In this paper, a new simultaneous localization and mapping (SLAM) method for micro aerial vehicles (MAVs) is put forward. Its main contributions are the hybrid iterative closest points and normal distribution transform (ICP-NDT) point cloud registration algorithm as well as the extended Kalman filter (EKF) algorithm for data fusion and estimation based on the dynamic model of the quadrotor. In this method, a 2-dimensional (2D) lidar is used to obtain surrounding obstacle information in the region. Its data can be turned into displacement by the hybrid ICP-NDT registration algorithm, and projected to a planar occupancy grid submap by the imaging algorithm. The displacement can be integrated into EKF for data fusion with the other sensors to get the optimal position for the MAV, and the submap can be inserted into this optimal position for updating the map. As the process repeats, the map can establish. The presented algorithm is tested in two pieces of the real scene, and the MAV is capable of getting its position and establishing the map for the region. In these tests, the maps can reflect the planar features of the environment with satisfactory accuracy.