Robust non-linear smoother for state-space models

This paper presents a robust, non-linear smoothing algorithm state-space models driven by noise and external inputs. This algorithm is extremely robust to outliers and handles missing data and state-dependent noise. Its implementation is straight-forward as it consists of two main components: (a) the Rauch-Tung-Striebel recursions (a.k.a. the Kalman smoother); and (b) a back-tracking line search strategy. Since the algorithm preserves the underlying structure of the problem, its computational load is linear in the number of data. Global convergence to a local optimum is guaranteed under mild assumptions.