Static and dynamic fusion for outdoor vehicle localization

The vehicle's localization is classically achieved by Bayesian methods like Extended Kaiman Filtering. Such a method provides an estimated position with its associated uncertainty. Bounded-error approaches using interval analysis work in a different way as they provide a possible set of positions. An advantage of such approaches is that the results are guaranteed and are not probabilistically defined. This paper focuses on constraints propagation techniques using static and dynamic fusion. Static fusion uses data redundancy to enhance proprioceptive data. Then dynamic fusion uses GPS in order to reduce the size of the localization box. The approach has been validated with a real outdoor vehicle.