Development of a statistical method for predicting human driver decisions.

PDFTech ReportATLAS-2015-06DTRT13-G-UTC54DriversDecision makingBehaviorKinematicsMathematical predictionAutomatic data collection systemsIntersectionsUnited StatesCenter for Advancing Transportation Leadership and Safety (ATLAS Center)Tan, Yaoyan VincentElliott, MichaelFlannagan, CarolUniversity of Michigan. Transportation Research InstituteUnited States. Dept. of Transportation. Office of the Assistant Secretary for Research and TechnologyUS Transportation CollectionAs autonomous vehicles enter the fleet, there will be a long period when these vehicles will have to interact withhuman drivers. One of the challenges for autonomous vehicles is that human drivers do not communicate theirdecisions well. However, the kinematic behavior of a human-driven vehicle may be a good predictor of driver intentwithin a short time frame. We analyzed the kinematic time-series data (e.g., speed) for a set of drivers making leftturns at intersections to predict whether the driver would stop before executing the turn or not. We used principalcomponents analysis (PCA) to generate independent dimensions that explain the variation in vehicle speed before aturn. These dimensions remained relatively consistent throughout the maneuver, allowing us to compute independentscores on these dimensions for different time windows throughout the approach to the intersection. We then linkedthese PCA scores to whether a driver would stop before executing a left turn using the Bayesian additive regressiontrees (BART). Our model achieved an area under the receiver operating characteristic curve (AUC) of more than 0.90by -25m away from the center of an intersection