Advances in Driving Research - Models, Data, and New Methods

Abstract:

The current project, collaboration between the MIT and the University of Michigan, concerns the development of complex models to describe driver decision making at intersections. MIT is focused on modeling and Michigan on data collection and method development. Two experiments have been conducted so far (24 subjects/experiment) using a NADS MiniSim driving simulator for which extensive programming was required. Subjects drove through 2 sets of 70 intersections following a lead vehicle (and being followed). At each intersection, the signal was green, yellow (3 time options), or red. For the yellow signals, vehicles at the intersection could be (1) stationary, (2) intrude into the intersection, or (3) turn in front of them unexpectedly. In experiment 2, augmented reality warnings were provided. While driving, the speed, acceleration, gaps, and lane positions for all vehicles were recorded, and in the latest experiment, eye fixations for younger drivers.

MIT is analyzing these studies and the models resulting are described in the MIT abstract. In addition, there were numerous significant advances in the methods for studying driving that resulted as a consequence of this project.

1. SAE Recommended Practice J2944, which defines driving performance measures and statistics has been revised and approved.
2. Very low cost eye fixations systems (less than $500) were interfaced to the driving simulator.
3. Very low cost driving simulations (free simulator, $250 in hardware) were developed for related educational activities
4. Improved design rules for urban driving scenarios were developed.