Real Time Application of Sensor Data Fusion in Automotive Safety System

Adla, Rawa, Nizar Al-Holou, and Youssef Bazzi

Automotive safety domain has become more attainable in the research and auto industry. One of the most important issues is to enhance the sensing capabilities, aiming at avoid or at least mitigate vehicles collision. The use of multiple sensors in the domain of vehicle collision, necessitate the need to design an efficient and reliable sensor fusion algorithm in order to reduce the user monitoring load, and a reliable reaction decision to avoid any potential collision.

     Since each sensor provides an entity of evidence presented by a probability value, these evidences must be integrated in order to provide a hypothesis on how to react. In this paper, we introduce a model of sensor fusion. This model applies Bayes’ probabilistic reasoning technique to multi sensor data fusion system in order to enhance the vehicle collision avoidance system in a real time without any interference from the driver. Our model works by integrating sensors output and in order to produce a confidence measure value, this confidence value was introduced in the decision making process scheme and proved to be more realistic which means more reliable when compared with other techniques of collision avoidance.