During a disaster, it may be difficult for the visually impaired to use infrastructure or to obtain care worker's support. To support the independence of the visually impaired, we have developed a wearable travel aid that assists in safe travel, including navigating stairs, without special infrastructure. Because safe navigation requires accurate position estimation, our system fuses data from different sensors to estimate a pedestrian's position. In this paper, we propose a floor estimation method based on fuzzy inference with a laser range finder (LRF). When walking on a floor, an image processing system and a LRF are used together to estimate the position, but when walking on stairs, a second LRF is used to estimate the floor and detect obstacles. The experimental results show that our floor estimation is also useful for position estimation.