Unmanned aerial vehicles often rely on GNSS for positioning, but that can fail in contested or obstructed environments. A new paper on arXiv proposes ScalLoc, a method that uses thermal imaging to geo-localize a UAV by matching oblique thermal images to a 2D satellite map. Because thermal cameras capture heat signatures rather than visible light, the approach is said to be robust to illumination changes.

The abstract frames this as cross-modal thermal geo-localization (TG), which it describes as an illumination-robust positioning solution for GNSS-denied environments. It notes that existing thermal geo-localization methods have limitations, though the abstract is cut off before detailing them. ScalLoc appears to be a continuous approach, meaning it likely provides ongoing localization rather than a one-time fix, but the abstract does not specify the technical implementation.

The paper is listed as arXiv:2610.14866v1 with a new announcement type. Since only the abstract is available, the specific contributions, experimental results, and comparisons with prior work are not described in the source. Readers interested in the full method will need to consult the paper.