Naturalistic Driving Study: Technical Coordination and Quality Control (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

87 C h a p t e r 8 Summary This study was undertaken as one of the major research ini­ tiatives supporting the Safety area of the second Strategic Highway Research Program (SHRP 2), authorized by Congress in the Safe, Accountable, Flexible, Efficient Transportation Equity Act. The idea of conducting a large­scale naturalistic driving study was underpinned by the success of the 100­Car Study (Dingus et al. 2006), a landmark study in its own right. That study not only demonstrated the feasibility of conducting a study of this nature but also showed that the resulting highly complex data set could be successfully mined and analyzed by a variety of researchers from a variety of perspectives. S06 study design parameters were based on the plan devised in the planning project (Antin et al. 2011), and all aspects of that plan were based on the identification, generation, and collection of several hundred research questions that guided all subsequent decisions and actions. Study sites were selected based on organizations that responded to an initial request for qualifications (RFQ); those responding were then down­ selected to the final six organizations (and their respective sites) that study sponsors determined were best suited to success­ fully conduct the study with the best possible geographical diversity. The goal was to have 1,950 cars on the road simulta­ neously collecting naturalistic driving data for 2 years or more. Study sites and the nominal number of DAS kits managed at each site are depicted in Figure 2.11 in Chapter 2. To address the breadth of the assembled research questions, the DAS had to be designed to • Be robust enough for long­term field deployment in climates that exposed them to extremes of heat, cold, and moisture; • Be robust enough to continuously collect data not only in the presence of the vibrations, accelerations, and decelera­ tions experienced in the automotive environment, but even under crash conditions; • Automatically begin continuous data collection when the vehicle was started and stop when the vehicle was turned off; • Collect data from the vehicle’s onboard network, as available; • Be installed and deinstalled in a way that would not perma­ nently damage or change the vehicle in any way; • Be installed so that it would not cause a safety hazard, even in the event of a serious crash; • Be able to communicate via cell technology automated health checks, programs, and a variety of video and data snippets; • Be integrated in such a way that it would neither affect nor interfere with the vehicle’s operation or handling in any way; • Be unobtrusive enough that participants could largely ignore it; and • Collect a broad range of data over a period of months without attention or maintenance. The data collected included four video channels (face, forward, rear, and center stack interactions). In addition, the DAS integrated the following sensors: accelerometers in three dimensions, front radar, GPS, and vehicle network data, among others. The study resulted in the successful collection of 2 peta­ bytes of real­world driving video and sensor data from more than 3,000 participants over a 3­year period beginning in fall 2010. This data set includes some 50 million miles and well over a million hours of naturalistic driving data. Participants represented an approximately equal mix of males and females ranging in age from 16 to 98. With a study of this nature and scope, there was a substantial concomitant effort to • Ensure that ethical procedures were implemented at every stage and site; • Ensure that human subjects protections (including a Cer­ tificate of Confidentiality) were in place; and • Work with all active stakeholder IRBs (six in all, as two site contractors chose to formally rely on the Virginia Tech IRB) to secure not only initial study approval but also approvals Study Summary and Future Research Implications

88 of today and well into the future, just as data from the 100­Car Study have been mined and analyzed for the past decade. Follow-On Studies Each participant was asked when exiting the study if he or she would permit his or her name to be included on a list of those who may be contacted for follow­on studies, studies in which additional data might be collected on the participant. Inclusion on the list in no way obligates individuals to participate in any such future study, and they can opt to have their names removed from this list at any point. It is expected that policies for use of the list by researchers will be developed under the guidance of an oversight committee managed by TRB. Cell Phone Records Study The first follow­on study has already been initiated: the Cell Phone Records Study (CPRS). As participants exited the driving study, they were given the opportunity to participate in this follow­on effort. The CPRS entailed acquiring as much participant cell phone data as possible—voice and text only— for the period covering the individual’s participation in the driving study. Once acquired, the data will be integrated into the larger naturalistic database to give researchers additional insights as to participants’ cell phone usage behaviors while driving, and the safety­related implications thereof. Figure 8.1 presents the total number of SHRP 2 NDS par­ ticipants, both primary and secondary, versus the number who agreed to be contacted for future studies and the number who agreed to participate in the CPRS. for a series of amendments that continued virtually to the end of the data collection period. It was the Coordination Contractor’s role to oversee and coordinate the entire study, developing, procuring, and con­ figuring all the hardware and software elements required, both onboard the vehicle and otherwise. The Coordination Contractor also provided training in a wide variety of areas for the site contractors with whom there was close collaboration throughout the duration of the study. Another major role of the Coordination Contractor was to manage the secure flow of data from the onboard data drives through several stages all the way to storage on servers at its location. Checks and redundancies were built into this largely automated process at each step to ensure no data would be lost in transmission. Key to the data management and storage process was a continual process of assessing the quality of not only the DAS data, but also all other types of data collected in the study (e.g., demographics, driver assessments, and post hoc crash investigations). Data anomalies were identified and corrected when possible, and quarantined when no remedia­ tion was possible. Future research Implications This project entailed very little data analysis per se, as vir­ tually the entire effort was founded on the notion of future research. The overarching goal was to collect a very large, extremely rich and detailed store of data to be mined and ana­ lyzed by a generation of transportation safety researchers and others to answer many of the key traffic safety–related questions Figure 8.1. Total participants versus participants who agreed to be contacted for future studies.