In this project, we have developed a method of automatic registration of point clouds in urban areas. The data is acquired by GeoSim, using a LiDAR scanner mounted on a car. Our project’s goal is to bring a sequence of clouds, acquired every 20-30 meters, to the same coordinate system. In the project, we present 2 solutions. The first solution is based on existing methods of identifying feature points according to geometric features of the local environment, creating descriptors which describe the local geometry of each feature point, comparing them to find correspondences, and calculating transformations from these correspondences. In the second solution, we introduce a new registration method by creating a grid of viewpoints at the center of each scan, calculating 1D range descriptors for each viewpoint, comparing the descriptors by a unique correlation method for finding correspondences, and calculating a transformation from these correspondences. The first method has resulted in 54% success rate, while the second method has resulted in 92% success rate. Success is defined as translation error smaller then 1.5m and total rotation error less than 1°.