Deterministic Target Selection

A major challenge in the design of active safety systems for vehicles is to approximate the reality collected by sensors into a set of reliable and useful properties. Inability to do so can cause the active safety system to perform in a way, that will put the vehicle in a situation where the driver behaves in an unpredictable way, Unpredictable behavior in these systems can cause the vehicle to enter an unsafe state for itself or others. In order to achieve highly dependable applications it is important not only to know the behavior of sensor data, it is also important to put feasible limitations of the input data to guarantee a safe and predictable function. This paper presents methods to improve and assess the decision material of active safety functions such as driver support systems with autonomous braking. The methods are introduced to increase the dependability of the function but also to set more accurate scope and also avoid so-called slightly out of specifications, SOS, fault. Deterministic Target Selection - Setting Requirements on Speed and Yaw Rate in Automotive Sensor Systems.