PathPainter: Transferring the Generalization Ability of Image Generation Models to Embodied Navigation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wang, Yijin, Tian, Yuru, Huang, Xijie, Gai, Weiqi, Zhu, Mo, Zhou, Xin, Wu, Yuze, Gao, Fei
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910200868372480
author Wang, Yijin
Tian, Yuru
Huang, Xijie
Gai, Weiqi
Zhu, Mo
Zhou, Xin
Wu, Yuze
Gao, Fei
author_facet Wang, Yijin
Tian, Yuru
Huang, Xijie
Gai, Weiqi
Zhu, Mo
Zhou, Xin
Wu, Yuze
Gao, Fei
contents Bird's-eye-view (BEV) images have been widely demonstrated to provide valuable prior information for navigation. Given the global information provided by such views, two key challenges remain: how to fully exploit this information and how to reliably use it during execution. In this paper, we propose a navigation system that uses BEV images as global priors and is designed for ground and near-ground robotic platforms. The system employs an image generation model to interpret human intent from natural language, identify the target destination, and generate traversability masks. During execution, we introduce cross-view localization to align the robot's odometry with the BEV map and mitigate long-term drift in conventional odometry. We conduct extensive benchmark experiments to evaluate the proposed method and further validate it on a UAV platform. Using only a conventional local motion planner, the UAV successfully completes a 160-meter outdoor long-range navigation task. This work demonstrates how the world-understanding capabilities of foundation models can be transferred to embodied navigation, enabling robots to benefit from the strong generalization ability of existing image generation models.
format Preprint
id arxiv_https___arxiv_org_abs_2605_07496
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle PathPainter: Transferring the Generalization Ability of Image Generation Models to Embodied Navigation
Wang, Yijin
Tian, Yuru
Huang, Xijie
Gai, Weiqi
Zhu, Mo
Zhou, Xin
Wu, Yuze
Gao, Fei
Robotics
Bird's-eye-view (BEV) images have been widely demonstrated to provide valuable prior information for navigation. Given the global information provided by such views, two key challenges remain: how to fully exploit this information and how to reliably use it during execution. In this paper, we propose a navigation system that uses BEV images as global priors and is designed for ground and near-ground robotic platforms. The system employs an image generation model to interpret human intent from natural language, identify the target destination, and generate traversability masks. During execution, we introduce cross-view localization to align the robot's odometry with the BEV map and mitigate long-term drift in conventional odometry. We conduct extensive benchmark experiments to evaluate the proposed method and further validate it on a UAV platform. Using only a conventional local motion planner, the UAV successfully completes a 160-meter outdoor long-range navigation task. This work demonstrates how the world-understanding capabilities of foundation models can be transferred to embodied navigation, enabling robots to benefit from the strong generalization ability of existing image generation models.
title PathPainter: Transferring the Generalization Ability of Image Generation Models to Embodied Navigation
topic Robotics
url https://arxiv.org/abs/2605.07496