Newsitems

- My early interest in technology was definitely stimulated by Lego toys, which allowed me to easily create and implement machines, games, cars or space craft, says Eric Demeester. My current interest in technology and robotics is of the same kind as back then, in the sense that it is interesting to search for solutions to challenging problems, to implement a design, and to actually see the devised machine move according to a desired plan (typically after days of trying to understand why it does not move the way it should).

First winner
Eric Demeester is 31 years old and a post-doctoral researcher at the Catholic University of Leuven in Belgium. When studying mechanical engineering he specialized in the field of mechatronics. A 2,5 year project on robotic wheelchairs followed, leading to PhD-studies and since this spring, Eric is a postdoctoral researcher at the Department of Mechanical Engineering. As the first winner of the Robotdalen Scientific Award he recieved €20 000 at a prize ceremony in Sweden in September.

How will winning € 20 000 help support your future research career?
- Winning the Robotdalen Scientific Award reads well on my CV and will have a positive impact on raising funds for research projects. Prior to winning this award, I was more inclined to go to industry and to leave my research results at the university. Winning the award definitely encouraged me to continue my Ph.D.’s research and pushed me towards bringing the research results in practice.

- I am not sure how this should be performed in the best way, by starting a spin-off company or by co-operating with an existing wheelchair manufacturer for example. Incorporating research results into a novel product involves many aspects that I am not familiar with, such as legislation, management, funding and above all selling the product.

New application areas lay ahead
- I’ve been lucky to learn about many recent and promising techniques, such as map building or trajectory planning for robots, reflects Eric on his education. These techniques will clear the road for many new application areas such as humanoids, flying robots, space robots or autonomous cars, to mention a few. Not only the emergence and maturing of new techniques in robotics captured my interest, but also the fact that robotics requires knowledge about many different technology fields, from mechanical engineering over electrical engineering to computer science and artificial intelligence.

- My interest also stems from the potential to assist people. Where robots were previously mainly adopted in factories, they are now at a point of invading our every-day life. This may have a very large impact on our lives. For example, given the ageing of society, robots may become an indispensable tool to enhance the mobility of elderly and disabled people. In this regard, one of the most interesting topics for me personally is human-robot interaction: how should robots behave in order to obtain natural and intuitive human-robot interaction?

----------------------------------------------------------------------------------------------------
The winning thesis 2007
Eric Demeester won the Robotdalen Scientific Award 2007 with his thesis "User-adapted plan recognition and shared control for wheelchair driver assistance under uncertainty". Here he explains what his research is about:
The main goal of my research is to assist elderly and disabled people with powered wheelchair driving. Initially we thought this could be performed by “simply” adding sensors and a computer to the wheelchair, and by applying state-of-the-art mobile robot techniques such as localisation and path planning. However, we soon found that issues regarding human-machine interaction were at least as important.

Therefore, our main research focus shifted towards gaining insight into how robots should behave in order to obtain natural and intuitive human-robot interaction, with robotic wheelchairs as test case. This inherently requires a multi-disciplinary approach to robotics. For example, in order to gain insight in human-robot interaction, a thorough insight in the behaviour of humans is required. Therefore, we co-operated with wheelchair drivers, physicians, physiotherapists, and human factor specialists.

More specifically, the thesis focuses on three main research topics: plan recognition, user modelling, and shared control.

Plan recognition corresponds to the problem of determining which goal a user has in mind and how he or she wishes to achieve this goal. For example, we want to estimate which manoeuvre a wheelchair driver is trying to execute without him or her telling us which manoeuvre should be executed. We represent manoeuvres as geometric paths from the current wheelchair location to a desired goal location, and we estimate the probability of these mental geometric paths based on the driver's joystick signals and based on observation of the environment.

User modelling concerns the building and learning of models of human behaviour, for example models that predict how wheelchair drivers control their wheelchair. We are trying to understand how to adapt user models automatically to different drivers, and to continuously adapt these user models on-line.

Shared control corresponds to the problem of determining how much assistance the robot should give at each instant, taking into account uncertainty regarding user plans and taking the user's abilities into account
The next big step in our research will be to perform tests with wheelchair drivers over prolonged periods of time, for example in house-like environments that are dedicated to evaluate new assistive technology. A next step will then be to incorporate our techniques into commercially available wheelchair platforms. For this two options exist. Either we produce ourselves an embedded system that can be easily mounted on any powered wheelchair in plug-and-play fashion, either we co-operate with existing wheelchair manufacturers and integrate our techniques directly into new wheelchair designs.

“Winning the award definitely encouraged me to continue”