In-Hand Cube Reconfiguration: Simplified

TL;DR

This paper introduces a simplified yet effective approach to in-hand cube reconfiguration, combining classical planning with environmental constraints and compliant end-effectors, outperforming complex deep learning systems.

Contribution

It demonstrates that a simplified, classical planning-based method can achieve robust in-hand cube reconfiguration, challenging the reliance on complex deep learning approaches.

Findings

Simplified approach outperforms deep learning systems in cube reconfiguration

Combining GOFAI planning with environmental constraints enhances dexterous manipulation

Inherent compliance and simplified perception improve robustness

Abstract

We present a simple approach to in-hand cube reconfiguration. By simplifying planning, control, and perception as much as possible, while maintaining robust and general performance, we gain insights into the inherent complexity of in-hand cube reconfiguration. We also demonstrate the effectiveness of combining GOFAI-based planning with the exploitation of environmental constraints and inherently compliant end-effectors in the context of dexterous manipulation. The proposed system outperforms a substantially more complex system for cube reconfiguration based on deep learning and accurate physical simulation, contributing arguments to the discussion about what the most promising approach to general manipulation might be. Project website: https://rbo.gitlab-pages.tu-berlin.de/robotics/simpleIHM/