Tracking Center of Mass with Limited Inertial Measurement Units

• Main Author: Morrow, Connor Nathaniel
• Format: Others
• Published: PDXScholar 2019
• Subjects: Mechanical Engineering
• Online Access: https://pdxscholar.library.pdx.edu/open_access_etds/5273; https://pdxscholar.library.pdx.edu/cgi/viewcontent.cgi?article=6345&context=open_access_etds

Wearable motion tracking systems pose an opportunity to study and correct human balance and posture during movement. Currently, these observations are either being conducted in laboratories with the use of camera systems and markers placed on the body, or through the use of suits containing large numbers (15-20) of inertial measurement units. However, to aid with rehabilitation of individuals with impaired balance, there needs to be an option to collect these observations outside of clinics and without incurring much cost from the user. I have focused on three inertial measurement units, one placed on each shank and one placed on the upper torso, to record posture and estimate where an individual's center of mass is located. The use of three inertial measurement units significantly reduces the cost and encumbrance for studying balance in an individual's primary environment. This reduction comes at the cost of accuracy, which is explored further through this thesis. First, I looked at different methods used for determining IMU orientation, such as a biologically inspired orientation algorithm and a gradient descent algorithm. I then incorporated those methods to create a model of the body that calculates an estimate on the center of mass, based on the user's weight and height. Finally, this estimate is compared to a force plate derived method of finding center of mass to determine viability of sensor reduction for practical applications. The created algorithm is able to generate a center of mass location that is able to match the actual location within 3 cm.