CLOUD: Contrastive Learning of Unsupervised Dynamics
This work addresses the problem of efficient dynamics learning for agents in complex control tasks, offering a novel method that could improve performance in robotics and AI applications.
The paper tackles the challenge of learning dynamics from high-dimensional observations by proposing a contrastive learning approach for unsupervised forward and inverse dynamics, demonstrating efficacy in tasks like goal-directed planning and imitation from observations.
Developing agents that can perform complex control tasks from high dimensional observations such as pixels is challenging due to difficulties in learning dynamics efficiently. In this work, we propose to learn forward and inverse dynamics in a fully unsupervised manner via contrastive estimation. Specifically, we train a forward dynamics model and an inverse dynamics model in the feature space of states and actions with data collected from random exploration. Unlike most existing deterministic models, our energy-based model takes into account the stochastic nature of agent-environment interactions. We demonstrate the efficacy of our approach across a variety of tasks including goal-directed planning and imitation from observations. Project videos and code are at https://jianrenw.github.io/cloud/.