|
Abstract:
|
Automated Vehicles (AVs) road tests are limited and contingent upon their safety validations, while the AV safety evaluation requires extensive road test data. This study addresses this chicken-and-egg paradox by proposing a methodology for evaluating AV safety, with limited road test data, compared to conventional vehicle safety as the benchmark. We define a new metric, mile-to-crash (MTC), to add another layer of information, the time of the crash, into the analysis. Then, we estimate the distribution of MTC for conventional vehicles and AV, which represents failure function, the complement of the survival function. Finally, we parametrize the conventional vehicle and AV’s empirical failure functions and test the hypothesis of whether the conventional vehicle and AV’s failure functions are statistically different. Our analysis of conventional vehicles and Level 3 AV crashes showed that, with 95% confidence, automated driving is safer in terms of MTC. Despite the uncertainties in AV crash reports, this study can be considered the most accurate verdict regarding Level 3 of automation safety. The proposed method is transferable for the safety evaluation of other AV technologies.
|
