A New Methodology for Vehicle Collision Avoidance using FMCW Radar and Critical Distance Estimations using K-Means Clustering Algorithm

摘要

In this paper we propose novel techniques to avoid vehicle collisions using a collision avoidance system in highway scenarios. A personalized time delay close to 2 seconds is maintained between the host and target vehicle. Compared to the conventional laser and radar system we use FMCW radar to track the speed parameters and position, with which a virtual boundary is created for two purposes. To maintain headway distance and provide braking when the target vehicle comes very close to the host vehicle. The system calculates the reaction time of the driver and applies K-means clustering algorithm to obtain a specific reaction time for different ranges of velocity, personalized for an individual driver. Unlike certain collision avoidance systems which take relative velocity as a major factor in determining the braking distance, we take into account of host vehicle velocity as a major parameter. This will provide a more comfortable distance between the host vehicle and the target vehicle. A graduated light display indicates the proximity of the target vehicle from the host vehicle enabling the driver to maintain an apt and comfortable distance.