125. ROLE OF HEAD DYNAMICS IN HUMAN-CENTERED ACTIVE SAFETY SYSTEMS: COMPARATIVE ANALYSIS OF VISION BASED APPROACHES

Department: Electrical & Computer Engineering
Faculty Advisor(s): Mohan Trivedi

Primary Student
Name: Sujitha Catherine Martin
Email: scmartin@ucsd.edu
Phone: 858-822-0002
Grad Year: 2015

Abstract
Since the advent of cars, safety of human beings inside and outside the vehicle has become an increasing concern. Many passive systems such as seat belts, air bags and laminated windshields, and non-human-centered active systems such as anti-lock brakes, lane departure warning and adaptive cruise control have indeed improved safety of humans. However, only recently have researchers incorporated the state of the driver in active safety systems. Among many parts of the body one can monitor, the human head in and of itself gives a wealth of information on driver behavior. Many studies have already shown the importance of using head dynamics to predict lane changes, to understand a driver's focus of attention, to infer state of alertness, etc. In this study, we further study head dynamics in the context of mulimodal alerts and distraction. We use techniques likes faceLAB and optical flow, to not only evaluate how the estimated head pose of these techniques can be used to infer head gestures, but also to study the behavior of a driver's head in the presence of multimodal alerts and distraction.

« Back to Posters or Search Results