Drift-Free Position Estimation for Periodic Movements Using Inertial Units

Abstract

Latest advances in microelectromechanical systems have made inertial units (IUs) a powerful tool for human motion analysis. However, difficulties in handling their output signals must be overcome. The purpose of this study was to develop the novel "PB-algorithm" based on polynomial data fitting, splines interpolation, and the wavelet transform, one after the other, to cancel drift disturbances in position estimation for periodic movements. High-accuracy position measurements from an optical system (Vicon Nexus 1.0) were used to validate the proposed method and comparison with another drift-correction algorithm was provided. Results indicate the accuracy with respect to the Vicon's reference signal (euclidean error lower than 54.62 × 10(-3) m and correlation coefficient higher than 0.968). A reduction of the root-mean-square error of 68.74% was obtained when the proposed two-step method was compared with a modified-band limited Fourier linear combiner. All methods were applied to data from the 30-s chair stand test, which is one of the most used clinical tests dealing with lower body strength assessment, falls prediction, and gait disorders in older adults. The relevance of this study is that after cancelling drift disturbances, and obtaining an accurate Z-position estimation, it is possible to evaluate the sit-to-stand and stand-to-sit transitions from the whole test.

The Validity of a Mixed Reality-Based Automated Functional Mobility Assessment

“…derived from the HoloLens and Opal sensors, we found good to excellent signal agreement for the majority of STS and TUG measures (Xcor 0.74-0.99, NRMSE ~10%), with better signal agreement observed in STS task, as each sit-to-stand cycle allows for displacement calibration adjustment (Zero Displacement Update) [24]. For sensor signal comparison in TUG task, however, due to the integration and drift error associated with IMU sensor and lack of viable calibration during the walking period (10s or longer), the vertical displacement derived from IMU sensors was biased in comparison to HoloLens output ( Fig.2e), resulting unsatisfactory NRMSE measures (14.07-19.56%).…”

“…STS and TUG tests, the gravity corrected vertical (VT) acceleration data from Opal sensors were double integrated over time to obtain the vertical displacement, with drift and integration error corrected using 1) a high pass filter (4th order Butterworth, 0.1Hz)[23] and 2) drift correction under the assumption that participants reach the same height when they make contact with the chair (zero displacement update-ZDU)[24]. The processed VT displacement from HoloLens and Opal sensors was then time-aligned using cross correlation analysis (calculating the similarity and time lag between signal).…”

Improving Displacement Measurement for Evaluating Longitudinal Road Profiles

“…The works carried out by the researchers in [19]- [25] confirm the negative effect of drift in applications that use IMU sensors to derive displacement. Various solutions were proposed in these studies to minimize the effect of this drift in their respective applications.…”

“…This technique works by estimating an error offset on the evaluated displacement, based on a mathematical model that integrates the measurements obtained from other related sensors. The correction models in [19], [21] are targeted to applications where the motion of the object is periodic, and its theory basically relies on pre-existing knowledge of the expected frequencies or band.…”

The Removal of EOG Artifacts From EEG Signals Using Independent Component Analysis and Multivariate Empirical Mode Decomposition

“…Consequently, the MEMD is superior to other signals processing method such as Fourier and wavelet transformation [20,30,31]. Moreover, it is demonstrated that the MEMD is clearly suitable for processing non-deterministic and non-stationary electrophysiological signals because the MIMFs decomposed by MEMD can reflect the local characteristics of signal at any time [32].…”