Wheeled mobile robot (WMR) is not a new subject but it has gained interest in recent years due to its high applicability in civil industries and military field. Autonomous robots can perform desired tasks in unstructured environments without continuous human guidance. A fully autonomous robot has the ability to gain information about the environment
and take decision according to this information independently. Vision is the human most powerful sense. In this paper, a vision-based lane detection algorithm is presented for tracking the road for autonomous wheeled mobile robot. The presented algorithm relays on feature based detection which detects the interested pixels by a mixture between image thresholding and edge detection. This mixture gives advantages of execution time reduction, overcomes shadow and dark areas problem and works with different day night time. Hough transform is used for line detection. The required heading angle is calculated by using these detected lines after the adjustment of the path following control law. This angle is used for controlling the WMR. Decision trees are utilized to increase the robust of robot behavior and have the benefit of solving the breaks in marking lines problem. The lane detection algorithm along with the decision trees are tested experimentally and they give high accuracy in different environment conditions. Finally at controller design step, WMR mathematical kinematic molding is hired to characterize the controller gains.