DSRC based collision warning for vehicles at intersections

The increasing number of collisions at intersections calls for a reliable and safe, real time collision warning system that can alert the drivers about a potential collision. Although many kinds of collision warning systems are available, the problems related to acceleration change and curved road scenarios are to be considered to improve the accuracy of the system. In the proposed paper, an efficient intersection collision avoidance system for straight and curved roads are considered. The objective of the project is to predict the location of the vehicle 10 seconds in future. The location, speed and heading information of a vehicle are available through a GPS device. Accelerometer provides the acceleration/deceleration information of the vehicles. Kalman filter with GPS and accelerometer as inputs can be used to predict the future location of the vehicle. These predicted locations are checked to be on the road or not using GIS map. If off road, the locations are projected onto the road with the help of the coordinate data. A warning message in the form of alarms or lights are generated when the distance between the predicted locations of the host vehicle and target vehicle falls below a threshold value. Position information are exchanged between the vehicles using Dedicated Short Range Communication. The MATLAB simulation results validate our proposed system by reducing the prediction error with the combination of GIS and kalman filter. False alarms are reduced considerably in the proposed system for collision avoidance.