Unified State Representation Learning under Data Augmentation
This addresses the challenge of generalization for RL agents in real-world applications, though it appears incremental as it builds on existing benchmarks and methods.
The paper tackles the problem of zero-shot policy transfer in reinforcement learning by proposing USRA, a framework that learns unified state representations using data augmentation, achieving 14.3% better domain adaptation performance on the DeepMind Control Benchmark.
The capacity for rapid domain adaptation is important to increasing the applicability of reinforcement learning (RL) to real world problems. Generalization of RL agents is critical to success in the real world, yet zero-shot policy transfer is a challenging problem since even minor visual changes could make the trained agent completely fail in the new task. We propose USRA: Unified State Representation Learning under Data Augmentation, a representation learning framework that learns a latent unified state representation by performing data augmentations on its observations to improve its ability to generalize to unseen target domains. We showcase the success of our approach on the DeepMind Control Generalization Benchmark for the Walker environment and find that USRA achieves higher sample efficiency and 14.3% better domain adaptation performance compared to the best baseline results.