Research projects

Visually monitoring assembly and factory halls is a challenge. Lighting conditions are often poor or highly changeable, areas can be dirty and shiny machine parts are moving constantly. At the Styrian automation and machinery company Hage Sondermaschinenbau, workers currently carry out safety checks, using padlocks that have to be hung up or strategically positioned buttons. However, the checks are time-consuming and not completely secure. With VRVis, Hage wants to digitalise these processes and prevent accidents or injuries among employees.

Detecting human presence automatically 

In this test-before-invest project from the EDIH "AI5production" funding programme, the partners are investigating whether a camera system can be combined with machine learning to reliably and automatically exclude people from the risky area around machines during everyday operations. Traditional computer vision is not always suitable in this context. VRVis and Hage Sondermaschinenbau want to identify appropriate camera systems and subsequently simulate and record typical situations. The research team will use this to train and test different ML components.

Detailed tests in a variable environment

Avoiding a functional limitation of the ML model to the training data (overtraining) is important. The project partners want to analyse this through follow-up tests under varying site conditions at customers' facilities.