Environment perception is an important component for autonomous driving. One of the main tasks is the detection and tracking of dynamic obstacles in the surroundings of the ego-vehicle. The paper proposes a method to achieve this process precisely, reliable and without Graphical-Processors-Unit (GPU) resources. A certified autonomous shuttle driving on two test sites in Heilbronn and Bad Wimpfen is used for evaluation. Starting with the multiple 3D-lidar-based sensor setup on the shuttle, the detection and tracking process is presented step by step. At first, the point clouds obtained by different sensors are preprocessed separately. In detail the point clouds are transformed into the vehicle coordinate system, cropped to the region of interest, statistically filtered to remove outlier and finally fused to a so-called total point cloud. The total point cloud is down sampled, filtered again and as a final step, clustered into coherent 2D point groups. An extended Kalman Filter, which estimates the velocities for the measured objects, is designed for the tracking procedure. The output of the filter is an object list, where each element has a position with center point, a bounding box, a velocity vector as well as an estimated classification. The setup and methods presented in this paper have been proven themselves over 6000 km on public roads in urban areas in the cities of Heilbronn and Bad Wimpfen.