Tony Belpaeme

Home » Articles posted by tbelpaeme

Author Archives: tbelpaeme

PhD position in social robotics and machine learning

The AIRO group (Artificial Intelligence and Robotics) of the IDLab at Ghent University is looking for a PhD student to help with our research on social robotics and machine learning.

Social robots are robots which interact with people in a natural manner by using speech, gestures, and facial expressions for example. As social robots use natural communication, they are easy to interact with and this knows many applications in entertainment, services, education, collaborative robotics and therapy.

Now however, these robots are programmed by hand, as automatically generating verbal and nonverbal behaviour is still largely impossible. Our research aims to change that. The project will study how a robot can use generative deep learning to observe video recordings of people interacting with each other, and learn which behaviours are appropriate under which circumstances. The goal is to have the robot learn to produce verbal and non-verbal behaviour at a level where people think the robot is a worthy conversation partner.

You will be supervised by Prof Tony Belpaeme (www.tonybelpaeme.me) together with Prof Francis wyffels and Prof Joni Dambre. You will be part of a new, vibrant and interdisciplinary research team, focusing on unconventional robotics, cognitive systems and machine learning.

Ghent University (www.ugent.be) is a top 100 university in Belgium situated in the historical and trendy city of Ghent. With 41,000 students and 9,000 staff, the university is one of the largest universities in the wide region. Ghent University’s mission is to combine high-quality education with internationally leading research and a pluralistic social responsibility. You will join the AIRO group which is part of IDLab, a research lab consisting of more than 300 researchers developing today the tech of tomorrow.

Profile of the candidate

You must have an MSc degree in computer science, electrical engineering, technical cognitive science or areas relevant to the research topic. Good programming skills are required (C++, Python, or other), as is training in machine learning. English will be the primary language used (spoken as well as written). The PhD position is highly interdisciplinary and requires an understanding and/or interest in psychology and social sciences.

Further details:

  • The position starts on 1 October 2018 and runs for 4 years.
  • The net amount of the scholarship will be approximately €1900 per month, increasing to €2100 per month in the fourth year. You will also receive a holiday allowance and an end-of-year bonus, and will enjoy full social security cover. Additional financial support is available for attending conference and workshop.
  • You will be affiliated with the Artificial Intelligence and Robotics group of the IDLab. While the research will be based in Ghent, occasional travel to international conferences will be required.
  • You will be enrolled in the doctoral training programme offered by the Doctoral School of Engineering.
  • Ghent University encourages equal opportunities and will consider applications only based on your potential as an early career researcher.

How to apply

For informal queries, do not hesitate to contact Tony Belpaeme (tony.belpaeme@ugent.be). Your application should include

  1. a letter motivating your application.
  2. a CV, copies of relevant exams, grades, master thesis work or publications.
  3. the names and contact details of at least 2 referees. Recommendation letter can be included with your application, but do not need to be at time of application.

Applicants should send their application to tony.belpaeme@ugent.be. The application deadline is 31 May 2018. Selected candidates will be invited for interview, which can be organised over Skype if necessary.

 

HRI2018

We descended on this year’s Human-Robot Conference (HRI) with 10 people from the Plymouth team: a research budget breaking record. I gave a keynote at the Personal Robots for Exercising and Coaching workshop and the Cognitive and Social Neuroscience Methods for HRI workshop, Daniel Hernandez co-organised the Social Robots in Therapy: Focusing on Autonomy and Ethical Challenges workshop, we presented our alt.HRI paper and had a clutch of Late Breaking Results to show.

This was the largest HRI edition so far, with about 650 delegates. What struck me was the large presence of industry at the conference, probably fuelled by the recent feeding frenzy around AI and robotics, but also showing that HRI is maturing and that applications are now becoming market ready.

We couldn’t resist shooting a band cover at the Chicago waterfront, from left to right Fotios Papadopoulos, Tony Belpaeme, Charlotte Edmunds, Christopher Wallbridge, Emmanuel Senft, Maddy Bartlett, Bahar Irfan, Mina Marmpena (Daniel Hernandez and Serge Thill not showing).

Band picture HRI2018

Arrived at Ghent University

Gent-Plymouth contrastI have been neglecting my website lately, so here is a belated post announcing that I have accepted a position at Ghent University. I maintain my post at the University of Plymouth, if only to see running projects to a good end and to supervise the many PhD and postdoctoral researchers still there. Still, a new chapter in my academic career, with new opportunities and new challenges. The reasons for moving were many, Brexit being one of them, the others I will have to regale you when we next meet. I am joining an ambitious team of machine learning (Deep Learning of course) and unconventional robotics enthusiasts here in Ghent, and I hope to build up the research in social robotics and Human-Robot Interaction here. If you wish to visit or collaborate, by all means get in touch!

Séverin Lemaignan studying Theory of Mind

I am very excited about Dr Séverin Lemaignan joining us in a few months. Séverin will be working on his DOROTHY project, short for “Donating Robots a Theory of Mind”, for which he received funding from the Marie Skłodowska-Curie programme. As humans we take a Theory of Mind very much for granted, the skill to understand what others know, what they are thinking and what they are about to do is almost magic. It is only when someone’s ToM is impaired or completely lacking that you realise how crucial a tool it is. No longer having access to what people around you know and think about their environment and about others leaves you very much on the back foot when trying to respond appropriately in a social environment. Unfortunately, as with so many skills that seem natural to us humans, a Theory of Mind is very difficult to build for a robot. Séverin will join us from EPFL in the summer of 2015 to try and do just that: build an artificial Theory of Mind.

PhD opportunity – we are looking for a team mate!

Update: this position has been filled. We’re welcoming Emmanuel Senft to the team from October 2014 onwards.


We are looking for a team mate!

This is the age of robots, where robots are not just found in factories picking and placing things, but will enter our daily lives: places where no robot has ever gone before. However, robots currently have limited autonomous social capabilities. For this to change, advances in artificial intelligence need to be matched up with advances in robotics.

This PhD looks into the science and technology of action selection for social humanoid robots. Given sensor input, prior knowledge and dose of social cognition, would it be possible for the robot to autonomously respond in a social manner? Currently almost all social robots need a human hand to operate well: the “Wizard of Oz” idea, where a human operator takes over when the robot is stuck. What is needed to move away from WoZ and build fully autonomous social interaction?

You will be working in the context of the DREAM project (www.dream2020.eu), a multinational European project which studies the science and tech behind the robot-enhanced therapy (RET), specifically targeted at young children with Autism Spectrum Disorders. DREAM uses a range of robots, but through its collaboration with Aldebaran the use of the Nao humanoid is central to our research.

Applicants should have (at least) a first or upper second class honours degree in an appropriate subject and preferably a relevant MSc or MRes qualification. We encourage applicants with a background in computer science, artificial intelligence, robotics and/or cognition science.

We expect candidates to be confident programmers (C++ experience recommended): you will spend a considerable amount of time programming and hacking robots and their artificial intelligence, which is not for the fainthearted. A keen interest in science and discovery is a prerequisite, as is not being afraid of international travel to meetings and events.

If you wish to have an informal chat, please contact Tony Belpaeme (tony.belpaeme@plymouth.ac.uk) or contact any of the mad people in the lab. However, applications must be made via the official application site.

More details on funding and the application process can be found at FindAPhD. Applications close at noon on Monday, 28 July 28 2014. Quick!

HRI2014

Together with Andrea Thomaz (Georgia Tech) we chair the Programme Committee of the 2014 edition of IEEE/ACM Human-Robot Interaction (HRI2014) in Bielefeld, Germany. This year’s edition focuses on (E)merging perspectives. HRI is a cross-disciplinary endeavour, marrying state-of-the-art AI and robotics with real-world applications in noisy and unstructured social environments. It is one of the final frontiers in robotics: after having planted robots on the moon and on Mars, and after having robot explore the deep seas, the field of HRI pushes to have robots closely interact with people. However, social HRI is as yet an unsolved scientific and technical challenge.