Description

Problem
In labor-intensive manufacturing systems (LIMS), there are “dark spots” in terms of data availability:

Many manual processes are not digitally captured.
This leads to limited transparency about worker activities, process execution, and bottlenecks.
As a result, it is difficult to monitor, optimize, or automate decision-making for these manual tasks.

Goal
Increase data availability and process transparency in LIMS by applying machine-learning-based Human Activity Recognition (HAR):

Automatically recognize and classify worker activities using accelerometer data.
Generate event logs from previously unobserved manual work.
Use these enriched datasets to support process analysis and performance monitoring.

Required Skills and Knowledge