How to Close Sim-Real Gap? Transfer with Segmentation!

TL;DR

This paper presents a segmentation-based transfer method to bridge the sim-real gap in robotic grasping, using domain-invariant perception and control policies trained in simulation and adapted to real environments.

Contribution

It introduces a novel segmentation interface for perception-control transfer, combining closed-loop control and perception learning without real-world supervision.

Findings

Achieved 88% success rate in real robot grasping tasks.

Demonstrated robustness to dynamic discrepancies and unseen scenarios.

Enabled transfer of control policies from simulation to real robots.

Abstract

One fundamental difficulty in robotic learning is the sim-real gap problem. In this work, we propose to use segmentation as the interface between perception and control, as a domain-invariant state representation. We identify two sources of sim-real gap, one is dynamics sim-real gap, the other is visual sim-real gap. To close dynamics sim-real gap, we propose to use closed-loop control. For complex task with segmentation mask input, we further propose to learn a closed-loop model-free control policy with deep neural network using imitation learning. To close visual sim-real gap, we propose to learn a perception model in real environment using simulated target plus real background image, without using any real world supervision. We demonstrate this methodology in eye-in-hand grasping task. We train a closed-loop control policy model that taking the segmentation as input using simulation. We show that this control policy is able to transfer from simulation to real environment. The closed-loop control policy is not only robust with respect to discrepancies between the dynamic model of the simulated and real robot, but also is able to generalize to unseen scenarios where the target is moving and even learns to recover from failures. We train the perception segmentation model using training data generated by composing real background images with simulated images of the target. Combining the control policy learned from simulation with the perception model, we achieve an impressive $\bf{88\%}$ success rate in grasping a tiny sphere with a real robot.