Randomized-to-Canonical Model Predictive Control for Real-world Visual Robotic Manipulation

TL;DR

This paper introduces KRC-MPC, a novel zero-shot sim-to-real visual MPC framework that leverages a Kalman-based model to extract task-relevant features from randomized images, reducing real-world data collection efforts.

Contribution

The paper presents a new KRC-model framework and KRC-MPC method enabling zero-shot transfer of visual MPC from simulation to real-world tasks, significantly reducing human effort.

Findings

KRC-MPC successfully transfers from simulation to real-world tasks.

The method achieves effective control in valve rotation and block mating tasks.

Experimental results demonstrate zero-shot applicability across various domains.

Abstract

Many works have recently explored Sim-to-real transferable visual model predictive control (MPC). However, such works are limited to one-shot transfer, where real-world data must be collected once to perform the sim-to-real transfer, which remains a significant human effort in transferring the models learned in simulations to new domains in the real world. To alleviate this problem, we first propose a novel model-learning framework called Kalman Randomized-to-Canonical Model (KRC-model). This framework is capable of extracting task-relevant intrinsic features and their dynamics from randomized images. We then propose Kalman Randomized-to-Canonical Model Predictive Control (KRC-MPC) as a zero-shot sim-to-real transferable visual MPC using KRC-model. The effectiveness of our method is evaluated through a valve rotation task by a robot hand in both simulation and the real world, and a block mating task in simulation. The experimental results show that KRC-MPC can be applied to various real domains and tasks in a zero-shot manner.