AI-Assisted Motion Analysis

GET STARTED
1
Request Info
2
Visit
3
Apply
ai-ama03

“Anatomy of a Fast-Ball”, Bryan Christie, from “Look Inside.” Wired © Gestalten 2016

ARTIFICIAL INTELLIGENCE & HUMAN MOTION

The Sports Medicine and Motion Analysis Lab, housed in the School of Education and Human Development’s Department of Kinesiology and Sports Science (KIN), is conducting cutting-edge research that focuses on alternative markerless motion capture technologies in an effort to replace traditional optical (marker-based) or sensor (IMU-based) systems. Our research is driven to improve the ability to accurately measure the human movement in natural environment, including clinic and home settings, without the need for external markers or devices.

Our vision is to provide an efficient method for clinicians, coaches, and movement professionals to obtain full biomechanical analyses in clinical or sporting environments. Through this, we hope to provide objective decision-making to boost athletic performance, reduce musculoskeletal injuries, improve post-surgical interventions, and aid management of chronic movement disorders.

MARKERLESS MOTION CAPTURE TECHNOLOGY

What is it? 

It is a method used to estimate the 3D positions of the human body over a period of time by tracking objects within image sequences moving in conjunction.

How does it work?

Using a high-speed video camera, this technology captures, analyzes, and provides a 3D reconstruction of a subject’s motion without the need to place markers or use other devices on the subject, resulting in the whole-body motion to be observed and behavior can be quantified.

CURRENT PROJECTS

Accuracy of KinaTrax markerless motion capture system during gait in Parkinson’s disease patients. PI: Moataz Eltoukhy, PhD.

Purpose: Determine if a markerless system can alleviate practical issues encountered when using marker-based systems to track human motion. These include time requirements for system and subject preparation, subject-researcher contact during marker placement, confinement to laboratory settings, and errors involved in the placement of markers on soft tissues. Additionally, we aim to test the validity of this new markerless motion capture system in quantifying the gait patterns of patients diagnosed with Parkinson’s disease.

Comparison of predicted ground reaction forces and moments during gait using full-body kinematics-driven musculoskeletal modelling and deep learning algorithms.

Purpose: Force plates are a common requirement for inverse dynamics calculations during gait. Specifically, during double support phase when the forces and moments under each foot become indeterminate. This confines full biomechanical analyses to a laboratory space and further to the specific areas of the embedded force plates. Deep learning networks have shown promise in solving the indeterminacy problem providing an alternative to plate measured or musculoskeletal ground-contact driven modeling prediction of ground reaction forces and moments. So, our aim is to be able to accurately predict the forces produced during the gait cycle without the need for force plates.

For more information, contact Dr. Moataz Eltoukhy, associate professor, Department of Kinesiology and Sport Sciences, and associate professor of industrial engineering.  

Learn more about Markerless Motion Capture:

TO FUND THIS RESEARCH, CONTACT:  

Angie Gonzalez-Kurver
Director of Development and Alumni Relations
305-284-5038 – ajgonzalez@miami.edu