A Perceived Environment Design using a Multi-Modal Variational Autoencoder for learning Active-Sensing

TL;DR

This paper introduces a multi-modal variational autoencoder framework for designing perceived environments that enable agents to learn active-sensing behaviors, and compares it with curiosity-driven learning approaches.

Contribution

It presents a novel integration of multi-modal variational autoencoders with environment design for active-sensing, highlighting differences from curiosity-driven methods.

Findings

Demonstrates effective environment design for active-sensing

Shows comparative advantages over curiosity-driven learning

Provides insights into perception-action coupling

Abstract

This contribution comprises the interplay between a multi-modal variational autoencoder and an environment to a perceived environment, on which an agent can act. Furthermore, we conclude our work with a comparison to curiosity-driven learning.