Theory of Collision Avoidance Capability in Automated Driving Technologies

This paper proposes a theory to analyze the collision avoidance capability of automated driving technologies. The theory gives answers to a fundamental question whether automated vehicles fall into extreme conditions at all rather than another question how a vehicle reacts under extreme conditions (is it as safe as driver?). The theory clarifies the following matters: There are two types of hazards to cause collisions, cognitive hazards and behavioral hazards. Cognitive hazards are handled by controlling the upper limit speed of the automated vehicle including when stopped. There are two methods for handling behavioral hazards, preparation and response. The response known well is the coping method activated when the hazard is detected in the dynamic (operational) level. The preparation is the coping method operating at all time in the semantic (tactical) level. The collision condition in the semantic level is as follows, a collision occurs when the paths of two vehicles have a crossing point and the two vehicles drive on the crossing point at same time. The condition can be formulated as collision avoidance equation. Solving the equation means that the automated vehicle has prepared for the behavioral hazard before the hazard occurs. It is concluded that a collision avoidance capability consists of not only a response capability that supports the accuracy of collision avoidance in extreme conditions in the dynamic level but also a preparation capability that supports the accuracy to avoid reaching those extreme conditions in the semantic level. The preparation capability can be evaluated through stability analysis of the automated vehicle behavior given by the temporal backward simulation from each extreme condition. A remaining problem is how determine the upper limit of the hazards growing speed to which the automated vehicles should react.