Worker Activity Recognition for Work Measurement and Ergonomics Using Wearable IMUs and Deep Learning

Time study and ergonomic assessment are traditionally done by an analyst with a stopwatch and a RULA sheet. This project recognises the work operations of packaging workers every second from four wearable motion sensors (wrists and upper arms) of the public OpenPack dataset, compares hand-crafted features with Random Forest and SVM against a 1-D CNN and a CNN-LSTM on workers the models never saw, smooths the sequence with the usual order of the operations, and turns it into work cycles, element times against standard times and an upper-arm posture-risk indicator. A web application with a timeline view, time-study sheets and ergonomics pages, and a complete MATLAB version are included.

Category: Industrial Engineering

Industrial Engineering Project IE06 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.