Planetary rover exploration requires three fundamental tasks: control, localization and mapping. The performance of autonomous navigation depends significantly on the robustness and reliability of the localization subsystem, which is responsible for closing the navigation control loop while driving. Due to inherent sensor errors, sensor fusion techniques need to be used to estimate the resulting position and attitude more accurately than if each sensor was considered individually. In addition to this, Space constraints implies limited resources at the same time more capable system are required. This is forcing us towards better solutions for the coming years

Statistical sensor fusion comprises four high-level primitives: fusion, defusion, estimation and transformation. This talk directs the attention to the importance of combining all those steps, presenting a more complete localization scheme that nowadays solutions. The proposed schema combines temporal and spatial probabilistic constraints while interacting with the environment. First qualitative and quantitative results will be analyzed during the talk. Critical aspects and future work will be discussed towards a preferable SLAM solution for planetary rovers.