Specific rehabilitation is a key period for a lower-limb amputee patient. While learning how to walk with a prosthesis, the patient needs to avoid any gait compensations that may lead to future comorbidities. To reach a gait pattern close to the one of a healthy person, objective data may be of great help to complement the experience of the clinician team. By using 6 IMUs located on the feet, shanks and thighs accompanied by 3 accelerometers on the pelvis, sternum and head, data could be recorded during walking exercises of 7 rehabilitation sessions of a patient. To compute the absolute symmetry index of the stance phase duration and the stride duration all over the instrumented sessions, the gait events defining the transitions between gait phases were determined thanks to several algorithms. By first comparing the error obtained in the calculation of the stance phase duration with all tested algorithms as compared to the data from pressure insoles considered as a reference system, the algorithm developed by Trojaniello and collaborators [1] was found to be the most adapted to this situation. Using this algorithm on the data from all sessions highlighted the possibility to detect changes in the symmetry of stance phase duration and stride duration, that are relative to the gait quality. This means that IMUs seem to be able to monitor the progress of a patient during his rehabilitation. Hence, IMUs have proven themselves to be a system of great interest in the analysis of the gait pattern of a lower-limb amputee patient in rehabilitation, by allowing for an embedded measurement of much more parameters than the pressure insoles, whose calibration constituted a real limitation.

Human motion capture technology is a powerful tool for advancing the understanding of human motion biomechanics (Andriacchi and Alexander, 2000). This is most readily accomplished by applying retroreflective markers to a participant’s skin and tracking the position of the markers during motion. Skin and adipose tissue move independently of the underlying bone during motion creating error known as soft tissue artifact (STA), the primary source of error in human motion capture (Leardini et al., 2005). (Solav et al., 2014) proposed and (Solav et al., 2015) expanded the triangular Cosserat point element (TCPE) method to reduce the effect of STA on derived kinematics through application of a marker cluster analyzed as a set of triangular Cosserat point elements. This method also provides metrics for three different modes of STA. Here the updated TCPE method (Solav et al., 2015) was compared to the established point cluster (PC) method of (Andriacchi et al., 1998) and the marker position error minimizing Procrustes solution (PS) method of (Söderkvist and Wedin, 1993) in two implant-based simulations, providing quantitative measures of error, and standard gait analysis, providing qualitative comparisons of each method’s determined kinematics. Both of these experiments allowed the TCPE method to generate observed STA parameters, informing the efficacy of the simulation. The TCPE method’s performance was similar to the PS method’s in the implant simulations and in standard gait. The PC method’s results seemed to be affected by numerical instability: simulation trial errors were larger and standard gait results were only similar to the other methods’ in general terms. While the PS and TCPE results were comparable, the TCPE method’s physiological basis provided the added benefit of non-rigid behavior quantization through its STA parameters. In this study, these parameters were on the same order of v magnitude between the standard gait experiments and the simulations, suggesting that implant simulations could be valuable substitutes when invasive methods are not available.

At least 2.6 billion people all over the world suffer from nearsightedness, among whom 312 million are under 19 years old. Just like other vision problems, uncorrected nearsightedness brings inconvenience to many human daily activities including walking. However, the influence of nearsightedness on gait patterns and gaze behaviours remains barely discovered. This project aimed to study the influence of nearsightedness on human visuomotor coordination in different environmental settings. An integrated system combining motion capture and eye-tracking was implemented for measuring gait and gaze simultaneously. Twelve participants were recruited to perform a protocol consisting of walking tasks in various visual and environmental conditions. Nine of the participants were eligible for data analysis. Gaze time distribution and gait cycle parameters were compared between participant groups (five with normal vision, four nearsighted), and among different walking tasks. Results revealed that comparing with the control group, the nearsighted participants made shorter and slower steps, as well as spent more time looking at the walking path. The walking path also affected the gait and gaze behaviours, with shorter step length and longer step time observed when the participants were walking uphill, while increased gaze attention was paid downhill. The practicality of combining gait analysis with eye-tracking was proved in this project, laying a foundation for future studies of visuomotor coordination.

No description available.

No description available.

No description available.

The extraction and analysis of human gait characteristics using image sequences are an important area of research. Recently, the focus of this research area has turned to computer vision as an unobtrusive way to analyze human motions. The applications for such a system are wide ranging in many disciplines. For example, it has been shown that visual systems can be used to identify people by their gait, estimate a subject's kinematic configuration and identify abnormal motion. The focus of this thesis is a system that accurately classifies observed motions without the use of an explicit spatial or temporal model. The visual detection of hidden loads through passive visual analysis of gait is presented as a test of the system. The major contributions of this thesis are in two areas. The first is a neural network based scheme that classifies walking styles based on simple image metrics obtained from a single, monocular gray scale image sequence. The powerful neural network classifier utilized in this system provides an efficient, robust and highly accurate classification using these image metrics. This eliminates the need for more complex and difficult to obtain measures that are required by many of the currently human visual analysis systems. This system uses computer vision and pattern recognition techniques combined with physiological knowledge of human gait to estimate an observed subject's hip angle. The hip angle is then used to calculate a normality index of the gait. The hip angle estimate and normality index are then used as inputs to a neural network. It is shown through experiment that this system provides an accurate classification of four different walking styles observed by a single camera. Secondly, a computer vision based approach is presented that provides an accurate pose estimate without the use of an explicit spatial or temporal model. A hybrid fuzzy neural network is used to assign contour points of a silhouette to kinematically relevant groups. These labeled points are used to estimate the joint locations of the subject. The joint angles are shown to be good estimates as compared to ground truth angles provided by a motion capture system. The effectiveness of the system to distinguish between subtle gait differences is demonstrated by detecting the presence of hidden loads when carried by walking people. / Ph. D.

Exoskeletons show great promise in aiding people in a wide range of applications. One such application is medical rehabilitation and assistance of those with spinal cord injuries. Exoskeletons have the potential to offer several benefits over wheelchairs, including a reduction in the risk of upper-body injuries associated with extended wheelchair use. To fully mitigate this risk of injury, exoskeletons will need to be fully self-balancing, able to move and stand without crutches or other walking aid. To accomplish this, the Orthotic Lower-body Locomotion Exoskeleton (OLL-E) will actuate 12 Degrees of Freedom, six in each leg, using custom design linear series elastic actuators. The placement of these actuators relative to each joint axis, and the geometry of the linkage connecting them, were critical to ensuring each joint was capable of producing the required outputs for self-balancing locomotion. In pursuit of this goal, a general model was developed, relating the actuator's position and linkage geometry to the actual joint output over its range of motion. This model was then adapted for each joint in the legs and compared against the required outputs for humans and robots moving through a variety of gaits. This process allowed for the best placement of the actuator and linkages within the design constraints of the exoskeleton. The structure of the exoskeleton was then designed to maintain the desired geometry while meeting several other design requirements such as weight, adjustability, and range of motion. Adjustability was a key factor for ensuring the comfortable use of the exoskeleton and to minimize risk of injury by aligning the exoskeleton joint axes as close as possible to the wearer's joints. The legs of OLL-E can accommodate users between 1.60 m and 2.03 m in height while locating the exoskeleton joint axes within 2 mm of the user's joints. After detailed design was completed, analysis showed that the legs met all long-term goals of the exoskeleton project. / Master of Science / Exoskeletons show great promise in aiding people in a wide range of applications. One such application is medical rehabilitation and assistance of those with spinal cord injuries. Exoskeletons have the potential to offer several benefits over wheelchairs, including a reduction in the risk of upper-body injuries associated with extended wheelchair use. To best reduce this risk of injury, exoskeletons will need to be fully self-balancing, able to move and stand without crutches or relying on any other outside structure to stay upright. To accomplish this, the Orthotic Lower-body Locomotion Exoskeleton (OLL-E) will use a set of custom designed motors to apply power and control to 12 joints, six in each leg. Where these motors were placed, and how they connect to the joints they control, were critical to ensuring the exoskeleton was able to self-balance, walk, and climb stairs. To find the correct position, a set of equations was developed to determine how different positions changed each joints’ speed, strength, and range of motion. These equations were then put into a piece of custom software that could quickly evaluate different joint layouts and compare the capabilities against measurements from people and robots walking, climbing stairs, and standing up out of a chair. This process allowed for the best placement of the motors and joints while still keeping the exoskeleton relatively compact. The rest of the exoskeleton was then designed to connect these joints together, while meeting several other design requirements such as weight, adjustability, and range of motion. Adjustability was very important for ensuring the comfortable use of the exoskeleton and to minimize risk of injury by ensuring that the exoskeleton legs closely matched the movements of the person inside. The legs of OLL-E can accommodate users between 1.60 m and 2.03 m in increments of 7 mm. After detailed design was completed, additional analyses were performed to check the strength of the structure and ensure it met other long-term goals of the project.

This thesis explores the use of machine learning (ML) and smartphone sensors to improve indoordistance estimation, a critical aspect of healthcare tests like the 6-minute walk test (6MWT). In order to make tests like the 6MWT more available, and lower the barrier for a patient toget tested, there are multiple problems which need to be solved: How can the distance data needed for these tests be collected reliably and remotely, and without having to rely on the patient reporting correct data; How can these tests be performed indoors, without relying on GPS or other GNSS, which are unreliable indoors. To tackle these challenges a convolutional neural network (CNN) trained on a dataset containing continuous ground truth was employed. An enhancement of an existing CNN model was done by collecting more training data, tuning hyper parameters, and testing it on a diverse dataset. The results of this thesis shows that when predicting distance walked on data from participants the CNN model has seen before, the precision meets clinical minimum for being able to show a change in the health condition. On real world data the performance suffers. Despite limitations due to the scope of data collection, the results still underscores the potential of ML for accurate and efficient indoor distance estimation and points to future research directions. / <p></p><p></p><p></p>

No description available.