Driver Workload as a Function of Road Geometry: A Pilot Experiment

Paul Green, Bernice Lin, and Tandi Bagian

September, 1994

This experiment examined the relationship between road geometry, workload ratings, and predictions from Wierwille's workload model, and provided representative performance data for driving a simulator. Eight people drove the UMTRI driving simulator for six sets of roads (varying in sight distance from 150 to 1140 feet (46 to 347 m)), each set having a fixed sequence of five winding road segments of varying width (lanes of 7.5 to 12 feet (2.3 to 3.7 m)). The six sets were presented in a counterbalanced order. Traffic was not shown.

The standard deviation of lateral position was 1.1 feet, with drivers using more of the road as lane width increased (0.9 feet (.27 m) for 7.5 foot lanes to 1.4 feet (.43 m) for 12-foot lanes). In general, lane variance decreased slightly with increasing sight distance, except for the 150-foot sight distance.

In terms of predicting performance, the mean and standard deviation of speed were not correlated with workload ratings, Wierwille workload estimates, or other performance measures. The standard deviation of lateral position was correlated with workload ratings. These data suggest that use of the standard deviation of speed and lateral position as measures of driving workload merits reconsideration when workload is light and traffic is absent.

Close This Window