How standard deviation of lane position is affected by driving environment and driver characteristics: implications...

The objective of this analysis was to use pooled driving performance data to characterize normal lane keeping driving performance as measured by standard deviation of lane position (SDLP) and lane crossings. There has been a lack of consistency in the magnitude of baseline SDLP reported in results from different studies in the published literature. Understanding what factors impact lane keeping will shed light on how sampling techniques, study design, and driving scenarios affect performance data.

Data collected from twenty-nine studies on the NADS-1, a high fidelity full-motion simulator, and from nine studies on the NADS miniSim^TM, a quarter cab limited field of view non-motion simulator, were combined for analysis by speed limit, number of lanes, and presence of traffic. Driver characteristics of age, sex, years of driving experience, and miles driven per year were captured from questionnaire data for each participant where available. A total of 1,719 participants were included in the analysis, with nearly eighty-five percent of the sample coming from NADS-1. Data were analyzed using the SAS GLM Select procedure with simulator, scenario, and driver characteristics as independent measures. Quadratic effects for age and years of driving experience were also included.

SDLP and lane crossings per minute are affected by simulator platform, driving scenario characteristics, and driver characteristics. The model predicts that SDLP on the NADS-1 will be approximately 2 cm greater than on the miniSim^TM. The model predicts that SDLP increases while lane crossings become less frequent on roadways with one lane in the driver's direction of travel and when traffic is present. Both SDLP and lane crossing are predicted to increase as posted speed limit increases. The model also identified a non-linear relationship between both SDLP and driver age and lane crossing and driver age. The model predicts fewer lane crossings as event duration increases.

Overall, the results indicate that differences in the simulator platform, the design of the simulator scenario, and the age of the participant sample can play a profound role in the absolute values observed when comparing between studies. The design of simulator scenarios must consider a consistent speed limit, number of lanes, and presence of traffic during the periods of driving to be analyzed. There exists a need to balance participant age in the samples of between subject designs to ensure parity to avoid a confounding effect on SDLP or on the rate of lane crossings per minute. Due to a non-linear relationship, particular care must be taken when including younger and older drivers in the design. Taken together, this suggests a need for careful study design to control the various factors that impact baseline driving performance.