Mueller and Baten 1995: An intelligent vehicle that navigates in a natural outdoor environment fully autonomously is described. To handle this demanding task, the system has to be able to combine measurements and static a-priori knowledge in order to act in an intelligent manner. Static knowledge occurs in different manifestations, e. g. in maps, in physical and geometrical models and in the knowledge about the basic skills the vehicle is capable of. Measurements are taken in real-time by fusing data from different sensor systems, such as vision, odometry, simple inertial sensors and GPS. Both form parameters and state variables are estimated exploiting dynamical models that inherit space and time conditions.
The system is built up in a hierarchical control structure. Measurement, state estimation, behavior decision, vehicle and gaze control work as parallel processes. They are mapped onto a transputer net. So, the entire hardware needs a volume of only about 30 liters.