Map-based free navigation for autonomous vehicles

A general method for the construction of a map-based navigation system is proposed, which is based on a Kalman filter and can be equipped with many commercially available navigation sensors. A prototype has been implemented using a laser range finder and a 2D model of the natural environment. No artificial landmarks are necessary. The computing time, required for matching the measurement data with the model elements is low and independent of the model size. This is achieved by precompiling the navigation map to form a lookup table. The classical Kalman filter algorithm is modified to take into account the different speeds of the absolute and relative sensors as well as the computing time of the algorithm itself. A systematic way is given to decouple the time when absolute measurement data are taken from the time when these data are processed. Loss of track resulting from errors in the navigation map or incorrect dead reckoning is detected by comparing the real measurement data with on-line simulated measu...