Solving Rubik's Cube with a Robot Hand

TL;DR

This paper presents a novel simulation training method called automatic domain randomization (ADR) that enables a humanoid robot hand to solve a Rubik's Cube in the real world, demonstrating advanced sim2real transfer and emergent meta-learning.

Contribution

Introduction of ADR, a new algorithm for automatic environment randomization, combined with a custom robot platform, to successfully transfer complex manipulation skills from simulation to real robots.

Findings

Control policies trained with ADR transfer effectively to real robots.

Memory-augmented models show signs of meta-learning at test time.

The approach enables solving complex tasks like Rubik's Cube with a humanoid robot hand.

Abstract

We demonstrate that models trained only in simulation can be used to solve a manipulation problem of unprecedented complexity on a real robot. This is made possible by two key components: a novel algorithm, which we call automatic domain randomization (ADR) and a robot platform built for machine learning. ADR automatically generates a distribution over randomized environments of ever-increasing difficulty. Control policies and vision state estimators trained with ADR exhibit vastly improved sim2real transfer. For control policies, memory-augmented models trained on an ADR-generated distribution of environments show clear signs of emergent meta-learning at test time. The combination of ADR with our custom robot platform allows us to solve a Rubik's cube with a humanoid robot hand, which involves both control and state estimation problems. Videos summarizing our results are available: https://openai.com/blog/solving-rubiks-cube/