Driving Assistance System at T-junctions in Mauritius

Road accidents somehow have made it to the headlines in the news or have been the hot topics on social media in the past few years in Mauritius. This work focuses on finding a solution to aid car drivers at T-junctions which had the highest number of accidents amongst all the other junctions in Mauritius. The main objective of this work is to assist a driver at a T-junction in decision-taking whether to move out or wait and hence to decrease the number of accidents at T-junctions. An advanced driving assistance system has been developed using Raspberry pi and Neural Compute Stick to achieve better frames per second during the processing of the live video streaming. The system detects, tracks the oncoming car and calculates the speed of the vehicle approaching a T-junction. The detection, tracking and speed calculation of a car were tested and the results obtained were satisfactory. In the case where no vehicle was being detected, the system successfully displayed a message that the lane was clear and the driver could move out of the junction. Also, the latter was asked to wait when the lane was blocked and was alerted of a danger whenever the speed limit was exceeded by the vehicle on the adjacent road at the T- junction. The average speed accuracy obtained is 84.96 %.