Abstract: An autonomous navigation method of robot in dynamic unknown environment is presented. Human's assistance is used instead of particular description of map. There are two phases in this method, which are human's guiding phase and autonomous navigation phase. In phase of human's guiding, a rough polar coordinate map of local environment is obtained by multi sensor information fusion and through it the global map is obtained. The method of eliminating sensor data error is also given here. According to map obtained in guiding phase, a robot moves in dynamic environment in the phase of autonomous navigation. Constraint condition of robot's motion control and dynamic obstacle avoiding method are also presented. Robot can dispose of emergent obstacles in this method, and optimization of path is also attained. Experiment results show the effectiveness of this method.