Abstract: In view of characteristics of no GPS signal, low illumination and structured environment in coal mine, a real-time pose estimation method for underground unmanned aerial vehicle (UAV) based on iterative closest point (ICP) was proposed. By establishing quadrotor UAV motion model and airborne laser radar observation model, the problem of position estimation of underground quadrotor UAV is converted into scanning matching problem of airborne laser point cloud data. 3D laser radar is used as airborne environment measurement sensor of quadrotor UAV, and observation point cloud data in current position of the UAV is obtained. Taking the first frame position as initial position,relative transformation matrix between two consecutive points of point cloud data is obtained by ICP method, and the continuous key frame point cloud data is solved iteratively to obtain the real-time pose estimation result of quadrotor UAV in underground coal mine.Filtering and downsampling methods are used to optimize point cloud data, and the solution of transformation matrix is accelerated to meet the real-time requirements of position estimation of quadrotor UAV. The experimental results show that the ICP-based real-time pose estimation method for underground UAV can quickly and effectively solve the pose of quadrotor UAV, and compared with normal distribution transform method, the ICP method is more suitable for real-time pose estimation of quadrotor UAV in underground coal mine.