Where2Act: From Pixels to Actions for Articulated 3D Objects

TL;DR

This paper introduces a novel approach to predict localized actionable regions on articulated 3D objects from images and depth data, enabling robots to interact meaningfully with objects like drawers by understanding possible elementary actions.

Contribution

It presents new neural network architectures for pixel-level action prediction on articulated objects, trained via a learning-from-interaction framework in simulation that generalizes across categories.

Findings

Network accurately predicts possible actions at each pixel.

Method generalizes across different object categories.

Framework enables simulation-based training for real-world application.

Abstract

One of the fundamental goals of visual perception is to allow agents to meaningfully interact with their environment. In this paper, we take a step towards that long-term goal -- we extract highly localized actionable information related to elementary actions such as pushing or pulling for articulated objects with movable parts. For example, given a drawer, our network predicts that applying a pulling force on the handle opens the drawer. We propose, discuss, and evaluate novel network architectures that given image and depth data, predict the set of actions possible at each pixel, and the regions over articulated parts that are likely to move under the force. We propose a learning-from-interaction framework with an online data sampling strategy that allows us to train the network in simulation (SAPIEN) and generalizes across categories. Check the website for code and data release: https://cs.stanford.edu/~kaichun/where2act/