Pre-training of Deep RL Agents for Improved Learning under Domain Randomization

Amiranashvili, Artemij, Argus, Max, Hermann, Lukas, Burgard, Wolfram, Brox, Thomas

arXiv.org Artificial Intelligence 

Visual domain randomization in simulated environments is a widely used method to transfer policies trained in simulation to real robots. However, domain randomization and augmentation hamper the training of a policy. As reinforcement learning struggles with a noisy training signal, this additional nuisance can drastically impede training. For difficult tasks it can even result in complete failure to learn. To overcome this problem we propose to pre-train a perception encoder that already provides an embedding invariant to the randomization. We demonstrate that this yields consistently improved results on a randomized version of DeepMind control suite tasks and a stacking environment on arbitrary backgrounds with zero-shot transfer to a physical robot.

Duplicate Docs Excel Report

Title
None found

Similar Docs  Excel Report  more

TitleSimilaritySource
None found