Line detection is one of the most important components in autonomous driving.
The presented work in this thesis achieves single-color line detection on an arbitrary surface, using the k-means machine learning algorithm. Program for detection and line tracking is written in Python programming language and tested in the Gazebo simulation environment. Experiment in a real-world setting was also carried out. Vehicle used for testing the algorithm is Husky, a ground vehicle robot with differential drive and four wheels.. Communication between Husky and the control algorithm is accomplished using ROS (Robot operating system). Husky is controlled by adjusting its wheel speeds according to the given linear and angular velocity speeds of the robot.