Trajectory-wise Multiple Choice Learning for Dynamics Generalization in Reinforcement Learning
This addresses the problem of dynamics generalization in reinforcement learning for control tasks, offering an incremental improvement over existing methods.
The paper tackles the challenge of learning generalizable dynamics models in model-based reinforcement learning by proposing a trajectory-wise multiple choice learning algorithm that clusters environments and adapts online, achieving superior zero-shot generalization performance across various control tasks compared to state-of-the-art methods.
Model-based reinforcement learning (RL) has shown great potential in various control tasks in terms of both sample-efficiency and final performance. However, learning a generalizable dynamics model robust to changes in dynamics remains a challenge since the target transition dynamics follow a multi-modal distribution. In this paper, we present a new model-based RL algorithm, coined trajectory-wise multiple choice learning, that learns a multi-headed dynamics model for dynamics generalization. The main idea is updating the most accurate prediction head to specialize each head in certain environments with similar dynamics, i.e., clustering environments. Moreover, we incorporate context learning, which encodes dynamics-specific information from past experiences into the context latent vector, enabling the model to perform online adaptation to unseen environments. Finally, to utilize the specialized prediction heads more effectively, we propose an adaptive planning method, which selects the most accurate prediction head over a recent experience. Our method exhibits superior zero-shot generalization performance across a variety of control tasks, compared to state-of-the-art RL methods. Source code and videos are available at https://sites.google.com/view/trajectory-mcl.