Solving the Real Robot Challenge using Deep Reinforcement Learning

TL;DR

This paper presents a pure reinforcement learning approach using domain randomization and Hindsight Experience Replay to enable a robot to carry a cube along specified trajectories, outperforming traditional control methods.

Contribution

The first pure learning-based method to successfully solve the Real Robot Challenge by training in simulation and transferring to a real robot with minimal expert knowledge.

Findings

Successfully transferred the learned policy from simulation to real robot.

Outperformed all other challenge submissions, including traditional control methods.

Demonstrated effective lifting and trajectory following of a cube with a robotic hand.

Abstract

This paper details our winning submission to Phase 1 of the 2021 Real Robot Challenge; a challenge in which a three-fingered robot must carry a cube along specified goal trajectories. To solve Phase 1, we use a pure reinforcement learning approach which requires minimal expert knowledge of the robotic system, or of robotic grasping in general. A sparse, goal-based reward is employed in conjunction with Hindsight Experience Replay to teach the control policy to move the cube to the desired x and y coordinates of the goal. Simultaneously, a dense distance-based reward is employed to teach the policy to lift the cube to the z coordinate (the height component) of the goal. The policy is trained in simulation with domain randomisation before being transferred to the real robot for evaluation. Although performance tends to worsen after this transfer, our best policy can successfully lift the real cube along goal trajectories via an effective pinching grasp. Our approach outperforms all other submissions, including those leveraging more traditional robotic control techniques, and is the first pure learning-based method to solve this challenge.