KI Methoden zur Eruierung optimaler Prozessparameter der Kühllinie (Tempcore-Prozess) 

Offende Projekte, Bachelor- und Masterarbeiten

• Machine Learning for Predicting Yield Strengths.

Der Lehrstuhl für Cyber-Physical-Systems widmet sich anwendungsorientierter Grundlagenforschung in den Bereichen der künstlichen Intelligenz, der Digitalisierung von Industrieprozessen und der Robotik. Ein Focus liegt dabei auf der Modelierung von intelligenten menschlichen Lernprozessen mit dem Ziel effiziente Lernmethoden und Vorhersagemodelle für cyber-physikalische Systeme zu entwickeln.

Hier finden sie den Foliensatz zum Vortrag.

Supervisors: Univ.-Prof. Dr Elmar Rückert,
Vedant Dave, M.Sc. 
Dr. Christoph Sorger und Dr. Luca Moderer
Univ.-Prof. Martin Stockinger (Chair of Metal Forming)
Start date: ASAP from June 2022

Theoretical difficulty: low
Practical difficulty: low

Abstract

In this thesis, the student has the unique opportunity to investigate supervised machine learning methods for predicting yield strengths using probabilistic regression models and deep learning approaches. The thesis is implemented with support of the MSC Software GmbH and the Stahl- und Walzwerk Marienhütte GmbH in Graz.

In the image above and below you see the production line at the Stahl- und Walzwerk Marienhütte GmbH in Graz.

To ensure the high quality standards, frequent ‘yield strength’ measurements are performed. These measurements have resulted in a large dataset which can now be analyzed and used to learn a prediction model.  First tests were promising and the thesis will be very likely a big success. 

The goal of this thesis is to analyze the data and to learn prediction models taking uncertainty estimates into account.

The models will be implemented and tested in Python.

Tentative Work Plan

To achieve our aim, the following concrete tasks will be focused on:

• Literature research on the underlying physical & chemical processes.
• Data analysis, filtering, preprocessing, visualization of the existing data. 
• Implementation of  deep neural networks (Variational Autoencoder), neural processes and GPs in Python. Baseline implementations are existing.
• Visualization and analysis of the prediction performance. An outlier detection and warning system should be implemented.
• (Optional) Implementation of neural time-series models like LSTMs.
• Analysis and evaluation of the provided data.

1 position for a fully employed University Assistant at the Chair of Cyber-Physical-Systems at the earliest possible date or beginning on 1^st of July 2022 in a full-time (100%) 4-year term of employment. Salary Group B1 to Uni-KV, monthly minimum charge excl. SZ.: € 3.058,60 for 40 hours per week (14 times a year), actual classification takes place according to accountable activity-specific previous experience.

We are looking for a motivated student interested in a PhD thesis on “Machine Learning of Robot Motor Skills”. The goal of the doctoral thesis is to develop probabilistic or deep neural networks for interactive learning of autonomous systems and industrial processes. This will involve processing complex sensor data such as from RGB-D cameras, tactile data and sensory data of the industrial machines for motion control of robotic systems. The developed models will be tested using realistic industrial applications for process modeling and with robotic systems within our AI Robot Lab.

What we offer

The opportunity to work on research ideas of exciting modern topics in artificial intelligence and robotics, to develop your own ideas, to be part of a young and newly formed team, to go on international research trips, and to receive targeted career guidance for a successful scientific career.

Job requirements

Completed master’s degree in computer science, physics, telematics, statistics, mathematics, electrical engineering, mechanics, robotics, mechanical engineering or an equivalent education in the sense of the desired qualification; Programming experience in one of the languages C++, Python or similar; Experience in robotics, with ROS, and reinforcement learning or in machine learning. Willingness and ability for scientific work in research including publications with the possibility to write a dissertation.

Desired additional qualifications

Experience with movement primitives, Keras, Tensorflow and graphical models. Basic knowledge of Linux is advantageous. Good English skills and willingness to travel for research and to give technical presentations.

Application

Application deadline: June 2nd, 2022

Online Application via: Montanuniversität Leoben Webpage (2205WPC)

The Montanuniversitaet Leoben intends to increase the number of women on its faculty and therefore specifically invites applications by women. Among equally qualified applicants women will receive preferential consideration.

You are interested in working with real robots and want to understand how such machines ‘learn’?

This project will enable you to dig into the fascinating world of robot learning.

You will work alone or in a team on modern, state-of-the art hardware at the Chair of CPS.

We offer complex robotic systems, powerful PCs and GPU clusters to work with.

The course provides a structured and well motivated overview over modern techniques and tools which enable the students to define learning problems in Cyber-Physical-Systems. 

Note that offer two courses, 190.014 and 190.019. You may select two project topics or continue your project in the second course. 

Selected Topics (Many more are available upon request)

Tasks

• Select a topic from the presented topics at the CPS chair 
• Build a team of 3-4 students.
• Send an E-Mail with all details (topic, members) to the supervisor (E-Mails are on the CPS webpage).
• Keep track of your tasks and working hours (spreadsheet with tasks, hours per day, and members).
• Send a 1-page concept of your project (tasks, outcomes, milestones, est. work hours, meetings with the supervisor) to us.
• Create a git repository on our studgit.cps.unileoben.ac.at
• Implement your project.
• Write a readme AND a wiki in the git repository. Dokument how you used AI tools like chatGPT. Translation and spelling tools can be neglected. 
• Create slides to present your work at 24.06.2024 12:00-15:00 HS Thermoprozesstechnik

Grading

• 80 points for the project results (code, effort, team management, git readme, and git wiki).
• 20 points for the final presentation (20 min per group + 5 min. questions).
• up to 20 bonus points for extraordinary efforts. 

Links and Resources

• studgit.cps.unileoben.ac.at
• MUOnline 190.014
• MUOnline 190.019

Location & Time

• Location: Independent project team work, or for real robot projects in the Laboratories at the Chair of Cyber-Physical-Systems.
• Dates:
  • Regular, typically weekly meetings, with your supervisor(s). 
  • Final presentation at the end of the semester.

Learning objectives / qualifications

• Students get a practical experience in working, modeling and simulating Cyber-Physical-Systems.
• Students understand and can apply advanced model learning and reinforcement  techniques to real world problems.
• Students learn how to write scientific reports.

Literature

• The Probabilistic Machine Learning book by Univ.-Prof. Dr. Elmar Rueckert. 
• Bishop 2006. Pattern Recognition and Machine Learning, Springer. 
• Barber 2007. Bayesian Reasoning and Machine Learning, Cambridge University Press. 
• Murray, Li and Sastry 1994. A mathematical introduction to robotic manipulation, CRC Press. 
• B. Siciliano, L. Sciavicco 2009. Robotics: Modelling,Planning and Control, Springer.
• Kevin M. Lynch and Frank C. Park 2017. MODERN ROBOTICS, MECHANICS, PLANNING, AND CONTROL, Cambridge University Press.

Univ.-Prof. Dr. Elmar Rueckert is organizing this research seminar. Topics include research in AI, machine and deep learning, robotics, cyber-physical-systems and process informatics. 

Language:
English only

Are you an undergraduate,  graduate, or doctoral student and want to learn more about AI?

This course will give you the opportunity to listen to research presentations of latest achievements. The target audience are non-experts. Thus no prior knowledge in AI is required.

To get the ECTS credits, you will select a research paper, read it and present it within the research seminar (10-15 min presentation). Instead of selecting a paper of our list, you can also suggest a paper. This suggestion has to be discussed with Univ.-Prof. Dr. Elmar Rueckert first.

After the presentation, the paper is discussed for 10-15 min.

Further, external presenters that  are leading researchers in AI will be invited. External speakers will present their research in 30-45 min, followed by a 15 min discussion. 

Links and Resources

• MUOnline

Location & Time

List of Talks and Dates

• 22.11.2024 11:15 (Seminar Room HdD – our new building)
  
  • M.Sc. thesis final presentation: Daniel Wagermaier, Predictive Models for Copper Content and Energy Consumption in Electric Arc Furnace Operations. 
• 06.12.2023 10:00 (Seminar Room HdD & Lab – our new building)
  • Tutorial: by NOYES GmbH, on our new autonomous warehouse system. 

Available Research Papers to Select

• Own research topics.
• Important tutorials.
• Predictive Whittle Networks for Time Series by Yu et al. (UAI 2022)
• Natural Gradient Shared Control by Oh, Wu, Toussaint and Mainprice
• Learning to solve sequential physical
  reasoning problems from a scene image by Driess, Ha and Toussaint (IJRR 2021)
• Deep Visual Constraints: Neural Implicit Models for Manipulation
  Planning from Visual Input by Ha, Driess and  Toussaint
• AW-Opt: Learning Robotic Skills with Imitation and
  Reinforcement at Scale by Lu et al. (CoRL 2021)
• Decision Transformer: Reinforcement Learning via Sequence Modeling (NeurIPS 2021) by L. Chen et al
• CURL: Contrastive Unsupervised Representations for Reinforcement Learning (ICML 2020) by A. Srinivas et al
• Offline-to-Online Reinforcement Learning via Balanced Replay and Pessimistic Q-Ensemble (CoRL 2021) by  S. Lee et al
• Deep Hierarchical Planning from Pixels (preprint) by D. Hafner et al
• Offline Meta-Reinforcement Learning with Online Self-Supervision by Pong et al. (ICML 2022)
• Information is Power: Intrinsic Control via Information Capture by Rhinehart et al. (NeurIPS 2021)
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks by Finn et al.
• Event-based Asynchronous Sparse Convolutional Networks by Messikommer et al.
• Gaussian-binary Restricted Boltzmann Machines on Modeling Natural Image Statistics by Wang et al.
• A Spiking Neural Network Model of Depth from Defocus for Event-based Neuromorphic Vision by Haessig et al.
• Dream to Control: Learning Behaviors by Latent Imagination by Hafner et al.

The exercise will enable the application of modern machine learning techniques and tools in robotics / cyber-physical systems. The following topics will be covered in the course:
   – Kinematics, dynamics & simulation of CPS.
   – Data representations & model learning.
   – Control techniques, priorities & torque control.
   – Planning & cognitive reasoning.
   – Reinforcement learning and black-box optimization.

The course provides a structured and well motivated overview over modern techniques and tools which enable the students to define learning problems in Cyber-Physical-Systems. 

Links and Resources

• Find all relevant information in the post on the lecture 190.001 ‘Cyber-Physical-Systems’.

• Optional Online Streaming:

  • Live Streaming of the lectures via our Nextcloud TALK App.

Exercises

• Python Basics Tutorial: Jupyter Notebook.

Assignments / Team work

• Team Work 1 [submission till 16.11.22 12:00]: Jupyter Notebook on forward and inverse kinematics.
  • Q&A Session on 9.11.22 12:15 in HS Themoprozesstechnik and online (live.ai-lab.science).
• Team Work 2 [submission till 21.12.22 12:00]: Assignment Description on real robot inverse kinematics. 
  • Q&A Session on 23.11.22 12:15 in HS Themoprozesstechnik and online (live.ai-lab.science).
  • Q&A Session on 30.11.22 12:15 in HS Themoprozesstechnik and online (live.ai-lab.science).
  • Q&A Session on 07.12.22 12:15 in HS Themoprozesstechnik and online (live.ai-lab.science).
  • Q&A Session on 14.12.22 12:15 in HS Themoprozesstechnik and online (live.ai-lab.science).
• Team Work 3 [submission till 01.02.23 12:00]: Assignment on Dynamic Movement Primitives for real robot motion control.

Location & Time

• Location: HS Thermoprozesstechnik
• Dates: Wednesdays 12:15 – 14:00

CPS Latex Templates

Find below Latex templates for student reports, B.Sc. theses, M.Sc. theses and doctoral theses.

You may use these templates also for other lectures, courses, seminars or doctoral theses at other chairs at the Montanuniversität Leoben. However, do not remove the acknowledgement or copyright statement.

Student Report or Assignment

We provide a professional scientific student report template using double columns.

Get the latest CPS report template from our cloud.

Student Presentation

We provide a scientific student presentation template for project and thesis reports. Get the latest CPS presentation (latex beamer) template from our cloud.

B.Sc. Thesis

We provide a thesis template using minitocs in a standard report format.

Get the latest B.Sc. template from our cloud.

The latex template also provides basic instructions on the content and the structure of a thesis.

However, every thesis is unique and may be adapted acordingly.

M.Sc. Thesis

We provide a thesis template using minitocs in a standard report format.

Get the latest M.Sc. thesis template from our cloud.

Ph.D. Thesis

We provide a Ph.D. thesis template using minitocs in a standard report format.

Get the latest Ph.D. thesis template from our cloud.

In Introduction to Latex

If you have never used Latex, you find a brief intorduction in these slides on Latex.

Ausstattung:

• 2 universal robotics UR3e Roboter,
• 2 FANUC CRX10iA Roboter
• eine Drehbank von ELMAG
• eine Industriebohrfräse von ELMAG und
• ein Rollenförderband.

Die Anforderungen an Industriebetriebe im Zeitalter der Digitalisierung sind enorm gestiegen und der Bedarf an individualisierten Losungen ist groß. Kleine Stückzahlen, komplexe Bauteilegruppen und der notwendige hohe grad der Automatisierung ist mit fest vorprogrammierten Roboterprogrammen nicht mehr umsetzbar.

Projektziel

Ziel des Projektes ist es sich als starken Partner für Forschungs- und Industriebetriebe zu positionieren. Dazu soll ein begehbares, autonomes Industrieroboterlabor aufgebaut werden, in dem Robotern praxisrelevante Arbeitsablaufe durch moderne Lernmethoden der kunstliche Intelligenz beigebracht werden. Werksmitarbeiter können innerhalb weniger Sekunden, durch Vorzeigen oder durch fuhrendes Anleiten, Maschinen komplexe Bewegungsabläufe beibringen. Für die Koordination multipler autonomer Robotereinheiten und die Prozessüberwachung werden moderne Datenmodellierungsmethoden entwickelt und über Tablets bedient.

Anwendungen

Die Anwendungsszenarien umfassen Manipulations- und Sortieraufgaben  an einem Rollenforderband, die automatische visuelle Objekterkennung und Vorhersage unter realen Industriebedingungen, der Warentransport durch mobile Roboter mit Greifarmen und die Bedienung komplexer Industriemaschinen, exemplarisch vorgefuhrt an einer Bohrfräsmaschine und an einer Kleindrehbank.

Kooperationen und öffentliche Events

Das autonome Industrieroboterlabor soll nachhaltig zu Kooperationen mit nationalen und regionalen Forschungs- und Industriepartnern fuhren und die Sichtbarkeit des Lehrstuhls fur Cyber-Physical-Systems (CPS) und der Montanuniversitat Leoben im Bereich der angewandten kunstlichen Intelligenz für CPS durch jährliche öffentliche Events steigern.

Der Lehrstuhl für Cyber-Physical-Systems widmet sich anwendungsorientierter Grundlagenforschung in den Bereichen der künstlichen Intelligenz, der Digitalisierung von Industrieprozessen und der Robotik. Ein Focus liegt dabei auf der Modelierung von intelligenten menschlichen Lernprozessen mit dem Ziel effiziente Lernmethoden und Vorhersagemodelle für cyber-physikalische Systeme zu entwickeln.

Gerade diese Schnittstelle zwischen fundamentaler Grundlagenforschung in tiefen neuronalen Netzen, probabilistischer Informationsverarbeitung und komplexen industriellen Anwendungen zeichnen den Lehrstuhl für Cyber-Physical-Systems aus.

Neben der Entwicklung von Algorithmen und Methoden zur Modelierung und Verarbeitung großer Datenmengen, baut der Lehrstuhl auch komplexe Roboter- und Sensorsysteme. Eines dieser Systeme wird in naher Zukunft autonom an der Universität navigieren, Besucher empfangen und mit ihnen über ein gelerntes Dialogsystem kommunizieren. Darüber hinaus entsteht gerade ein begehbares KI Roboter Labor, dass die anwendungsorientierte Grundlagenforschung anhand von Aufgaben mit Industrieroboterarmen an einem Rollenförderband greifbar macht.