Lane Detection and Tracking Algorithms for Driver Assistance System

The majority of fatalities on the roads each year are caused by drivers who are careless and disregard traffic laws. Safety is the system’s primary goal, and lane detection devices help prevent accidents. These systems’ primary function is identifying lane markings and alerting the driver if a vehicle is about to veer off the road. The recent technologies in advanced driver assistance systems lead to the development of various techniques for improving driver safety and automating driving. One such development is the Lane Departure Warning (LDW) system, wherein lane detection techniques are employed for the detection of lanes to avoid road accidents. Accordingly, three methods are discussed for lane detection in the research work. The first method for multiple lane detection is discussed based on image transformation and CSA-based Deep Convolution Neural Network (DCNN). The multiple lane images are initially transformed into Bird’s eye view images. Next, the detection of the lane is carried out by an Earth Worm-Crow Search Algorithm (EW-CSA) based on DCNN. The Earth Worm-Crow Search Algorithm was discussed by merging EWA and CSA. This merged algorithm can be used as the training algorithm in the DCNN. The performance parameters are also discussed to compare the different lane detection techniques.