Graduate Student Researcher
Brian Phegley received his BS from UC Berkeley in 2010 in mechanical engineering. He is currently pursuing a PhD in mechanical engineering, with a focus on traffic control and optimization.
Current research: Fault Detection of Freeway Loop Detectors
In the construction of traffic models for use in planning and prediction, a source of quantitative data on the ground is needed for calibration and verification purposes. The most common source of consistent data for freeway models are loop detectors that regularly update information; however, such loop detectors have noise errors and biases that can make model fitting unfeasible. Direct application of such data will lead to faulty and poor models. Before this time, analysis needed to be made by hand to identify faulty detection. The current research is developing more automated methods to make such identification by statistically comparing results from model simulations to real data and flagging locations of significant error. Currently, a functional algorithm has been developed to consider mainline freeway data, and tests are being made to an algorithm being considered for on and off ramp identification. This project is joint work with Professor Roberto Horowitz.