Collision warning using neighboring vehicle information.

In a typical year, millions of rear-end collisions occur, resulting in hundreds of thousands of injuries and thousands of fatalities. As the population and the number of miles driven continue to grow, these numbers will likely increase as well. This dissertation proposes a previously unexplored vehicle collision-warning approach aimed at reducing the number and severity of rear-end collisions, thereby reducing the number of associated injuries and deaths. The system uses information broadcast by nearby vehicles equipped with the system, as well as information available from on-board sensors, to advise the driver of any required deceleration. The proposed system does not require additional roadway infrastructure or dedicated lanes, and can be used in the presence of equipped and unequipped vehicles. The requirements of this system were evaluated, and it was found that the requirements could be met with currently available technology. A candidate algorithm for the system was developed which utilizes the information available to the device to determine the deceleration required by the vehicle to avoid an impact with the vehicle immediately ahead. The algorithm is memoryless in that it does not need any of the information used in the previous calculations to make the current calculation of the required deceleration and corresponding driver warning. This attribute makes the algorithm insensitive to losses of information from a specific vehicle caused by communications interruptions or the vehicle exiting the local environment. The amount of information required by the device to provide benefit, and the potential reductions in collisions, impact velocities, and deceleration rates of the system in operation, were evaluated using computer simulation. In addition, the impact of the system on the number of yearly fatalities is also discussed. The results of this evaluation show that the system has the potential to provide almost immediate benefits even with a low ratio of equipped vehicles to unequipped vehicles. The magnitude of these benefits is non-trivial. In some cases, the reduction in the number of collisions could be as much as 50%, and the reduction in the number of fatalities is estimated to be as much as 75%.Ph.D.Applied SciencesAutomotive engineeringComputer scienceSocial SciencesTransportationUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/130196/2/9721942.pd