WiFi-based Driving Activity Recognition in Vehicles

With the increasing requirements for smart vehicles in recent years, driving activity recognition in vehicles becomes a promising direction given its unique advantages in safe driving and human-car interaction. It can not only alert the driver when there exist distracted driving behaviors but also enable novel human-car interactions with gesture-based control. Most state-of-the-art systems use camera-based approaches for activity recognition, which highly rely on enough visible light and have a high risk of user privacy leakage. We propose that WiFi signals can be leveraged for driving activity recognition in vehicles as WiFi is replacing Bluetooth for in-car deployment. The objective of this research is to develop a WiFi-based driving activity recognition framework using advanced signal processing and deep learning techniques for driving assistance applications. This research is closely related to PANOMOTION TECHNOLOGY INC., a local startup working on intelligent driving sensing and assistance. Our research can greatly benefit the company by applying the research outcomes to its main products.

Faculty Supervisor:

Jane Wang

Student:

Fangxin Wang

Partner:

PanoMotion Technology Inc

Discipline:

Engineering - computer / electrical

Sector:

Professional, scientific and technical services

University:

University of British Columbia