Embodied Visuomotor Representation

TL;DR

This paper introduces Embodied Visuomotor Representation, a method enabling robots and agents to infer distances based on action, allowing quick adaptation to unknown environments without external calibration.

Contribution

It presents a novel approach for inferring spatial information through action-based embodiment, bypassing the need for precise external calibration in robotics and simulation.

Findings

Robots can learn to touch obstacles within seconds without external scale knowledge.

Agents can jump across unknown gaps after minimal test oscillations.

Behavior mimics natural strategies of animals like bees and gerbils.

Abstract

Imagine sitting at your desk, looking at objects on it. You do not know their exact distances from your eye in meters, but you can immediately reach out and touch them. Instead of an externally defined unit, your sense of distance is tied to your action's embodiment. In contrast, conventional robotics relies on precise calibration to external units, with which vision and control processes communicate. We introduce Embodied Visuomotor Representation, a methodology for inferring distance in a unit implied by action. With it a robot without knowledge of its size, environmental scale, or strength can quickly learn to touch and clear obstacles within seconds of operation. Likewise, in simulation, an agent without knowledge of its mass or strength can successfully jump across a gap of unknown size after a few test oscillations. These behaviors mirror natural strategies observed in bees and gerbils, which also lack calibration in an external unit.