Vehicle guidance with control action computed by a rao-blackwellized particle filter

Particle filters can be used in navigation and state estimation problems. They can approximate arbitrary posterior distributions and do not require assumptions of Gaussianity or linearizations. This paper suggests to use the discretized posterior distributions that are produced within the filter beyond the state estimation, to compute control actions based on the information about uncertainty of the states. A case study is presented, and this indicates that the suggested approach can give robustness against uncertainty and reduce the amount of required control action.