In order to improve the accuracy of simultaneous localization and mapping (SLAM) based on laser radar, an efficient and high-precision SLAM framework based on factor graph is proposed in this paper. A factor graph method based on sliding window is used to match the current frame to get the relative pose, and then the key frame is selected according to certain rules, and the absolute pose is obtained by matching the key frame with the global map. A factor graph is constructed, and the relative pose between consecutive frames and the absolute pose of key frames are taken as optimization factors, and the pose of the robot is taken as a state node to be put into the factor graph for pose optimization, so that the pose of the robot with high frequency and a globally consistent environment map are obtained. The results show that the algorithm can reduce the accumulation of errors and has higher positioning accuracy.