One of the goals in the field of mobile robotics is the development of mobile platforms which operate in populated environments and offer various services to humans. For many tasks it is highly desirable that a robot can determine its own position, the states of various objects as well as the positions of the humans in its surrounding. In this talk we present Bayesian filtering as a probalistic state estimation method for mobile robots. We discuss different techniques to represent the underlying densities and present applications to robot pose estimation and people tracking. For the problem of tracking multiple objects, we introduce a new sample-based variant to track features originating from individual objects and to solve the correspondence problem between the detected features and the filters. We present several experiments illustrating the robustness of our approach even in dynamic environments