Ceiling Vision Based Autonoumous Navigation of a Mobile Robot uisng Hybrid Visual Features in a Large Indoor Environment

摘要: This paper describes the research results on an indoor robot navigation system using visual features extracted from ceiling images of a large building environment. To reduce cost and complexity system, we developed single camera-based that utilizes acquired camera installed to point upwards. make operation reliable, proposed method hybrid which include natural in scene artificial observable infrared ray domain. Compared with works utilizing only based (1), our reduced required number as exploit both features. In addition, compared (2), has advantage convergence speed robustness observation feature provides crucial clue for pose estimation. experiment challenging situations real environment, was performed impressively terms accuracy. knowledge, is first vision visible rays domains. Our can be easily utilized variety service applications