Advanced Driver Assistance System (ADAS) are systems developed to enhance vehicle safety and assist the driver in avoiding collisions and unwanted accidents by providing alerts and/or warnings whenever required. The development of autonomous or self-driving vehicles depends on the development of ADAS. Various features are being offered in personal and commercial vehicles which are currently available, such as adaptive cruise control, autonomous navigation which is used in applications like self-parking, driver drowsiness detection, and collision avoidance systems, to name a few. Some of the above-mentioned features are dependent on less complex computations like traffic sign detection and recognition, lane detection and departure warning, and vehicle proximity warning systems, each of which are developed using the concepts of Computer Vision in real-time.
This paper attempts to develop a robust lane detection and tracking system, which identifies the lane in which the vehicle is currently traveling with the help of lane markings on the road, and tracks them in real time. The system should be capable of handling challenging scenarios such as worn-out markings or other distracting objects like shadows and other vehicles. This will be achieved with the help of footage imported from a front-facing dashboard camera. A variety of different algorithms have been developed over the years in various journals and conferences on computer vision and intelligent vehicles and this paper will compare the implementation and results of a few of these to find the most robust one based on criterion that will be clearly laid out.