Learning Closed-loop Dough Manipulation Using a Differentiable Reset Module

TL;DR

This paper introduces a novel differentiable reset module for trajectory optimization, enabling effective closed-loop manipulation of deformable objects like dough, with successful real-world application and transfer from simulation.

Contribution

It presents a new trajectory optimizer with a differentiable reset module and trains a closed-loop policy for dough manipulation from RGB-D data.

Findings

Successful flattening of dough into target shapes in real-world experiments

Effective transfer of the policy from simulation to real-world scenarios

Outperforms naive trajectory optimization methods

Abstract

Deformable object manipulation has many applications such as cooking and laundry folding in our daily lives. Manipulating elastoplastic objects such as dough is particularly challenging because dough lacks a compact state representation and requires contact-rich interactions. We consider the task of flattening a piece of dough into a specific shape from RGB-D images. While the task is seemingly intuitive for humans, there exist local optima for common approaches such as naive trajectory optimization. We propose a novel trajectory optimizer that optimizes through a differentiable "reset" module, transforming a single-stage, fixed-initialization trajectory into a multistage, multi-initialization trajectory where all stages are optimized jointly. We then train a closed-loop policy on the demonstrations generated by our trajectory optimizer. Our policy receives partial point clouds as input, allowing ease of transfer from simulation to the real world. We show that our policy can perform real-world dough manipulation, flattening a ball of dough into a target shape.