Bayesian network based collision avoidance systems

Motor vehicle collisions are the leading cause of death in the Unites States. Rear-end crashes alone occur approximately 1.6 million times each year [1]. These statistics demonstrate the obvious need to reduce the number of vehicle collisions and save lives. In response, the government, automobile industry, and academia have conducted intensive research in an effort to enhance the safety in the U.S. transportation system. Such research has led to a recent trend to develop the next generation driverless car. This paper proposes a new methodology for use in vehicle safety system that has the potential to be used in autonomous driving (driverless vehicles). The new method applies Bayes' probabilistic reasoning technique to a multi sensor data fusion system in order to enhance a vehicle collision avoidance system in real time. The proposed methodology integrates multiple sensor readings, such as the speedometer of the host vehicle, and other sensors mounted on the vehicle to measure the speed of the leading vehicle. This methodology was modeled by using MATLAB and proved to produce a more reliable and certain decision for the host vehicle to react in order to avoid any potential collision.