In this paper, we describe a robust and fast self position estimation technique for mobile robot in an environment where many unknown obstacles exist. The free-space observation model is the basis of the proposed technique. Although the free-space observation achieves robust self-position estimation, it has a complicated likelihood evaluation, and particles continue to spread to a direction where are no landmarks. In this research, we solve these problems with two likelihood evaluations based on the area of free space and occupied space. We evaluated the robustness and verify the validity of the proposed method by a simulation and an experiment in a real environment.
