Motion Prediction by Optimal Paths Through Disordered Landscapes

Abstract

The ability to navigate safely and efficiently through a given landscape is relevant for any intelligent moving object. Examples range from robotic science and traffic analysis to the behavior within an ecosystem. Through this thesis, methods for finding traffic patterns and predicting future motion, have been constructed based on theory of optimal paths. The algorithms are applied to maritime traffic, in terms of recorded vessel coordinates. \newline By considering the structure of a given traffic situation as a disordered energy landscape, one can define optimal routes within the area. An algorithm for finding hierarchies of optimal paths in a disordered energy landscape is implemented. The algorithms are used in two settings, one for detecting patterns of motion within a given area, and a method for estimating single vessel trajectories. The results found in the thesis, show that the methods have great potential for analyzing traffic patterns and predict future motion.