Flying in Clutter on Monocular RGB by Learning in 3D Radiance Fields with Domain Adaptation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Huang, Xijie, Li, Jinhan, Wu, Tianyue, Zhou, Xin, Han, Zhichao, Gao, Fei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908722097291264
author Huang, Xijie
Li, Jinhan
Wu, Tianyue
Zhou, Xin
Han, Zhichao
Gao, Fei
author_facet Huang, Xijie
Li, Jinhan
Wu, Tianyue
Zhou, Xin
Han, Zhichao
Gao, Fei
contents Modern autonomous navigation systems predominantly rely on lidar and depth cameras. However, a fundamental question remains: Can flying robots navigate in clutter using solely monocular RGB images? Given the prohibitive costs of real-world data collection, learning policies in simulation offers a promising path. Yet, deploying such policies directly in the physical world is hindered by the significant sim-to-real perception gap. Thus, we propose a framework that couples the photorealism of 3D Gaussian Splatting (3DGS) environments with Adversarial Domain Adaptation. By training in high-fidelity simulation while explicitly minimizing feature discrepancy, our method ensures the policy relies on domain-invariant cues. Experimental results demonstrate that our policy achieves robust zero-shot transfer to the physical world, enabling safe and agile flight in unstructured environments with varying illumination.
format Preprint
id arxiv_https___arxiv_org_abs_2512_17349
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Flying in Clutter on Monocular RGB by Learning in 3D Radiance Fields with Domain Adaptation
Huang, Xijie
Li, Jinhan
Wu, Tianyue
Zhou, Xin
Han, Zhichao
Gao, Fei
Robotics
Modern autonomous navigation systems predominantly rely on lidar and depth cameras. However, a fundamental question remains: Can flying robots navigate in clutter using solely monocular RGB images? Given the prohibitive costs of real-world data collection, learning policies in simulation offers a promising path. Yet, deploying such policies directly in the physical world is hindered by the significant sim-to-real perception gap. Thus, we propose a framework that couples the photorealism of 3D Gaussian Splatting (3DGS) environments with Adversarial Domain Adaptation. By training in high-fidelity simulation while explicitly minimizing feature discrepancy, our method ensures the policy relies on domain-invariant cues. Experimental results demonstrate that our policy achieves robust zero-shot transfer to the physical world, enabling safe and agile flight in unstructured environments with varying illumination.
title Flying in Clutter on Monocular RGB by Learning in 3D Radiance Fields with Domain Adaptation
topic Robotics
url https://arxiv.org/abs/2512.17349