ARM-Based Denoising of the Accelerometer and Gyroscope via Lifting Wavelet Transform

The Inertial Navigation System (INS) is a fundamental navigational apparatus, employing a processor in conjunction with motion and rotation sensors (accelerometers and gyroscopes) to continually compute a vehicle's positional coordinates, orientation, and velocity devoid of external reference points. Over time, however, the system's precision degrades due to accumulated sensor noise. To address this, the lifting scheme algorithm, specifically the second-generation wavelet variation, is deployed as an effective noise reduction method, known for its low computational demands amenable to real-time implementation. This paper aims to implement the lifting design algorithm on ARM hardware, targeting low-cost INS sensors to enhance performance. Empirical results demonstrate substantial 30% and 41% improvements in accelerometer and gyroscope performance in real-time scenarios. Furthermore, hardware and software calculation comparisons reveal the RMSE and MAE are X,Y for accelerometer measurements and X,Y for gyroscope, respectively.