Schmid and Thomanek 1993: An object recognition system for an autonomous vehicle guidance application is presented for interpreting monocular real-time image sequences of motorway scenes. The approach described is based on an integrated spatio-temporal 4D model for object tracking and relative state estimation. This task is solved by combining a fast obstacle detection and tracking algorithm using only 2D image information with an extended object recognition module, which is able to handle two objects represented as 3D models, even in the case of partial occlusion. Up to three independently moving objects in the perceived scene can be processed in real time by the obstacle detection module. The system implemented on a transputer cluster has been tested by using noise corrupted measurements of synthetic and real images. The 2D obstacle detection and tracking part has already proved its reliability in an autonomous guidance and control system on German motorways.