Label-Free Long-Horizon 3D UAV Trajectory Prediction via Motion-Aligned RGB and Event Cues
Hanfang Liang, Shenghai Yuan, Member IEEE, Fen Liu, Yizhuo Yang, Bing Wang, Zhuyu Huang, Chenyang Shi, Jing Jin · 2025
Read Label-Free Long-Horizon 3D UAV Trajectory Prediction via Motion-Aligned RGB and Event Cues by Hanfang Liang, Shenghai Yuan, Member IEEE, Fen Liu, Yizhuo Yang, Bing Wang, Zhuyu Huang, Chenyang Shi, Jing Jin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-03365.txt.
Description
Hanfang Liang and seven co-authors forecast where a drone will fly seconds ahead, without hand-labeled training data. Small drones are nimble and erratic, and counter-drone work needs their future paths, not just current spots. The method first pulls drone tracks out of raw laser-scanner clouds with an unsupervised routine. It then matches those tracks to camera images by motion agreement, creating trustworthy stand-in labels. A predictor blends motion physics with a visual network and trains itself against those labels.
On a hard public dataset with wide-view sensors and urban flight, the system beat supervised rivals using images or sound. Its five-second-ahead position error ran about forty percent lower, despite using no manual three-dimensional labels. The authors pitch the system as a cheap, scalable option for live airspace defense. The catalog creator field carries stray membership text after one name.
Catalog record
| CREATOR | Hanfang Liang, Shenghai Yuan, Member IEEE, Fen Liu, Yizhuo Yang, Bing Wang, Zhuyu Huang, Chenyang Shi, Jing Jin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-03365