^4fModelling the Vehicle Kinematics as Key Element for the Design of Integrated Navigation Systems

摘要

^7fEspecially represented by the very popular Kalman filter, the observer principle is a commonly used basis for integrating navigation systems. This applies both for such "simple" devices like a gyrostabilized magnetic compass and for the well known, complex combination of inertial sensors and satellite navigation receivers. The realisation of an observer is possible by using mechanical parts or an analog electrical circuit (like for the first case) as well as by means of special software implemented on microprocessors (like for the latter case). Correspondingly, one major advantage of observers is the independence from the sensor principles employed (e.g. optical gyros, micro-mechanical accelerometers, or radar units as typical elements of integrated navigation systems). Thus, it is not the technology, which determines mainly the character and the basic design of an integrated navigation system; it is rather the modelling of the kinematics for the vehicles considered. For instance, a gyrostabilized magnetic compass takes into account only the vehicle rotation about its vertical axis. Conventional dead reckoning systems assume the motion of a small, rigid body on a surface with two or three degrees of freedom. The classical combination of an inertial measurement unit and a satellite navigation receiver considers the motion of a small, rigid body with six degrees of freedom (and employs normally error states to describe the vehicle movement). Yet, there is no reason to restrict the vehicle modelling to these familiar examples. Alternative and more detailed models can lead to further improvements and new applications of integrated navigation systems. Using an observer variant with total state variables for the combination of inertial measurement units and satellite navigation receivers improves the system stability as it fulfils especially the observer principle. Additionally assuming large vehicles allows distributing several satellite navigation antennas over an extensive area and enhances considerably the system stability as well as the attitude accuracy. However, this can cause problems with structural flexibilities of the vehicle. Enlarging, on the other hand, the vehicle model by additional degrees of freedom for the distortions can not only overcome this difficulty but also enables a more complete recording of the vehicle motion. This leads to integrated navigation systems which have spatially distributed sensors and which are simultaneously beneficial to structural control as well. Likewise, the same strategy is also useful for multibody structures like robots. Based on simulated as well as on experimental data, the paper covers the following topics to illustrate the potential of vehicle modelling for improvements and new applications of integrated navigation systems: 1. observer principle and simple integrated navigation systems, 2. benefit of distributing several satellite navigation antennas over large vehicle structures, 3. design of integrated navigation systems for vehicle structures with significant flexibility and for multibody systems.