Sensitivity analysis and tuning for active safety systems

Active safety systems utilize information about the vehicle´s state and possibly the surrounding environment to assist the driver if a traffic situation is considered hazardous. The decision to assist the driver is based on sensors that are more or less subjected to errors. In this paper, we study the influence of input perturbations on decision functions in active safety systems. We present a framework for evaluation of system performance, specification of input requirements and decision function tuning. By introducing a robustness measure, describing the robustness to error for the decision function, we derive efficient methods for analyzing the system performance. The framework is demonstrated on experimental data.