Thomas Hellström

Research 

I  lead the Intelligent Robotics group at the Department of Computing Science at Umeå University. The general research focuses on the use of AI for the development of robots for social and industrial applications. Our current main track of research is Human-Robot Interaction (HRI). Between 2010 and 2014 I coordinated the EU project INTRO (FP7/ITN), with six international partners working on intelligent robots operating in close interaction with humans. In our lab, work focused on cognitive architectures and robot learning. Between 2016 and 2020 I coordinated SOCRATES, an H2020-financed project with seven international partners and 15 PhD students. The overall research topic was Social Robotics with applications in eldercare. In our group we developed solutions for understandability issues for robots, natural language solutions, and interaction in human-robot teams.  We applied deep learning techniques for object detection and natural language processing.

A second focus in the Intelligent Robotics Group is Field Robotics. We have done a lot of research on unmanned forest vehicles, such as autonomous navigation, simulator-based path planning, and simulation of system configurations. In the EU projects CROPS and SWEEPER, we worked on developing robots for sweet pepper harvesting, and on sensor solutions for the detection of humans, neural network classification of objects, and tree diameter estimation.

I have also published work on robot ethics, bias in machine learning, and occasionally contribute to the public debate on Artificial Intelligence.

Between 2014 and 2016, I lead a project in collaboration with CMTS, MT-FoU, and Umeå Stroke Center at the University Hospital of Umeå, Sweden. In this project, we developed an intelligent rollator, with improved usability for people with cognitive limitations, e.g. after a stroke.

In my earlier research, leading up to my PhD degree in 2001, I worked with machine learning for financial modeling and prediction. A complete description of my research profile has to include my industrial work with the development of nautical control systems. For several years I worked as a self-employed with the development of the Seapacer system, with advanced functions for propeller pitch optimization, route planning, and dynamic speed and fuel consumption control. Besides the engineering and commercial endeavor, I published a few scientific papers on the project.

Funding

• 2023  3700 TSEK  ROCC - Vetenskapsrådet
• 2022  3700 TSEK  SCAI - Vetenskapsrådet (co-applicant)
• 2019  3470 KSEK  VINNOVA (co-applicant)
• 2016  3875 K€      EC  SOCRATES  
• 2014    200 KSEK  Promobilia
• 2014    811 K€      EC SWEEPER
• 2013  1900 KSEK  Kamprad foundation
• 2011    200 KSEK  KSLA  (co-applicant)
• 2010    782 K€      EC CROPS
• 2010    100 KSEK  Innovationsbron
• 2009  2083 K€      EC INTRO  
• 2006    325 KSEK  Process-IT
• 2005  4000 KSEK  Kempestiftelserna
• 2004    177 KSEK  Carl Tryggers stiftelse
• 2004    250 KSEK  Alvis Hägglunds
• 2004    286 KSEK  LKAB
• 2003    372 KSEK  VINNOVA
• 2002  3000 KSEK  Kempestiftelserna

PhD student supervision

I was the main supervisor for

• Michele Persiani - PhD degree September 2022
• Ahmad Ostovar - PhD December 2019
• Benjamin Fonooni - PhD January 2015
• Mostafa Pordel - PhLic, May 2013
• Ola Ringdahl - PhD May 2010 

I was co-supervisor for

• Jussi Manner - PhD December 2015 (SLU, Umeå)
• Efi Vitzrabin - PhD February 2015 (Ben Gurion University, Israel)
• Erik Billing - PhD January 2012

Some presentations

• The Reasonable Ineffectiveness of Data. SweDS2018, Nov 20, 2018.
• MACHINE LEARNING - Vad och Varför. Plenary talk at Medicinsktekniska dagarna, Nov 10, 2018.
• Quality in research
• Learning in robotics - a biased overview

Media

• Dangerous over-confidence in AI that so far is too unintelligent. DN Debatt Oct. 8, 2018
• Vad är rätt och fel för krigsrobotar?
• What is right and wrong for war robots?

Ohh right, I am proud holder of an Erdös-Bacon-Sabbath number≤ 13