DETROIT: Data Collection, Translation and Sharing for Rapid Vehicular App Development

Abstract

Developing vehicular applications is bottlenecked by the fragmentation of in-vehicle data: each car model uses proprietary CAN message formats, making it nearly impossible to write portable apps that work across vehicle brands and model years. DETROIT is an open-source framework that automates the collection, translation, and sharing of vehicular CAN data.

DETROIT provides a vehicle-agnostic abstraction layer — app developers program against standardized signals (speed, RPM, fuel level) while DETROIT handles the per-vehicle translation under the hood. It also includes a crowdsourcing component that continuously expands coverage to new vehicle models through community contributions.

Key Contributions

• Open-source, vehicle-agnostic framework for CAN data collection and translation.
• Crowdsourced database of CAN signal mappings, continuously expanding with new vehicles.
• Standardized API enables portable vehicular apps across brands and model years.
• Accelerates research and commercial development of automotive applications.

BibTeX

@inproceedings{pese2022detroit,
  title     = {DETROIT: Data Collection, Translation and Sharing for Rapid Vehicular App Development},
  author    = {Pes\'{e}, Mert D. and Chen, Dongyao and Campos, C. Andr\'{e}s and Ying, Alice and Stacer, Troy and Shin, Kang G.},
  booktitle = {Proceedings of the IEEE International Conference on Sensing, Communication, and Networking (SECON)},
  year      = {2022}
}