Generating High-Accuracy Transportation Datasets with Unmanned Aerial Vehicles

Generating High-Accuracy Transportation Datasets with Unmanned Aerial Vehicles

Gabor Orosz Headshot. The link directs to their profile page.
Gabor Orosz
The University of Michigan Logo. The link directs to the funded research led by this institution.

Principal Investigator(s):

Gabor Orosz, Associate Professor of Mechanical Engineering – University of Michigan
Associate Professor of Civil and Environmental Engineering – University of Michigan

Project Abstract:
A new trend in the area of connected and automated vehicles (CAVs) is infrastructure-based sensing and tracking. By installing roadside cameras at the infrastructure, one can monitor the overall traffic situation and, after processing the collected data via edge computing, the information can be shared with CAVs via V2X connectivity. This has a huge potential to improve traffic safety and efficiency. We identify two main challenges to deploying such systems: (i) to obtain high-precision data and (ii) to label the obtained data. It is necessary to have sufficient high-precision labeled datasets for training the underlying machine learning (ML) algorithms to detect, identify, localize, and track road participants. In other words, one needs to know the “ground truth” (with high precision) for a large variety of different scenarios. As of now, this may be done by hand-labeling images, which is immensely labor intensive, or by using probe vehicles equipped with high-precision GPS, which can only provide data about a few specific vehicles.

Institution(s): University of Michigan – Ann Arbor

Award Year: 2023

Research Focus: Safety, Mobility, Equity

Project Form(s):