The propose of this paper is position tracking and map establishment of Autonomous Guided Vehicle (AGV). Odometry is widely applied to track AGV position but it exists cumulative errors. Map offsets is presence if cumulative errors not corrected immediately. To solve this problem, multi-sensors are fused in this proposal. Firstly, an inertia measurement unit(IMU) is applied for providing the odometry with a correct driving direction. Moreover, the iterative closest point algorithm is applied to match the scanned information of Laser Range Finder(LRF) over time. Meanwhile, the results of scanning registration are used to eliminate odometry errors and updated the new information into global map. In experiment process, a simulation software called Virtual Robot Experiment Platform is firstly used to simulate the proposed algorithm before real experiment. It can make sure the hardware system can work well which will reduce the time on algorithm validation and software fault debugging time of practical testing.