WorldMAP: Bootstrapping Vision-Language Navigation Trajectory Prediction with Generative World Models

TL;DR

WorldMAP introduces a framework that leverages generative world models to produce supervision signals for vision-language navigation, significantly improving trajectory prediction accuracy.

Contribution

It presents a novel teacher-student approach that converts generated futures into semantic-spatial supervision for navigation learning.

Findings

Achieves the best ADE and FDE on Target-Bench, reducing errors by over 18% and 42%.

Lifts a small open-source VLM to competitive performance with proprietary models.

Demonstrates structured supervision from world models enhances embodied navigation.

Abstract

Vision-language models (VLMs) and generative world models are opening new opportunities for embodied navigation. VLMs are increasingly used as direct planners or trajectory predictors, while world models support look-ahead reasoning by imagining future views. Yet predicting a reliable trajectory from a single egocentric observation remains challenging. Current VLMs often generate unstable trajectories, and world models, though able to synthesize plausible futures, do not directly provide the grounded signals needed for navigation learning. This raises a central question: how can generated futures be turned into supervision for grounded trajectory prediction? We present WorldMAP, a teacher--student framework that converts world-model-generated futures into persistent semantic-spatial structure and planning-derived supervision. Its world-model-driven teacher builds semantic-spatial memory from generated videos, grounds task-relevant targets and obstacles, and produces trajectory pseudo-labels through explicit planning. A lightweight student with a multi-hypothesis trajectory head is then trained to predict navigation trajectories directly from vision-language inputs. On Target-Bench, WorldMAP achieves the best ADE and FDE among compared methods, reducing ADE by 18.0% and FDE by 42.1% relative to the best competing baseline, while lifting a small open-source VLM to DTW performance competitive with proprietary models. More broadly, the results suggest that, in embodied navigation, the value of world models may lie less in supplying action-ready imagined evidence than in synthesizing structured supervision for navigation learning.