Visual servoing line following robot: Camera-based line detecting and interpreting

Autonomous self-driving car has gained a lot of interest with the advance of robotic technology. An autonomous car using only visual information for autonomous navigation is expected to mimic the way human driving capability. The car has to navigate robustly and completing the tasks in a specified environment. These tasks for example detecting road lanes, pedestrians, or road signs. In this paper, as a step toward the goal of developing an autonomous self-driving car, we develop a prototype of autonomous line following robot. Using only visual information, we study a visual servoing technique for which the robot able to detect road lanes and navigate autonomously. An image processing technique is employed to obtain effective features for feedback control. The controller utilizes features extracted from the image plane, without using the complete geometric representation. Motion is controlled only on x-axis or heading angle control. The proportional integral and derivative controller is used to maintain the heading of the robot. Experiments show encouraging result for the robot to navigate autonomously given the road lanes and maintain heading angle.