Catch & Carry: Reusable Neural Controllers for Vision-Guided Whole-Body Tasks

TL;DR

This paper presents a neural-network based framework for creating flexible, realistic humanoid controllers capable of performing diverse whole-body tasks with object interactions, using physics-based simulation and active sensing.

Contribution

It introduces an integrated approach combining motor primitives, demonstrations, and reinforcement learning for versatile humanoid control in realistic environments.

Findings

Controllers successfully perform goal-conditioned tasks like box carrying and ball catching.

The approach demonstrates behavioral robustness across various tasks.

Controllers can operate in real-time on standard hardware.

Abstract

We address the longstanding challenge of producing flexible, realistic humanoid character controllers that can perform diverse whole-body tasks involving object interactions. This challenge is central to a variety of fields, from graphics and animation to robotics and motor neuroscience. Our physics-based environment uses realistic actuation and first-person perception -- including touch sensors and egocentric vision -- with a view to producing active-sensing behaviors (e.g. gaze direction), transferability to real robots, and comparisons to the biology. We develop an integrated neural-network based approach consisting of a motor primitive module, human demonstrations, and an instructed reinforcement learning regime with curricula and task variations. We demonstrate the utility of our approach for several tasks, including goal-conditioned box carrying and ball catching, and we characterize its behavioral robustness. The resulting controllers can be deployed in real-time on a standard PC. See overview video, https://youtu.be/2rQAW-8gQQk .