Burgett 1 AN EMPIRICALLY BASED PROCESS FOR ESTIMATING SAFETY BENEFITS August Burgett URC Enterprise

The advent of new crash prevention technologies has made it necessary to develop new methodologies capable of estimating safety benefits before they are introduced in the market. This paper describes the development of one such empirically based methodology for estimating safety benefits. The developed methodology uses detailed engineering descriptions of the crash prevention system performance, in conjunction with a universal description of crash causal factors and resulting relevant crash types. This study also establishes objective tests to evaluate systems; a brake-assist system is used to describe the process. Crash data files such as the General Estimate System (GES) and the Fatality Analysis Reporting System (FARS) are used to develop Measures of Performance (MOP) and are used as the basis for objective tests. Naturalistic driving data are used to estimate Exposure Ratio and to refine elements of the objective tests. Finally data from the objective tests are used to estimate benefits of the crash prevention system. The process developed here has not been applied to a specific system. For that reason a hypothetical system is used as an example to demonstrate the data processing required to convert test outcomes to number of crashes that would be prevented. A generic brake-assist system is used as an example. This hypothetical example suggests that such a system could prevent 50% of rear-end crashes in which the following vehicle brakes to avoid crashing into a decelerating lead vehicle. The new process developed here has not been applied at this time. Challenges that will arise during application are not addressed in detail in this paper. INTRODUCTION In 2004, more than 10 million drivers were involved in a police-reported-crash. Each of these crashes began with a critical event, defined as a causal factor of the crash. The drivers who were confronted with these events responded by braking, steering or both; or by taking no action. The distribution of critical events and driver responses from the 2004 GES 1 are shown in Tables 1 and 2, respectively. It is the combination of critical event and driver response that occasionally (one crash for every 450,000 miles of travel) results in a crash. A basic foundation for assessments of crash avoidance systems is the distribution of these critical events for each crash related vehicle and driver responses associated with crashes. Table 1 Distribution of critical events for all vehicles involved in crashes (rounded to nearest thousands) Critical Event Total Subject vehicle loss of control 671,000 Action by subject vehicle 3,082,000 Action by another vehicle in subject vehicle‘s lane 4,114,000 Encroachment by another vehicle in subject vehicle’s lane 2,510,000 Pedestrian and other non-motorist 113,000 Object or animal 455,000 Total 10,945,000 The complete timeline of the chain of events for each vehicle in the GES file (critical event, driver response, first harmful event) is summarized in Table 3. This Universal Description provides a high-level, but complete, picture of each vehicle experience. A focused analysis of the underlying details of the Universal Description provides a solid foundation for developing test procedures and