Emergence of Spatial Coordinates via Exploration

TL;DR

This paper demonstrates that a naive agent can autonomously develop an internal spatial coordinate system by learning sensorimotor predictions, without prior knowledge of external space or predefined models.

Contribution

It introduces a self-supervised method enabling agents to discover spatial coordinates autonomously, aligning internal representations with external space.

Findings

Agents can build internal coordinate systems through sensorimotor learning.

The learned coordinates match the external space in dimension and metric properties.

The approach requires no prior spatial knowledge or predefined models.

Abstract

Spatial knowledge is a fundamental building block for the development of advanced perceptive and cognitive abilities. Traditionally, in robotics, the Euclidean (x,y,z) coordinate system and the agent's forward model are defined a priori. We show that a naive agent can autonomously build an internal coordinate system, with the same dimension and metric regularity as the external space, simply by learning to predict the outcome of sensorimotor transitions in a self-supervised way.