Human motion collected in real-world environments without instruction from researchers - or natural motion - is an understudied area of the field of motion capture that could increase the efficacy of assistive devices such as exoskeletons, robotics, and prosthetics. With this goal in mind, a natural motion dataset is presented in this thesis alongside algorithms for analyzing human motion. The dataset contains more than 36 hours of inertial motion capture data collected while the 16 participants went about their lives. The participants were not instructed on what actions to perform and interacted freely with real-world environments such as a home improvement store and a college campus. We apply our dataset in two experiments. The first is a study into how manual material handlers lift and bend at work, and what postures they tend to use and why. Workers rarely used symmetric squats and infrequently used symmetric stoops typically studied in lab settings. Instead, they used a variety of different postures that have not been well-characterized such as one-legged lifting and split-legged lifting. The second experiment is a study of how to infer human motion using limited information. We present methods for inferring human motion from sparse sensors using Transformers and Seq2Seq models. We found that Transformers perform better than Seq2Seq models in producing upper-body and full-body motion, but that each model can accurately infer human motion for a variety of postures like sitting, standing, kneeling, and bending given sparse sensor data. / Master of Science / To better design technology that can assist people in their daily lives, it is necessary to better understand how people move and act in the real-world with little to no instruction from researchers. Personal assistants such as Alexa and Google Assistant have benefited from what researchers call natural language processing. Similarly, natural motion processing will be useful for everyday assistive devices like prosthetics and exoskeletons. Unscripted human motion in real-world environments - or natural motion - has been made possible with recent advancements in motion capture technology. In this thesis, we present data from 16 participants who wore a suit that captures accurate human motion. The dataset contains more than 36 hours of unscripted human motion data in real-world environments that is usable by other researchers to develop technology and advance our understanding of human motion. In addition, we perform two experiments in this thesis. The first is a study into how manual material handlers lift and bend at work, and what postures they tend to use and why. The second is a study into how we can determine what a person's body is doing with a limited amount of information from only a few sensors. This study could be useful for making commercial devices like smartphones, smartwatches, and smartglasses more valuable and useful.

MScEng / Thesis (MScEng (Mechanical and Mechatronic Engineering))--University of Stellenbosch, 2009. / Clinical gait analysis has been proven to greatly improve treatment planning and monitoring of patients suffering from neuromuscular disorders. Despite this fact, it was found that gait analysis is still largely underutilised in general patient-care due to limitations of gait measurement equipment. Inertial motion capture (IMC) is able to overcome many of these limitations, but this technology is relatively untested and is therefore viewed as adolescent. This study addresses this problem by evaluating the validity and repeatability of gait parameters measured with a commercially available, full-body IMC system by comparing the results to those obtained with alternative methods of motion capture. The IMC system’s results were compared to a trusted optical motion capture (OMC) system’s results to evaluate validity. The results show that the measurements for the hip and knee obtained with IMC compares well with those obtained using OMC – with coefficient-of-correlation (R) values as high as 0.99. Some discrepancies were identified in the ankle-joint validity results. These were attributed to differences between the two systems with regard to the definition of ankle joint and to non-ideal IMC system foot-sensor design. The repeatability, using the IMC system, was quantified using the coefficient of variance (CV), the coefficient of multiple determination (CMD) and the coefficient of multiple correlation (CMC). Results show that IMC-recorded gait patterns have high repeatability for within-day tests (CMD: 0.786-0.984; CMC: 0.881-0.992) and between-day tests (CMD: 0.771-0.991; CMC: 0.872-0.995). These results compare well with those from similar studies done using OMC and electromagnetic motion capture (EMC), especially when comparing between-day results. Finally, to evaluate the measurements from the IMC system in a clinically useful application, a neural network was employed to distinguish between gait strides of stroke patients and those of able-bodied controls. The network proved to be very successful with a repeatable accuracy of 99.4% (1/166 misclassified). The study concluded that the full-body IMC system produces sufficiently valid and repeatable gait data to be used in clinical gait analysis, but that further refinement of the ankle-joint definition and improvements to the foot sensor are required.

Accurate risk assessment tools and methods are necessary to understand the relationship between occupational exposure to physical risk factors and musculoskeletal disorders. Ergonomists typically consider direct measurement methods to be the most objective and accurate of the available tools. However, direct measurement methods are often not used due to cost, practicality, and worker/workplace disruption. Inertial measurement units (IMUs), a relatively new direct measurement technology used to assess worker kinematics, are attractive to ergonomists due to their small size, low cost, and ability to reliably capture information across full working shifts. IMUs are often touted as a field-capable alternative to optical motion capture systems (OMCs). The error magnitudes of IMUs, however, can vary significantly (>15°) both within and across studies. The overall goals of this thesis were to (i) provide knowledge about the capabilities and limitations of IMUs in order to explain the inconsistencies observed in previous studies that assessed IMU accuracy, and (ii) provide guidance for the ergonomics community to leverage this technology. All three studies in this dissertation systematically evaluated IMUs using a repetitive material transfer task performed by thirteen participants with varying movement speeds (15, 30, 45 cycles/minute) and magnetic disturbance (absent, present). An OMC system was used as the reference device. This first study systematically evaluated the effects of motion speed and magnetic disturbance on the spatial orientation accuracy of an inertial measurement unit (IMU) worn on the hand. Root-mean-square differences (RMSD) exceeded 20° when inclination measurements (pitch and roll) were calculated using the IMU’s accelerometer. A linear Kalman filter and a proprietary, embedded Kalman filter reduced inclination RMSD to < 3° across all movement speeds. The RMSD in the heading direction (i.e., about gravity) increased (from < 5° to 17°) under magnetic disturbance. The linear Kalman filter and the embedded Kalman filter reduced heading RMSD to < 12° and < 7°, respectively. This study indicated that the use of IMUs and Kalman filters can improve inclinometer measurement accuracy. However, magnetic disturbances continue to limit the accuracy of three-dimensional IMU motion capture. The goal of the second study was to understand the capability of IMU inclinometers to improve estimates of angular displacements and velocities of the upper arm. RMSD and peak displacement error exceeded 11° and 28° at the fastest transfer rate (45 cycles/min) when upper arm elevation was calculated using the IMU accelerometer. The implementation of a Kalman filter reduced RMS and peak errors to < 1.5° and < 2.3°, respectively. Similarly, the RMS and peak error for accelerometer-derived velocities exceeded 81°/s and 221.3°/s, respectively, at the fastest transfer rate. The Kalman filter reduced RMS and peak errors to < 9.2°/s and < 25.1°/s, respectively. The third study was conducted to evaluate the relationship between magnetic field strength variation and magnetic heading deviation. In this study, the presence of the metal plate increased magnetic heading deviations from < 12° (90th-10th percentile) to approximately 30°. As expected, the magnetic field strength standard deviation increased from 1.0uT to 2.4uT. While this relationship may differ across other sources of magnetic disturbance, the results reinforce the notion that local magnetic field disturbances should be minimized when using IMUs for human motion capture. Overall, the findings from this thesis contribute to the ergonomics community’s understanding of the current capabilities and limitations of IMUs. These studies suggest that while the touted capabilities of the IMUs (full-body motion capture in workplace settings) may be unattainable based on current sensor technology, these sensors are still significantly more accurate than the accelerometer-based inclinometers commonly used by ergonomists to measure motions of the upper arms.

Joint center location is the driving parameter for determining the kinematics, and later kinetics, associated with human motion capture. Therefore the accuracy with which said location is determined is of great import to any and all subsequent calculation and analysis. The most significant barrier to accurate determination of this parameter is soft tissue artifact, which contaminates the measurements of on-body measurement devices by allowing them to move relative to the underlying rigid bone. This leads to inaccuracy in both bone pose estimation and joint center location. The complexity of soft tissue artifact (it is nonlinear, multimodal, subject-specific, and trial specific) makes it difficult to model, and therefore difficult to mitigate. This thesis proposes a novel method, termed Single Frame Optimization, for determining joint center location (though mitigation of soft tissue artifact) via a linearization approach, in which the optimal vector relating a joint center to a corresponding inertial sensor is calculated at each time frame. This results in a time-varying joint center location vector that captures the relative motion due to soft tissue artifact, from which the relative motion could be isolated and removed. The method’s, and therefore the optimization’s, driving assumption is that the derivative terms in the kinematic equation are negligible relative to the rigid terms. More plainly, it is assumed that any relative motion can be assumed negligible in comparison with the rigid body motion in the chosen data frame. The validity of this assumption is investigated in a series of numerical simulations and experimental investigations. Each item in said series is presented as a chapter in this thesis, but retains the format of a standalone article. This is intended to foment critical analysis of the method at each stage in its development, rather than solely in its practical (and more developed) form.

A realtime wireless motion tracking system is developed. The system is capable of tracking the orientations of multiple wireless sensors, using a semi-distributed implementation to reduce network bandwidth and latency, to produce real-time animation of rigid body models, such as the human skeleton. The system has been demonstrated to be capable of full-body posture tracking of a human subject using fifteen devices communicating with a basestation over a single, low bandwidth, radio channel. The thesis covers the theory, design, and implementation of the tracking platform, the evaluation of the platform’s performance, and presents a summary of possible future applications.

Cieľom tejto práce je navrhnúť, zhotoviť a implementovať vlastný systém pre zachytávanie pohybu založený na inerciálnych meracích jednotkách. V rámci budovania teoretického základu bolo preskúmaných viacero metód, avšak primárne bola pozornosť venovaná samotnému inerciálnemu snímanu - jeho kladom a nedostatkom, kľúčovým vlastnostiam a jednotlivým komponentom potrebným pre zostrojenie systému na jeho báze. Tento úvodný zber informácií je nasledovaný fázami návrhu, zhotovenia a zhodnotenia, ktoré sa zaoberajú procesom vývoja a testovania daného riešenia. Hlavným prínosom realizácie systému je zostrojenie zariadení pre snímanie pohybu - jedná sa o malé, ľahké, batériovo napájané zariadenia, ktoré sú kompletne bezdrôtové, či už z hľadiska komunikácie s okolitým svetom, alebo vďaka napájaniu kompatibilnému so štandardom Qi.