Visual Tracking for the Analysis and Measurement of Human Motion During Repetitive Tasks

Faculty Researcher: Nicola J. Ferrier, PhD, Assistant Professor of Mechanical Engineering at the University of Wisconsin at Madison

The ultimate goal of this project is to develop an automated system to identify and measure human position and motion during the execution of repetitive tasks. In assessing industrial hazards, human motion is often recorded on videotape. In evaluating the degree of repetition for assessing work-related risk, trained human observers view the video and score the repetitiveness of the task. This project aims to replace the human observers with a computer vision system. In other fields, such as traffic surveillance, human analysis of videotapes has been replaced with automated video analysis. Automated analysis and classification of human motion during repetitive tasks has not been successfully demonstrated, yet preliminary work in analysis of human activity suggests that automatic measurement of human motion is feasible. This project will concentrate on the measurement of hand and upper limb motion. Experiments will be performed to demonstrate feasibility and validate the system by comparing measured results to previously-studied motion data obtained using conventional methods. Successful completion of the project will produce the means to automatically analyze videos of humans performing tasks and produce quantitative data to determine whether injury risk exists, and if so, which tasks or sub-tasks may require further analysis or intervention. Such a system would provide a robust, repeatable measurement of repetition that is not subject to human subjectivity, fatigue or other difficulties.


Publications resulting from this project:
Lu C, Ferrier NJ. A digital video system for the automated measurement of repetitive joint motion. IEEE Trans Inf Technol Biomed. 2004;8(3):399-404.

Lu C, Ferrier NJ. Automated analysis of repetitive joint motion. IEEE Trans Inf Technol Biomed. 2003;7(4):263-273.

Lu C, Ferrier NJ. Repetitive motion analysis: segmentation and event classification. IEEE Trans Pattern Anal Mach Intell. 2004;26(2):258-263. doi:10.1109/TPAMI.2004.1262196.


Research trainee’s current position:
Nicola Ferrier is currently the Principle Automation Engineer for Argonne National Laboratory’s Mathematics and Computer Science Division.