26MOVE | Fall 2008
Until recently, transportation research relied on analy- ses of police-reported crash data and studies conducted on test tracks and simulators. While these methods have been helpful and effective, there are still many unan- swered questions. The only way to get to the heart of the driving problem is to conduct real-world—or, naturalis- tic—driving research.
To collect the needed data, researchers at VTTI ob- served the daily driving habits of 241 drivers in more than 100 cars in a completely naturalistic setting. This research project was the first instrumented-vehicle study of its kind, taken on with the primary purpose of collecting data while the driver is in a natural and comfortable driving environ- ment. This kind of study provides researchers with a more realistic and in-depth look at what causes a crash.
“Due to the unpredictability of driver performance and the random nature of automobile crashes, the col- lection of naturalistic data gives a more accurate per- spective of why crashes occur,” said Dr. Tom Dingus, director of VTTI and principal investigator for the study.
For years the National Highway Traffic Safety Admin- istration (NHTSA) has recognized the need for collecting naturalistic on-road driving data. This data was especially important for understanding what leads to crash and near- crash situations. In 2000, NHTSA, in partnership with the Intelligent Transportation Systems (ITS) Joint Program Of- fice, the Intelligent Vehicle Initiative, the Virginia Depart- ment of Transportation (VDOT), the Virginia Transporta- tion Research Council (VTRC) and Virginia Tech, awarded VTTI $3.7 million to study driver performance and behav- iors that lead to crashes and near-crashes.
The study consisted of more than 100 leased and privately-owned cars in the Northern Virginia/met- ropolitan Washington, DC area. These volunteers had their cars equipped with specialized data collection in- strumentation for 12 to 13 months each. They were ad- vised to use their vehicles in their normal daily routines and were given no special instructions. There were no researchers present and the data collection instrumen- tation was unobtrusive.
“The participants appeared to disregard the presence of vehicle instrumentation quickly,” said Dr. Vicki Neale, director of the Institute’s Center for Vehicle-Infrastruc- ture Safety. Participants were captured in their natural environments of driving to and from work, picking their kids up from school, going to the store, etc.
“This project provided a unique opportunity to study drivers’ performance in their own vehicles in real traffic conditions,” said Neale.
In order to get the information needed to better un- derstand the reasons crashes occur, but still keep the environment naturalistic, researchers instrumented the cars with an advanced and portable data collection sys-
Nearly 80 percent of all crashes and 65 percent of all near-crashes involved driver inattention just prior (i.e., within 3 seconds) to the onset of the incident.
tem. This system collected the data for a year and pro- vided video footage and vehicle-based details about the events leading up to a crash or near-crash.
There are two traditional approaches to collecting and analyzing human factors data related to driving. One ap- proach is to use data gathered from large-population stud- ies that are often done at the national level. However, these databases lack sufficient details that are helpful for many applications, such as the development of countermeasure systems or the assessment of interactions between con- tributing factors that lead to crashes, said Dingus.
The second approach uses data gathered through controlled experiments, including driving simulators and test tracks. However, according to Dingus, these studies cannot avoid a certain level of artificiality and do not always capture the complexities of the driving envi- ronment or of natural behavior.
“Test subjects are often more alert and more cautious in a simulation environment or when an experimenter is present in a research vehicle than when they are driving alone in their own cars,” said Dr. Charlie Klauer, project manager for the study.
The data acquisition systems are often compared to an airplane’s black box. However, the system is much more advanced and includes five channels of digital video, front and rear radar sensors, accelerometers, machine vision- based lane tracker, GPS, and a vehicle speed sensor.
“Our extensive naturalistic driving study became a reality thanks to advances in technology and data collec- tion systems,” said Andy Peterson, director of the Center for Technology Development at the Institute.