AssembleRL: Learning to Assemble Furniture from Their Point Clouds

TL;DR

This paper introduces AssembleRL, a reinforcement learning approach that assembles furniture from point clouds with minimal supervision by using a novel reward signal based on incorrectness and incompleteness measures.

Contribution

It proposes a new reward mechanism for furniture assembly from point clouds that reduces the need for human-labeled connection data.

Findings

Successfully trains deep networks for furniture assembly

Effective in assembling various furniture types

Reduces supervision requirements

Abstract

The rise of simulation environments has enabled learning-based approaches for assembly planning, which is otherwise a labor-intensive and daunting task. Assembling furniture is especially interesting since furniture are intricate and pose challenges for learning-based approaches. Surprisingly, humans can solve furniture assembly mostly given a 2D snapshot of the assembled product. Although recent years have witnessed promising learning-based approaches for furniture assembly, they assume the availability of correct connection labels for each assembly step, which are expensive to obtain in practice. In this paper, we alleviate this assumption and aim to solve furniture assembly with as little human expertise and supervision as possible. To be specific, we assume the availability of the assembled point cloud, and comparing the point cloud of the current assembly and the point cloud of the target product, obtain a novel reward signal based on two measures: Incorrectness and incompleteness. We show that our novel reward signal can train a deep network to successfully assemble different types of furniture. Code and networks available here: https://github.com/METU-KALFA/AssembleRL