Numerical Evaluation of the Safety of Self-Driving Vehicles: Functionality Involving Vehicle Detection

Due to recent accidents, safety has surfaced as the number one concern for the acceptance and adoption of autonomous vehicles. The development of safety systems in selfdriving vehicles has some of the most complex requirements. How does one know when a self-driving vehicle is safe enough? And equally important, how do we measure the safety level of a given vehicle design? In this paper we present a mathematical model that is used to numerically evaluate the safety level of five different self-driving vehicle designs, in terms of important parameters related to the design process. The main results show that strict adherence to safety standards, and the use of faulttolerant techniques, particularly for the perception system, can considerably improve the safety level of a self-driving vehicle.