Highly automated vehicles are expected to perform ordinary driving tasks as well as improve safety, in emergency situations when in their desired domain of operation. This poses challenges in testing such systems. Conventional methods of testing, which recreate specific scenarios may address some emergency situations but simply do not cover all the scenarios a highly automated vehicle is expected to handle.

This paper proposes a method that can be applied to normal driving that quantifies risk at any instance. This method analyzes the current situation to determine the probability of an unavoidable collision occurring. This probability is described as an Instantaneous Safety Metric (ISM). This type of evaluation allows for the presence of traffic configurations with a high collision probability to be identified at any point in time, and even if no collision occurs.

Simple vehicle models are used to project possible future positions of each vehicle in a scenario and the probability of a crash estimated. This document presents results from development of this method, to this point, and the current view of a path to completion.