Pedestrian dynamics, which has been vigorously studied in traffic engineering, architecture and psychology, also started to attract interest of physicists in the end of the twentieth century. It is almost impossible to predict the movement of an individual pedestrian in detail since she has own will. Physicists, however, have elucidated the macroscopic collective behaviors of pedestrians by dealing them as “self-driven” particles whose destination is clear. In this paper, the author introduces the research on three fundamental pedestrian dynamics, i.e., unidirectional flow, bidirectional flow and egress process, with analysis of cellular automaton models. In congested unidirectional flow, slow rhythm improves pedestrian flow. In bidirectional flow, appropriate anticipation smooth the flow, while excessive anticipation hinders it. Furthermore, in egress process, an obstacle increases the flow if it is set at proper position.