Filtering MEMS Gyroscope Data: A Comparative Analysis of Noise Reduction Methods : Student paper

This study compares three filtering techniques for removing noise in the MEMS MPU-6050 gyroscope, including Kalman, Bessel, and Exponential Moving Average filters. Parameter selection was conducted through experimental, iterative testing for every algorithm. The UAV (drone) experiment with a mounted MPU-6050 gyroscope obtained 171 raw, noisy tilt angle measurements around the x-axis. The UAV hovered above the ground while taking angle measurements, and the effects of high vibrations from the drone’s motor to MPU-6050 were observed with filter solutions. Analysis techniques like variance and cross-correlation with quadratic interpolation showed the preferable filtering solution in dynamic scenarios where latency and smoothness are critical.