SLAM Algorithms In Dynamic Environments

In this work the Kalman filter and the Particle filter are described and their performance in the Simaltaneous Localization And Mapping (SLAM) problem in static environments is discussed. Furthermore, this paper presents how derivatives of these filters are applied in order to solve the SLAM problem in dynamic environments. The Particle filter makes less assumptions about the probability distribution than the Kalman filter, and in that sense it is more general. However, the Kalman filter is optimal for linear Gaussian systems and has a smaller computational complexity if the number of states is high. Recent paper which offer a solution for the SLAM problem in the dynamic case often combine multiple methods which individually did not fully solve it. It is possible that in the near future the SLAM problem can fully be solved by integrating a Dynamic Vision Sensor into an existing SLAM algorithm.