Hierarchical visuomotor control of humanoids

TL;DR

This paper presents a hierarchical control architecture for humanoid robots that combines pre-trained low-level motor controllers with a high-level visual task controller, enabling flexible, task-specific behaviors in complex environments.

Contribution

It introduces a novel hierarchical visuomotor control system that integrates perception and motor skills for humanoids, with pre-trained sub-policies and a high-level controller for task switching.

Findings

Effective control of a high-DoF humanoid in simulation

Successful integration of visual perception during locomotion

Demonstrated flexible, task-directed motor behavior

Abstract

We aim to build complex humanoid agents that integrate perception, motor control, and memory. In this work, we partly factor this problem into low-level motor control from proprioception and high-level coordination of the low-level skills informed by vision. We develop an architecture capable of surprisingly flexible, task-directed motor control of a relatively high-DoF humanoid body by combining pre-training of low-level motor controllers with a high-level, task-focused controller that switches among low-level sub-policies. The resulting system is able to control a physically-simulated humanoid body to solve tasks that require coupling visual perception from an unstabilized egocentric RGB camera during locomotion in the environment. For a supplementary video link, see https://youtu.be/7GISvfbykLE .