Summary

A driving scenario takes place in an environment. A human driver on a specific day in some certain weather condition, happens to be on the road. But there is more to that. Is it just the environment or the human that builds the driving scenario? There is also the factor of time. And yes, the driving scenario is an interaction of the environment, and human which takes place in time. When you pass by a bike or reach an intersection, or when you are driving in a specific weather condition, these all matter to your driving behaviors. In this project, we are working on building a context-aware framework for modeling driving scenarios to enhance future autonomous vehicle’s decision while preserving driver’s privacy. We take advantage of our naturalistic driving study platform to collect multi-modal data about both environment and humans. The collective summary of all sensor measurements in our data collection defines a context for a driving scenario. Effectively monitoring drivers and taking actions requires models that are trained on real-life driving scenarios. Additionally, such systems should take privacy of the drivers into account for enhanced acceptance.

A driving situation can be hierarchical, having multiple layers, and varying over time. For instance, a driving scenario as shown in the following video includes passing through an intersection, while following/passing a bike. We are interested to know, how these situational elements define our driving behaviors.

Arsalan Heydarian

Assistant Professor

Principal Investigator

Arash Tavakoli

PhD Student

Graduate Research Assistant

Xiang Guo

PhD Student

Graduate Research Assistant