Abstract: Abstract: Dried fruits should be stored in warehouse by placing and stacking on the shelves. By using intelligent forklift to store and take goods on the shelves, the warehouse efficiency could be effectively solved, and the warehouse management of dried fruits could be promoted to be standardized and intelligent. Obstacle detection is the primary guarantee for the safe operation of intelligent forklifts, and the detection effect is also related to the efficient operation of intelligent forklifts in warehouse, as a key technology of intelligent vehicles, it has gradually become a research hot topic. However, the current researches focuse on small multi-degree-freedom intelligent vehicles, there is no research on obstacles detection methods for large intelligent forklift in dried fruit warehouse. Considering the limitations of warehouse layout, the detection region of traditional detection methods were mostly fixed shape, that means that the safety distance was fixed, so it was more suitable when forklift going straight in an open space, on the contrary, in the dried fruit warehouse with limited channel width, especially when turning, there would be false alarm, which would easily cause the large intelligent forklift to misjudge the objects that could be bypassed into potential obstacles, thus causing the forklift to change the road or stop sharply. In order to solve the false detection and realize the obstacle dynamic detection for large intelligent forklift in dried fruit warehouse, taking the reversing process of intelligent forklift as an example, an obstacles dynamic detection method based on dynamic change of laser scanning area with the speed and steering angle of large intelligent forklift in dried fruit warehouse was proposed in this paper. The real-time position and direction information of forklift in the global Cartesian coordinate system of warehouse was obtained by using on-board laser sensor SICK-NAV350, combining with the motion geometry model of forklift, the horizontal laser ranging sensor (SICK-LMS111) and the inclined laser ranging sensor (SICK-TIM561) scanning the obstacle in 2 planes, forming a dynamic detection area changing with the speed and steering angle of forklift. The real vehicle test results showed that the proposed method without error checking, the error detection rate of the sector method was 50.00% and that of the rectangle method was 10.00% in the testing of horizontal scan ranging sensor, the error detection rate of the sector method was 30.77% and that of the rectangle method was 69.23% in the testing of tilted scan ranging sensor. With tilted scanning range sensor as the auxiliary and horizontal scanning range sensor as the main part, a dynamic obstacle detection area based on fusion of 2 planes was formed, the minimum height of obstacles could be detected was about 31 mm when the installation angle of sensor SICK- TIM561 was 25.60°, the height to ground was 2 000 mm and the detection region width was set to 2 183 mm. The proposed method effectively solved the false alarm of intelligent forklift when driving in the warehouse, and was more suitable for warehousing and transportation than the traditional detection method, and improved the mobility and safety of intelligent forklift in warehouse. The research can provide reference for obstacle detection of large warehouse intelligent transport vehicles.