Predicting route targets based on optimality considerations

In this paper we present an approach to predict a target of a mobile user on the move. After observing the movement from a starting point, we are able to create possible extrapolations of routes. Our basic assumption: a mobile user tries to move efficiently, thus only a certain set of destinations is reasonable. We use a road network that contains information about movement costs to detect reasonable movements, but we do not expect theoretical optimal paths. We are able to model different efficiency goals and different degrees of optimality. We present an efficient algorithm to actually compute the set of reasonable targets that avoids brute force computation. In contrast to existing work to predict route destinations, we do not require a learning phase to collect an archive of former routes.