The further step that we want to reach is the modelling of trajectories by sequences of sub-trajectories. We want to develop an algorithm that is able to find coherent sub-trajectories in a sequence representing a trajectory. All the similar trajectories will be grouped in a set of clusters. The idea that we certainly have to use for this clustering is that, if two trajectories are similar, a point taken randomly on one will be close to another point taken randomly on the other trajectory.