SculptBot: Pre-Trained Models for 3D Deformable Object Manipulation
This addresses the challenge of robotic sculpting for materials with plastic behavior, but it is incremental as it builds on existing point cloud and Transformer methods.
The paper tackles the problem of robotic manipulation of deformable objects like clay by developing a system that uses point clouds and a pre-trained Transformer to learn a latent dynamics model for predicting material deformations, achieving successful creation of simple shapes in real-world experiments.
Deformable object manipulation presents a unique set of challenges in robotic manipulation by exhibiting high degrees of freedom and severe self-occlusion. State representation for materials that exhibit plastic behavior, like modeling clay or bread dough, is also difficult because they permanently deform under stress and are constantly changing shape. In this work, we investigate each of these challenges using the task of robotic sculpting with a parallel gripper. We propose a system that uses point clouds as the state representation and leverages pre-trained point cloud reconstruction Transformer to learn a latent dynamics model to predict material deformations given a grasp action. We design a novel action sampling algorithm that reasons about geometrical differences between point clouds to further improve the efficiency of model-based planners. All data and experiments are conducted entirely in the real world. Our experiments show the proposed system is able to successfully capture the dynamics of clay, and is able to create a variety of simple shapes.