Safety Research at the Data-Supported Transportation Operations and Planning (D-STOP) UTC

Safety Research at the Data-Supported Transportation Operations and Planning (D-STOP) UTCDr. Jen DuthieMarch 19, 2015COLLABORATE. INNOVATE. EDUCATE.Crash PredictionSpatial multivariate count model to jointly analyze traffic-related counts of pedestrians and bicyclists by injury severity

Source: http://crashstat.org/

COLLABORATE. INNOVATE. EDUCATE.Predict injury counts at Census tract level, based on data from Manhattan (via CrashStat website)2Crash PredictionPredicting crash frequency at urban intersections with a count data model with endogenous covariates

Source: hotpads.comCOLLABORATE. INNOVATE. EDUCATE.Irving Texas3Vehicle Tracking with Fine-Grained GPS

COLLABORATE. INNOVATE. EDUCATE.- Consumer-grade Global Navigation Satellite System (GNSS) receiver accuracy has stagnated at 2-3 metersOur device is solar-powered and wirelessly networked so that it can be mounted to the host vehicle without the need for invasive connections. The devices GPS processing is implemented on a software-defined GPS receiver that runs on a low-cost off-the-shelf computer module. The total cost of each prototype unit, including computer module, GPS antenna, battery, solar-powered charger, and miscellaneous electronics, is less than $1000. *Why we may want our vehicles to know their position at sub-decimeter accuracy:safety and security for connected veh. Our quantitative measure of passenger comfort will be based, among other correlates, on:Acceleration and deceleration profilesCornering styleLane change frequencyGeneral smoothness in transitionOur analysis of driver safety will discover risk signatures - small-scale features of driving behavior that are strongly correlated with driving consistency and driver safety. This includes lane preference, lane keeping precision, path repeatability, respect for the stop bar at intersections, cornering style, etc. Such signatures will make it possible to evaluate an unknown driver's safety risk after only a few weeks of monitoring.

4Autonomous Vehicle Intersection Modeling

COLLABORATE. INNOVATE. EDUCATE.Thank you.

jduthie@mail.utexas.eduCOLLABORATE. INNOVATE. EDUCATE.