Title:  Towards Novel Sensing Technologies for the Ubiquitous Assessment of Joint Kinematics.

Date: Wednesday, November 16th, 2022

Time: 9:00 AM – 12:00 PM EST

Location: Zoom meeting

https://us05web.zoom.us/j/81076713249?pwd=a0dZQUxoM0dHYllNVmpwaEZZdjhjUT09&from=addon

Fereshteh Shahmiri

Computer Science. PhD Student, School of Interactive Computing

College of Computing, Georgia Institute of Technology

Committee:

Dr. W. Keith Edwards (Advisor),  School of Interactive Computing, Georgia Institute of Technology, USA

Dr. Omer T. Inan (Co-Advisor), School of Electrical and Computer Engineering, Georgia Institute of Technology, USA

Dr. Elizabeth Mynatt, Dean of Khoury College of Computer Sciences, Northeastern University

Dr. Rosa Arriaga, School of Interactive Computing, Georgia Institute of Technology, USA

Dr. Thad E. Starner, School of Interactive Computing, Georgia Institute of Technology, USA

Dr. Gierad Laput, Apple.

Summary:

Towards Novel Sensing Technologies for the Ubiquitous Assessment of Joint Kinematics

How in the field of medical sensing, tuning the accuracy based on applications and anatomical constraints can reduce the intrusiveness of the required sensing hardware?

Our body movements and postures are key indicators of our health and well-being. In many cases in our daily lives, those movements and postures need accurate, low-cost, long-term, and unobtrusive monitoring and quantitative assessments. From preventing a poor posture or a knee injury to restoring fine motor activities for rehabilitation purposes, assessing the particular poses and movements of our body joints is an indispensable requirement, and a focal point in the development of many motion tracking technologies including wearable and portable sensing devices. It is key to emphasize the fact that the notion of motion tracking, by itself, is inherently undefined and has to be scoped in a set of requirements that are imposed by each specific application and its accompanying context, restrictions and necessities. 

This thesis has taken steps towards novel sensing methodologies to ubiquitously assess body motions in various approaches and application domains. It aims to answer a key, yet unanswered question on how to set a common ground to maximize the usability factor as we tune the trade-off between (pose) motion tracking accuracy vs the excessiveness of on-body sensing instrumentation. That said, it asks how in the field of medical sensing, tuning the accuracy based on application and anatomical constraints can reduce the intrusiveness of the required sensing hardware.

To answer this key question, this thesis adopted a case-study research approach, and proposed 4 novel wearable and portable sensing systems to assess the joint kinematics. Effective data acquisition and processing pipelines, sensing algorithms, software platforms and hardware prototypes are implemented. This thesis has leveraged two implementation approaches with respect to the nature of sensor data and requirements of the applications. The first approach consists of pattern recognition models that map motion related features to an activity set, and the second includes pose estimation models that recover the joint kinematic and pose trajectories. In some cases, proposed algorithms are further reinforced by sensor fusion algorithms and anatomical kinematics constraints for higher accuracy.

To sum, presented research projects in this dissertation highlight tangible interactive sensing technologies in the area of body motion sensing, and their continuous development toward actual and real-world products could contribute to a larger population in the near future. There are tremendous opportunities to expand such novel solutions to other application domains. Health and rehabilitation, sport medicine, entertainment, and input technologies have vast potentials for such extensions and innovations.