Research

A significant part of my research activities have been related to the investigation of crash risk under low ambient light conditions using an analysis technique that exploits annual daylight saving time transitions. This technique supplements crash datasets with geographical and astronomical data related to abrupt lighting change across daylight saving time (DST) transitions, allowing dark/light comparisons across the same time interval. The resulting analysis eliminates many confounds that have plagued day-night comparative analyses: driver fatigue, demographic shifts, and alcohol involvement.

This research has produced compelling and counterintuitive results. For example, pedestrians have emerged as a key population that is particularly vulnerable to dark road conditions, with an estimated fatal crash risk to be 7 times that of daylight. This has helped refocus forward vehicle lighting design efforts to directly address the issue of nighttime pedestrian safety. This research has also suggested that darkness does not appear to contribute to road departure crashes as many have believed. Instead road departure crashes appear to be related to driver fatigue, excessive speed, or alcohol use.

Later work in this area has also demonstrated that pedestrian risk in darkness is multiplied by vehicle speed, a finding consistent with the hypothesis that drivers are poor in recognizing when the forward illumination range is insufficient to permit safe avoidance of roadway obstacles. One consequence of this is that drivers often fail to use their high beams appropriately, a result confirmed in nighttime observational studies I conducted as part of the Industry Affiliation Program for Human Factors in Transportation Safety. The result has led to the renewed interest among OEMs in “smart beams” that perform the switching automatically.

Other applications of the DST analysis technique has showed that risk of fatal rear end collisions are also elevated in darkness. In particular, trucks have an especially high risk of being struck in darkness compared to light vehicles, prompting some of my work investigating the effectiveness of conspicuity treatments on trucks.

One of the greatest challenges in vehicle safety research is the objective measurement of the influence of safety technologies on driver behavior. Over the last 20 years, I have applied several analysis strategies to understand this influence. In my role as Co-PI in the Field Operational Test of Rollover Stability Advisor (RSA), I designed the experimental procedures by which the RSA’s influence on commercial drivers could be assessed. This activity included management of driver training, development of survey instruments, design and execution of analysis plans, and authoring chapters and presentations associated with the human factors aspects of the study. Using a temporal windowing analysis technique, I compared a driver’s naturalistic performance before and after advisories were issued, finding that drivers respond to advisories with turning maneuvers that produce lower levels of lateral acceleration over a limited period after the advisory. A similar analysis was later applied to data collected from Road Departure Crash Warning (RDCW) Field Operational Test. That study operationalized the construct of “subjective false alarm” and constructed a statistical model that related this to a driver’s response to future warnings.

Other technology assessment activities included analyses of driver performance using a night vision enhancement system, subjective judgments of advanced lighting systems during turn maneuvers, simulator studies measuring driver response to prototype blind-zone detection systems as part of the Advanced Collision Avoidance Technology-2 project, and studies of advanced lateral and longitudinal control systems in vehicles

Most recently, my colleagues and I have been looking at how drivers come to understand and learn to use advanced driver assistance systems (ADAS). In particular, we have been looking at drivers' understanding of adaptive cruise control (ACC) and lane keeping assist (LKA), finding gaps in their knowledge of these systems in potentially safety-critical areas.