Aiming at the problem that automated guided vehicle (AGV) is difficult to locate accurately due to the influence of environment and time drift when it works in the indoor intelligent storage system. In this paper, an extended kalman filtering (EKF) framework is designed. In order to make full use of the original ranging values of ultra wideband (UWB) and inertial measurement unit (IMU), the framework realizes the fusion positioning between UWB module and IMU module in a tight coupling manner, so as to ensure that the system can still work when the available base station signal is inaccurate. Firstly, for the problem that the traditional UWB positioning method is easily affected by the non-line of sight (NLOS) error indoors, the calculated positioning coordinate value is unstable. With the help of different NLOS probability distribution curves of different obstacles, the weighted least square method is applied to the UWB positioning method to determine the positioning coordinate value of UWB, which improves the sudden change of AGV positioning coordinate in the static environment. Then the data fusion algorithm is optimized, and the error value of IMU and UWB coordinate is taken as the observation value of EKF, which reduces the influence of cumulative error on IMU positioning results, provides the global optimal estimation of the system optimal state, and improves the fusion positioning accuracy. Finally, the measured data of UWB and IMU systems in indoor complex environment are simulated in MATLAB. The experimental results show that when NLOS signal seriously affects the positioning effect, the UWB and IMU combined positioning system can provide more reliable positioning results than the single IMU positioning system. It improves the positioning accuracy of AGV and provides a new idea for indoor positioning mode.