Active World Model Learning with Progress Curiosity

TL;DR

This paper introduces a curiosity-driven Active World Model Learning system using a novel $ ext{ extgamma}$-Progress signal, enabling an agent to explore and learn complex dynamics in a 3D environment more effectively than existing methods.

Contribution

The paper proposes a new $ ext{ extgamma}$-Progress curiosity signal for AWML, improving exploration efficiency and model learning in complex environments.

Findings

$ ext{ extgamma}$-Progress guides balanced exploration of learnable dynamics.

The proposed controller outperforms RND and Model Disagreement in AWML tasks.

Enhanced learning of environment dynamics and agent behavior patterns.

Abstract

World models are self-supervised predictive models of how the world evolves. Humans learn world models by curiously exploring their environment, in the process acquiring compact abstractions of high bandwidth sensory inputs, the ability to plan across long temporal horizons, and an understanding of the behavioral patterns of other agents. In this work, we study how to design such a curiosity-driven Active World Model Learning (AWML) system. To do so, we construct a curious agent building world models while visually exploring a 3D physical environment rich with distillations of representative real-world agents. We propose an AWML system driven by $\gamma$-Progress: a scalable and effective learning progress-based curiosity signal. We show that $\gamma$-Progress naturally gives rise to an exploration policy that directs attention to complex but learnable dynamics in a balanced manner, thus overcoming the "white noise problem". As a result, our $\gamma$-Progress-driven controller achieves significantly higher AWML performance than baseline controllers equipped with state-of-the-art exploration strategies such as Random Network Distillation and Model Disagreement.