RAE-NWM: Navigation World Model in Dense Visual Representation Space

TL;DR

This paper introduces RAE-NWM, a novel navigation world model that operates in a dense visual feature space, improving structural stability and control in visual navigation tasks.

Contribution

The paper proposes RAE-NWM, a new model that leverages dense visual features and a diffusion transformer to enhance navigation accuracy and stability.

Findings

Dense DINOv2 features have strong linear predictability for transitions.

Modeling in dense feature space improves structural stability.

Enhanced navigation performance in complex environments.

Abstract

Visual navigation requires agents to reach goals in complex environments through perception and planning. World models address this task by simulating action-conditioned state transitions to predict future observations. Current navigation world models typically learn state evolution under actions within the compressed latent space of a Variational Autoencoder, where spatial compression often discards fine-grained structural information and hinders precise control. To better understand the propagation characteristics of different representations, we conduct a linear dynamics probe and observe that dense DINOv2 features exhibit stronger linear predictability for action-conditioned transitions. Motivated by this observation, we propose the Representation Autoencoder-based Navigation World Model (RAE-NWM), which models navigation dynamics in a dense visual representation space. We employ a Conditional Diffusion Transformer with Decoupled Diffusion Transformer head (CDiT-DH) to model continuous transitions, and introduce a separate time-driven gating module for dynamics conditioning to regulate action injection strength during generation. Extensive evaluations show that modeling sequential rollouts in this space improves structural stability and action accuracy, benefiting downstream planning and navigation.