Potential Crash Rate Benchmarks for Automated Vehicles

Most automobile manufacturers and several technology companies are testing automated vehicles (AVs) on public roads. While automation of the driving task is expected to reduce crashes, there no consensus as how safe an AV must be before it can deployed. An should at least average driver, but national crash rates include drunk distracted driving, meaning that crashes rate somewhere between sober. In this paper, safety benchmarks for AVs explored from three perspectives. First, naturalistic studies used determine risk model (i.e., sober, rested, attentive, cautious) driver. Second, stated preference surveys in literature reviewed estimate acceptable public. Third, crash, injury, fatality other transportation modes compared baseline levels. A range potential targets presented a guide policymakers, regulators, developers assist evaluating technologies use.