PaperBot: Learning to Design Real-World Tools Using Paper

Abstract

Paper is a cheap, recyclable, and clean material that is often used to make practical tools. Traditional tool design either relies on simulation or physical analysis, which is often inaccurate and time-consuming. In this paper, we propose PaperBot, an approach that directly learns to design and use a tool in the real world using paper without human intervention.

We demonstrated the effectiveness and efficiency of PaperBot on two tool design tasks: 1. learning to fold and throw paper airplanes for maximum travel distance 2. learning to cut paper into grippers that exert maximum gripping force. We present a self-supervised learning framework that learns to perform a sequence of folding, cutting, and dynamic manipulation actions in order to optimize the design and use of a tool. We deploy our system to a real-world two-arm robotic system to solve challenging design tasks that involve aerodynamics (paper airplane) and friction (paper gripper) that are impossible to simulate accurately.

Video

Learning Curves Comparison

We ran experiments directly in the real world for our method and each of the baseline methods. Here shows the performance per optimization iteration for both tasks, designed by our method (blue) versus baselines.

Method

Data-efficient optimization framework based on neural surrogate model and epsilon-greedy exploration.

Learning happens 100% in the real world without any simulation. Rendering is for visualization only.

Learning Process

Top: paper airplane landing position periteration. Bottom: 3-hour experimental videos captured by 2 different cameras

Automation

We automate the process of folding and throwing paper airplanes using a two-arm robotic system and a cheap Injket printer.

Fast Adaptation

Our approach can learn to optimize the Kirigami gripper design for objects of different sizes with only 50 trials.

BibTeX

@misc{liu2024paperbot,
  title={PaperBot: Learning to Design Real-World Tools Using Paper}, 
  author={Ruoshi Liu and Junbang Liang and Sruthi Sudhakar and Huy Ha and Cheng Chi and Shuran Song and Carl Vondrick},
  year={2024},
  eprint={2403.09566},
  archivePrefix={arXiv},
  primaryClass={cs.RO}
}

PaperBot

Learning to Design Real-World Tools Using Paper

PaperBot learns in real-world settings to create and utilize paper-based tools for practical tasks.