Q&A: VUB engineer on making robot mouths more human
Gabriël Van de Velde used hundreds of photographs of mouths to fine-tune the way robots move theirs
Mechatronics combines electronics with several kinds of engineering, including mechanical telecommunications and computer. For his master’s thesis at the Free University of Brussels (VUB), Van De Velde made significant improvements to robotic mouths and artificial speech articulation, making them simpler to implement and more human-like in appearance.
His research was runner-up for the Agoria Prize, awarded to the best technology related thesis in Flanders.
Are human-like features that important in robots?
Indeed they are. Even if many of us don’t realise it, artificial intelligence is present everywhere. We may not be afraid of Google or Facebook’s AI algorithms, but robots still frighten us – even though they are helping us in so many different ways, from robot-assisted surgery to elderly care. They play an increasingly important role in our everyday lives.
How did you improve the movement of their mouths?
I started with photographs of the human mouth. I transformed them and was able to compare the pictures with the robotic mouth that I was building. I used the differences in features to slightly adjust the robotic mouth. I used this principle over and over, until there were no noticeable differences anymore.
The crucial aspect of my improvement is that technicians who build robots will no longer have to write a manual or implement a computer-guided calibration model. These techniques are expensive and time-consuming. Imagine that you have to calibrate a robotic head – fitted with more than 30 motors – for every mouth position. That’s a lot of work! My method was tested with only five motors, which is good for 52 reference positions. It only took an hour.
You’re now working on a PhD in rotor technology – which has nothing to do with robots.
For mechatronic engineers, life isn’t only about robots. I see a variety of potential study areas. Rotary dynamics has always fascinated me. I’m studying a new method to improve the speed of rapidly turning machines – beyond 60,000 revolutions per minute. That’s a big challenge!