Physics-Informed Diffusion Models for Vehicle Speed Trajectory Generation
Vadim Sokolov, Farnaz Behnia, Dominik Karbowski · 2026
Read Physics-Informed Diffusion Models for Vehicle Speed Trajectory Generation by Vadim Sokolov, Farnaz Behnia, Dominik Karbowski (2026) 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-2602-05028.txt.
Description
Testing vehicle control software needs realistic driving data, but collecting it on real roads is expensive. Vadim Sokolov and 2 co-authors generate synthetic speed profiles instead in this 2026 applied statistics paper. Older Markov chain generators leave discretization artifacts and cannot capture the variety of real trips.
Their method uses a diffusion model informed by physics, representing each trip as paired speed and acceleration channels with soft physical constraints. Trained on 6,367 GPS-derived micro-trips, a transformer-based model beat a convolutional rival at matching real distributions. A classifier could barely tell its output from real data, and the synthetic trips proved useful for downstream energy estimates. The authors present the approach as a scalable stand-in for field data in transport research.
Catalog record
| CREATOR | Vadim Sokolov, Farnaz Behnia, Dominik Karbowski |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-05028