Road to Vienna!
We are excited to announce that our work "Thermal Image Refinement with Depth Estimation using Recurrent Networks for Monocular ORB-SLAM3" has been accepted to ICRA 2026!
In this work, we propose using a monocular thermal camera on a UAV for real-time depth estimation and SLAM in GPS-denied, visually degraded environments. Our approach uses a lightweight supervised network with recurrent blocks to capture temporal dependencies and improve robustness.
Experimental results on datasets and real-time UAV flights demonstrate competitive depth accuracy and robust SLAM performance under low-light conditions.
Finally, our paper also provides a non-radiometric thermal–depth UAV dataset to evaluate our framework.
Project page: https://lnkd.in/ekBcqBFv
Video: https://lnkd.in/eNTVbjC7
