Abstract: In view of problem of autonomous location of inspection robot without GPS in underground coal mine, a method of simultaneous localization and mapping based on lidar was studied. Firstly, the observation model of lidar and prediction model of odometer are established, and the actual problems of robot localization and mapping are transformed into the logical reasoning problems of probabilistic mathematical model. At the same time, the adaptive Monte Carlo localization algorithm is used to estimate the real-time pose of the robot,the resampling method based on particle weight(maps matching degree) is proposed to remove particles with small weight, accurate representation of posterior probability distribution of robot posture with fewer and better particles is realized, requirement of using sensors to realize the real-time positioning of robots on raster maps is met. Fast-SLAM algorithm is optimized to reduce the number of particles, and mitigate particle dissipation,so as to improve accuracy of mapping. The experimental results show that the method effectively solves the problem of real-time pose estimation and environment mapping of inspection robot, and improves the self-adaptability of robot localization and accuracy of mapping combining with adaptive Monte Carlo localization algorithm and optimized Fast-SLAM algorithm.