Dickmanns 1992b : A paradigm for machine perception is presented which takes time and 3D space in an integrated manner as the underlying framework for internal representation of the sensorially observed outside world. This world is considered to consist of material and mental processes evolving over time. The concept of state and control variables developed in the natural sciences and engineering over the last three centuries is exploited to find a new, more natural access to dynamic real-time vision and intelligence. A. Schopenhauer’s conjecture of ‘The world as evolving process and internal representation’ (1819) is combined with modern recursive estimation techniques [Kalman 60] and some components from geometry and AI in order to arrive at a very efficient scheme for autonomous robotic agents dealing with evolving processes in the real world in real time. Application to autonomous mobile robots is discussed.