Thesis work: Computer Vision for improvement of manufacturing process

Thesis background


Volvo Group Digitalization & IT at Tuve modernizes IT system landscape, integrates data from legacy silos systems and automates business processes. Now it’s time to increase utilization of smart sensors, Big Data, cloud computing and Business Intelligence to monitor, model and simulate manufacturing processes. In our journey towards Digital Factory we want to adopt various advanced technologies like Artificial Intelligence, Machine Learning and Computer Vision.


Thesis scope of work


In this master thesis, you will develop models that could be used to support manufacturing processes at Volvo trucks’ factory in Gothenburg. The work will start with revision of production stations and choosing those where Computer Vision can improve quality, cost efficiency, safety or reduce environmental impact.
Some example use cases to evaluate might be:
detect parts with defects
validate if all parts have been correctly assembled or packed
identify parts
find incorrectly stored packages
check if most dangerous zones are entered by employees wearing safety clothes or are not entered at all
track incoming/outgoing trucks and position of shipping containers at the factory’s yard

The work includes preparation of image datasets and evaluation of different model learning approaches (with various pre-trained models, epochs, datasets) for stationary cameras/computers and mobiles (depending on use case, usability and performance)


Required skills:
Courses in Machine Learning, Computer Vision, Data Science, Software Engineering or similar.
At least basic knowledge of machine learning models
Programming in Python with some experience of relevant frameworks/libraries such as Tensorflow, Pytorch, Keras and OpenCV.



Contact: Roman Dolanski, Delivery Manager Tuve, roman.dolanski@volvo.com


Thesis Level: Master
Language: English
Number of students: 2