The scene of large factories is complex, and the requirements for production safety are higher. At present, complex factories mainly use manual operation mode to carry out safety inspection, that is, manual inspection and manual recording mode to inspect factory equipment, which has the disadvantages of high labor intensity, low work efficiency and unstable inspection quality. In recent years, with the development of robot navigation related technologies, it has provided a new idea for solving autonomous navigation in large-scale and complex factory environment with variable obstacles. Therefore, we design a slam and path planning method based on multi-sensor fusion, including the establishment of complex and large-scale factory maps, dynamic obstacle avoidance and local path planning, which can achieve accurate and efficient autonomous positioning and effectively generate smooth and safe paths. The security, smoothness, flexibility and efficiency of this method are verified in various complex simulation scenarios and challenging real world. It is proved that this method can meet the performance requirements of positioning and path planning in large-scale complex industrial scenes, and has a good application prospect.