Shaghayegh Zihajehzadeh

Integration of MEMS Inertial and Pressure Sensors for Vertical Trajectory Determination

Integration of a low-cost global positioning system (GPS) with a microelectromechanical system-based inertial measurement unit (MEMS-IMU) is a widely used method that takes advantage of the individual superiority of each system to get a more accurate and robust navigation performance. However, because of poor observations as well as multipath effects, the GPS has low accuracy in the vertical direction. As a result, the navigation accuracy even in an integrated GPS/MEMS-IMU system is more challenged in the vertical direction than the horizontal direction. To overcome this problem, this paper investigates the integration of a MEMS barometric pressure sensor with the MEMS-IMU for vertical position/velocity tracking without the GPS that has applications in sports. A cascaded two-step Kalman filter consisting of separate orientation and position/velocity subsystems is proposed for this integration. Slow human movements in addition to more rapid sport activities such as vertical and step-down jumps can be tracked using the proposed algorithm. The height-tracking performance is benchmarked against a reference camera-based motion-tracking system and an error analysis is performed. The experimental results show that the vertical trajectory tracking error is less than 28.1 cm. For the determination of jump vertical height/drop, the proposed algorithm has an error of less than 5.8 cm.

A cascaded two-step Kalman filter for estimation of human body segment orientation using MEMS-IMU.

Orientation of human body segments is an important quantity in many biomechanical analyses. To get robust and drift-free 3-D orientation, raw data from miniature body worn MEMS-based inertial measurement units (IMU) should be blended in a Kalman filter. Aiming at less computational cost, this work presents a novel cascaded two-step Kalman filter orientation estimation algorithm. Tilt angles are estimated in the first step of the proposed cascaded Kalman filter. The estimated tilt angles are passed to the second step of the filter for yaw angle calculation. The orientation results are benchmarked against the ones from a highly accurate tactical grade IMU. Experimental results reveal that the proposed algorithm provides robust orientation estimation in both kinematically and magnetically disturbed conditions.

UWB-Aided Inertial Motion Capture for Lower Body 3-D Dynamic Activity and Trajectory Tracking

This paper introduces a novel method for simultaneous 3-D trajectory tracking and lower body motion capture (MoCap) under various dynamic activities such as walking and jumping. The proposed method uses wearable inertial sensors fused with an ultrawideband localization system using a cascaded Kalman filter-based sensor fusion algorithm, which consists of a separate orientation filter cascaded with a position/velocity filter. In addition, to further improve the joint angle tracking accuracy, anatomical constraints are applied to the knee joint. Currently, available self-contained inertial tracking methods are not only drift-prone over time but also their accuracy is degraded under fast movements with unstable ground contact states due to the lack of external references. However, our experimental results, which benchmark the system against a reference camera-based motion tracking system, show that the proposed fusion method can accurately capture the dynamic activities of a subject without drift. The results show that the proposed system can maintain similar accuracies between fast and slow motions in lower body MoCap and 3-D trajectory tracking. The obtained accuracies of the system for 3-D body localization and knee joint angle tracking for fast motions were less than 5 cm and 2.1°, respectively.

Robust Biomechanical Model-Based 3-D Indoor Localization and Tracking Method Using UWB and IMU

This paper proposes a robust sensor fusion algorithm to accurately track the spatial location and motion of a human under various dynamic activities, such as walking, running, and jumping. The position accuracy of the indoor wireless positioning systems frequently suffers from non-line-of-sight and multipath effects, resulting in heavy-tailed outliers and signal outages. We address this problem by integrating the estimates from an ultra-wideband (UWB) system and inertial measurement units, but also taking advantage of the estimated velocity and height obtained from an aiding lower body biomechanical model. The proposed method is a cascaded Kalman filter-based algorithm where the orientation filter is cascaded with the robust position/velocity filter. The outliers are detected for individual measurements using the normalized innovation squared, where the measurement noise covariance is softly scaled to reduce its weight. The positioning accuracy is further improved with the Rauch–Tung–Striebel smoother. The proposed algorithm was validated against an optical motion tracking system for both slow (walking) and dynamic (running and jumping) activities performed in laboratory experiments. The results show that the proposed algorithm can maintain high accuracy for tracking the location of a subject in the presence of the outliers and UWB signal outages with a combined 3-D positioning error of less than 13 cm.

Regression Model-Based Walking Speed Estimation Using Wrist-Worn Inertial Sensor

Walking speed is widely used to study human health status. Wearable inertial measurement units (IMU) are promising tools for the ambulatory measurement of walking speed. Among wearable inertial sensors, the ones worn on the wrist, such as a watch or band, have relatively higher potential to be easily incorporated into daily lifestyle. Using the arm swing motion in walking, this paper proposes a regression model-based method for longitudinal walking speed estimation using a wrist-worn IMU. A novel kinematic variable is proposed, which finds the wrist acceleration in the principal axis (i.e. the direction of the arm swing). This variable (called pca-acc) is obtained by applying sensor fusion on IMU data to find the orientation followed by the use of principal component analysis. An experimental evaluation was performed on 15 healthy young subjects during free walking trials. The experimental results show that the use of the proposed pca-acc variable can significantly improve the walking speed estimation accuracy when compared to the use of raw acceleration information (p<0.01). When Gaussian process regression is used, the resulting walking speed estimation accuracy and precision is about 5.9% and 4.7%, respectively.

A Novel Biomechanical Model-Aided IMU/UWB Fusion for Magnetometer-Free Lower Body Motion Capture

The available human body inertial motion capture (MoCap) systems are aided by magnetometers to remove the drift error in yaw angle estimation, which in turn limits their application in the presence of long-lasting magnetic disturbances in indoor environments. This paper introduces a magnetometer-free algorithm for lower-body MoCap including 3-D localization and posture tracking by fusing inertial sensors with an ultrawideband (UWB) localization system and a biomechanical model of the human lower body. Using our novel Kalman filter-based fusion algorithm, the UWB localization data aided by the biomechanical model can eliminate the drift in inertial yaw angle estimation of the lower-body segments. This magnetometer-free yaw angle estimation makes the algorithm insensitive to the magnetic disturbances. The algorithm is benchmarked against the optical MoCap system for various indoor activities including walking, jogging, jumping, and stairs ascending/descending. The results show that the proposed algorithm outperforms the available magnetometer-aided algorithms in yaw angle tracking under magnetic disturbances. In a uniform magnetic field, the algorithm shows similar accuracies in localization and joint angle tracking when compared with the magnetometer-aided methods. The localization accuracy of the proposed method is better than 4.5 cm in a 3-D space and its accuracy for knee angle tracking is about 3.5° and 4.5° in low and high dynamic motions, respectively.

Fitness activity classification by using multiclass support vector machines on head-worn sensors

Fitness activity classification on wearable devices can provide activity-specific information and generate more accurate performance metrics. Recently, optical head-mounted displays (OHMD) like Google Glass, Sony SmartEyeglass and Recon Jet have emerged. This paper presents a novel method to classify fitness activities using head-worn accelerometer, barometric pressure sensor and GPS, with comparisons to other common mounting locations on the body. Using multiclass SVM on head-worn sensors, we obtained an average F-score of 96.66% for classifying standing, walking, running, ascending/descending stairs and cycling. The best sensor location combinations were found to be on the ankle plus another upper body location. Using three or more sensors did not show a notable improvement over the best two-sensor combinations.

Experimental evaluation of regression model-based walking speed estimation using lower body-mounted IMU

This study provides a concurrent comparison of regression model-based walking speed estimation accuracy using lower body mounted inertial sensors. The comparison is based on different sets of variables, features, mounting locations and regression methods. An experimental evaluation was performed on 15 healthy subjects during free walking trials. Our results show better accuracy of Gaussian process regression compared to least square regression using Lasso. Among the variables, external acceleration tends to provide improved accuracy. By using both time-domain and frequency-domain features, waist and ankle-mounted sensors result in similar accuracies: 4.5% for the waist and 4.9% for the ankle. When using only frequency-domain features, estimation accuracy based on a waist-mounted sensor suffers more compared to the one from ankle.

Pedestrian Dead Reckoning With Smartglasses and Smartwatch

Wearable miniature inertial sensors have been widely used for pedestrian dead reckoning (PDR). Typical low-cost PDR systems use sensors attached to either the human trunk or feet. The recent emergence of smartglasses and smart watches provides an opportunity to use both types of wearable devices in position tracking. This paper proposes a novel method of utilizing both a smartwatch and smartglasses for PDR. The general idea is to use the relative angle between arm swing direction and head direction to detect any head-turn motion that would otherwise skew the position dead reckoning propagation. A complete PDR solution that includes step detection, step length estimation, head-rotation detection, and dead reckoning using a smartwatch and smartglasses that are currently available in the market is presented. Using the smartglasses, step detection with an error rate less than 0.4% and a cumulative distance error of less than 2.4% on 800 m walks and runs is achieved. In the dead reckoning field experiments, the proposed algorithm produces result that closely track the actual path when plotted on Google Maps, outperforming solutions that only use the smartwatch or smartglasses alone.

Adaptive Kalman filter for indoor localization using Bluetooth Low Energy and inertial measurement unit.

This paper proposes a novel indoor localization method using the Bluetooth Low Energy (BLE) and an inertial measurement unit (IMU). The multipath and non-line-of-sight errors from low-power wireless localization systems commonly result in outliers, affecting the positioning accuracy. We address this problem by adaptively weighting the estimates from the IMU and BLE in our proposed cascaded Kalman filter (KF). The positioning accuracy is further improved with the Rauch-Tung-Striebel smoother. The performance of the proposed algorithm is compared against that of the standard KF experimentally. The results show that the proposed algorithm can maintain high accuracy for position tracking the sensor in the presence of the outliers.