Event details

Driving vehicles is a dynamic activity that exposes passengers to rapidly changing driving scenes. These scenes can be characterized by, e.g., the weather, the traffic state, the road conditions, the daytime, and the land-use around the road (residential, industrial, urban, or forest). With the dawn of the era of AI, where intelligent systems will become ubiquitously available, users will increasingly expect in-vehicle systems to adapt to their needs and state. Thus, to understand them and augment interactions or cabin states, it is crucial for such systems to understand the context passengers are exposed to.

In this talk, Philipp will present answers to questions of how to capture, encode, and use driving context in downstream systems to create novel driver experiences. He will present how driving context influences drivers’ ability to handle interruptions such as phone calls, and present a privacy-aware approach to assessing driver interruptibility based on this context. Further, he will present RouteLLM, the first Large Language model with native driving context understanding that enables context-aware reasoning.

Event hosted by Dr. Mohamed Kari at Princeton HCI.