A Comprehensive Lane Detection and Following System in Autonomous Vehicles
Hannah Grace Mason (Electrical and Computer Engineering, Lipscomb University)
Joe MacInnes (Computer Science, The College of Wooster)
Landon Bentley (Computer Science, The University of Alabama)
Lane detection and following are important aspects for the future of autonomous vehicles. Their project's goal was to explore a comprehensive implementation for lane detection and following using hardware-in-the-loop testing on a hybrid autonomous vehicle. Hue, lightness, and saturation are used with a perspective transform to detect lanes from an image and produce a bird’s eye view. A modiﬁcation to existing path extraction algorithms is proposed that improves upon the shortcomings of current algorithms. The resulting algorithm generates waypoints from extracted lane markings with higher accuracy in curvy lanes with discontiguous markings. The resulting waypoints are sent to a Hoffmann steering controller which performs real-time trajectory correction. The project details the implementation of these subcomponents and a comparison to the results of competing approaches.