Demonstration of Forward Collision Warning System Based on Real-Time Computer Vision

This paper demonstrates the software and hardware of a forward-collision warning system using techniques of realtime computer vision which helps self-driving cars and autonomous vehicles systems to merge with the road environment safely and ensure the reliability of these systems. The software approach of the paper consists of five parts: car detection, depth estimation, lane assignation, the relative speed of other cars and their corresponding speed limit and finally ultrasonic sensors which completes the front of the vehicle as the camera can't cover it alone. Besides these five objectives, a real-time notification system exists which interacts with the driver. In this paper, a hardware approach is proposed which implements the software approach techniques on a NVidia Jetson TX2 development kit and AVR microcontroller to prove the feasibility of the proposed collision warning system. The results show the reliability and promptness of the proposed approach by achieving up to 23 FPS (frame per second).