Thesis work in the area of human-robot interaction:

Dialog System for Machine Diagnostics

Supervisor: Maria Krantz (Epiroc Surface) and Franziska Klügl (Örebro University)

This thesis shall look into how operators interact with machines when an error occurs: The operator remotely controls multiple machines (e.g. drilling rigs). The machines operate autonomously until they discover that they need support. If one of the machines encounter a problem (e.g. machine gets stuck), the human operator needs to start investigating what went wrong, while currently the machine just displays error codes. The goal of this thesis is eventually to develop a more natural – that means natural language based – way of interacting to find out about what when wrong and address the errors identified.

This means in detail the student (pair of students) needs to do the following

  • A short study looking into how the operators currently work when an error occurs, what does the operator look for, etc….
  • Development of a concept how the machine instead of displaying a cryptic error code, can use natural language to explain what has happened and in which state the system is now.
  • Implementation of a proof of concept based LLMs based on a simple diagnostic model, if possible integrated with automation software.
  • If possible, a short evaluation with operators/experts at Epiroc Surface

This thesis forms a kind of test-balloon developing a prototype to illustrate whether such a system can be helpful for an operator and help to determine whether further development is interesting for Epiroc. An integration on real machines is not foreseen as that would need coupling to sensors as well as elaborated diagnostic models of the machine; Such elements would be too large and complicated, but give the final visionary idea.

Questions and further inquiries at Franziska Klügl (Franziska.klugl@oru.se) or Maria Krantz (maria.krantz@epiroc.com). The exam work can be done in a CiviIng thesis or HIng thesis. Work in a team of two students might be advisable.

 

Annonsuppgifter

Annonsör: Örebro universitet

Ansök senast:

Annonskategori: Examensarbete, praktik, uppsats

Intresseområde: Data och IT, Teknik och matematik

Kontaktperson: Franziska Klugl (Professor) franziska.klugl@oru.se