Robust navigation systems are of great importance in the field of mobile robotics. In order for a mobile robot to be useful, it must know its surroundings at all times to be able to perform its task. The complexity of creating navigation is highly dependent on the environment the robot is to work in and whether or not geographical data of this environment is known beforehand. The key focus in this thesis was to find fast and time efficient robust algorithms and models for driving autonomously along an unknown road using minimal equipment. It also strives to find a trade off with the least amount of parameters necessary for doing so. A range finding laser scanner (lidar from SICK) was used as a single sensor for interpreting the surroundings of the robot. The scanning laser measures distances in a single plane in front of the car and these data are then interpreted by a road finding algorithm. The goal is to distinguish the road from the terrain. When the system has found what is interpreted as the road it will pass that information to the driver model which is responsible for controlling the car. After the development of a steering model and road finding algorithms, the car successfully managed to navigate autonomously over 1 kilometre along a curved and hilly forest road. While remaining in constant low speed (<20 km/h) the system was robust. higher speeds and acceleration showed weaknesses in the system. including a rate gyro could increase the stability and speed.