Machine Learning & Machine Vision for quality control and picking

Thesis proposal


Virtual Manufacturing is a supplier of lean-based production development services. We believe in and are willing to take on the challenge of combining technology, methods and hard work around production development in order to achieve results.

Problem description

During the summer of 2018 Virtual has installed a collaborative robot for assembly at a customer. Next step is to investigate whether quality check during bin-picking can be efficient. The challenge is to integrate machine learning software with a time constrained robot program in order to check the quality simultaneous to the picking procedure.


Proposed research questions

Research question 1: How well would machine learning software perform in the same conditions?
Research question 2: What is the required data-size to get a satisfactory performance?

 (Left) Kuka IIWA. (Right) Machine Learning work flow.

Number of students

2 students

Torbjörn Danielsson

Anton Johannesson