Contrastive Variational Reinforcement Learning for Complex Observations
This addresses the problem of robust reinforcement learning in natural environments with complex visual inputs for robotics applications, representing an incremental improvement over existing model-based methods.
The paper tackles the challenge of complex visual observations in deep reinforcement learning by introducing Contrastive Variational Reinforcement Learning (CVRL), a model-based method that learns a contrastive variational model to avoid unnecessary modeling of observation space, achieving comparable performance on standard Mujoco tasks and significantly outperforming state-of-the-art methods on Natural Mujoco and robot box-pushing tasks with dynamic shadows.
Deep reinforcement learning (DRL) has achieved significant success in various robot tasks: manipulation, navigation, etc. However, complex visual observations in natural environments remains a major challenge. This paper presents Contrastive Variational Reinforcement Learning (CVRL), a model-based method that tackles complex visual observations in DRL. CVRL learns a contrastive variational model by maximizing the mutual information between latent states and observations discriminatively, through contrastive learning. It avoids modeling the complex observation space unnecessarily, as the commonly used generative observation model often does, and is significantly more robust. CVRL achieves comparable performance with state-of-the-art model-based DRL methods on standard Mujoco tasks. It significantly outperforms them on Natural Mujoco tasks and a robot box-pushing task with complex observations, e.g., dynamic shadows. The CVRL code is available publicly at https://github.com/Yusufma03/CVRL.