments. We present a new approach that handles stochastic and partially-observable environments. Our key insight is to use discrete autoencoders to capture the multiple possible effects of an action in a stochastic environment. We use a stochastic variant of Monte Carlo tree search to plan over both the agent's actions and the discrete latent variables representing the environment's response. Our approach significantly outperforms an offline version of MuZero on a stochastic interpretation of chess where the opponent is considered part of the environment. We also show that our approach scales to DeepMind Lab, a first-person 3D environment with large visual observations and partial observability.