Improving the Efficiency of the MPU-6050 Sensor Module for Inertial Drone Navigation Elkhan Sabziev Laboratory of modeling and control of continuous processes Institute of Control Systems Baku, Azerbaijan elkhan.sabziev@gmail.com https://doi.org/10.31713/MCIT.2023.023 Tahir Alizada Laboratory of methods of recognition and intelligent analysis of signals Institute of Control Systems Baku, Azerbaijan tahiralizada1973@gmail.com Namig Heydarov Innovative research group Institute of Radiation Problems Baku, Azerbaijan nmq.heydarov91@gmail.com Abstract—We investigate the problems of noise reduction in navigation data supplied by an accelerometer and a gyroscope. The MPU-6050 sensor module, which is often a component of drone navigation systems, was used in the experiments. To smooth the data, the use of "simple averaging method" and "moving average method" is recommended. We show experimentally that the data smoothing methods reduce the errors by more than 10 times. It is also shown that the "simple averaging method" can reduce the hardware load by more than 10 times compared to other data processing methods. Keywords—navigation; drone; unmanned aerial vehicle; MPU-6050; inertial sensor; accelerometer; gyroscope; noise. I. INTRODUCTION As a rule, global navigation satellite systems (GNSS) such as GPS (USA), GLONASS (Russia), BeiDou (China), Galileo (Europe) and DORIS (France) are used to determine the location of drones [1], [2], [3]. All five systems are available with the use of dedicated navigation or geodetic receivers anywhere on the Earth's surface. In addition to global navigation systems, there are regional systems: IRNSS (India) and QZSS (Japan). These satellite navigation systems can be used only in India and Japan, respectively. Despite the seemingly universal accessibility of satellite navigation, in certain conditions its use can be unreliable and sometimes impossible. One such condition is the use of drones (kamikaze drones, reconnaissance drones, etc.) in armed conflict areas. Satellite navigation signals in such areas are jammed by special radio electronic systems. Thus, there is a need to develop alternative navigation systems. Among alternative navigation systems with respect to radio electronic (including satellite) systems, inertial navigation is considered to be the most effective [3], [4], [5], [6]. In recent years, many miniaturized, relatively cheap and quite accurate inertial sensors (accelerometers and gyroscopes) have appeared on the world market. It is especially convenient to use them as part of a single module, such as MPU-6000 or MPU- 6050 (Fig. 1) [7]. Figure 1. MPU-6050 sensor module [8] II. PROBLEM STATEMENT In this paper, we aim to improve the efficiency of the MPU-6050 sensor module by selecting the optimal sampling frequency F of inertial sensors for collecting navigation data and the amount of navigation data N taken for averaging (in order to reduce the noise interference). Preliminary studies have shown that both parameters have an impact on the system and its result. For instance, an overestimation of the F parameter leads to a strain on the computational resources of the system, while an underestimation reduces the accuracy of the computed navigation data due to the influence of noise. An overestimation of the parameter N in turn also overloads the CPU of the system, while an underestimation increases the influence of noise. Thus, in order to find the optimal values of the parameters F and N, appropriate computational experiments should be carried out. For this purpose, we built a hardware unit with the MPU-6050 module and the ability to connect to a computer via USB 2.0 Type A connector. Special software was also developed to control this hardware unit and record data from the