Abstract: Aiming at the problem that the number of cone obstacles obtained in the view of a curved racing track is small for driverless FSAC racing cars for college students in China under a high-speed turning condition, which is easy to lead to failure of path planning, this paper designs a dynamic vision algorithm based on lidar and inertial navigation. The algorithm detects the cone obstacles based on the information of lidar. At the same time, the algorithm obtains the change of the heading angle by using inertial navigation solution, and performs weighted moving average filtering on the change of the heading angle to predict its change trend at the next moment. Combined with the speed and wheel angle information, the algorithm calculates and adjusts the range of the region of interest so as to keep the perceptual vision in the center of the racing track and obtain more information of cone obstacles at the turning. The algorithm is verified by real vehicle experiments. The results show that it effectively improves the perception efficiency and the path planning ability of the car.