Closing the Sim-to-Real Loop: Adapting Simulation Randomization with Real World Experience

TL;DR

This paper introduces a method to adapt simulation parameters using real-world experience, enhancing the transfer of policies from simulation to various real robots in tasks like peg-in-hole and drawer opening.

Contribution

It proposes an adaptive simulation randomization technique that uses real-world roll-outs to improve policy transfer without manual tuning.

Findings

Policies successfully transferred to different robots.

Improved simulation-to-real transfer performance.

Effective adaptation of simulation parameters with minimal real-world data.

Abstract

We consider the problem of transferring policies to the real world by training on a distribution of simulated scenarios. Rather than manually tuning the randomization of simulations, we adapt the simulation parameter distribution using a few real world roll-outs interleaved with policy training. In doing so, we are able to change the distribution of simulations to improve the policy transfer by matching the policy behavior in simulation and the real world. We show that policies trained with our method are able to reliably transfer to different robots in two real world tasks: swing-peg-in-hole and opening a cabinet drawer. The video of our experiments can be found at https://sites.google.com/view/simopt