Studies have shown that fatalities due to unintentional roadway departures can be significantly reduced if Lane Departure Warning (LDW) and Lane Keep Assist (LKA) systems are used effectively. However, these systems are not yet popular because the systems are not robust due, in part to the lack of suitable standards for pavement markings that enable reliable functionality of the sensor system. The objective of this project is to develop a reference Lane Detection (LD) system that will provide a benchmark for evaluating different lane markings and perception algorithms. The project will also validate the effectiveness of lane markings as well as the vision algorithms through a systematic development of LD metrics, and testing of LD algorithms in a robust test/vehicle environment.
