It's always night or we wouldn't need light

- Thelonius Monk

Of course this time of year up here it's always day, but it's not meant to be taken literally.

With Luke the Duke in Amsterdam

On my flight home from the Netherlands, for the first time in my life I felt "if this plane crashes and I die, human knowledge will be set back at least a few years". In retrospect, months might be more accurate than years, but still. Had I been a cartoon character, the light in question, in this case, would have come from a bulb above my head.

The meeting was a mixture of roboticists, bio-mechanicists, prostheticists, orthoticicists, and a few of us control theorists, all with a common interest in walking and balance. I got to try walking with a prosthetic leg (not so easy), I met Theo Jansen (see post below) and got to play with his kinetic sculptures and see how they are made.

Me playing with a kinetic sculpture

I also discovered that a big secret embarrassment for the walking biped robot community is that none of their robots can walk on rough ground. This inevitably raises the question: if they can only walk on flat ground, why not just use wheels? The best I've seen can handle a step down of about 8 cm, but that's more arse than class, and most will collapse if the floor varies by millimeters. The control problems are hard, and all current solutions are ad-hoc.

(turn your head 90 degrees to the right for optimal viewing)

The video above is of one of the better current robots. Note that the well-known robots by Honda, Sony, Toyota, etc have such big feet and take such small steps that they don't make any attempt to solve the problem of balance. We consider that cheating.

The thing is, I know how to solve the problem, how to systematically design stabilizing controllers for this type of system. We here in Umeå have just submitted a whole bunch of papers about it. At the meeting, after a few hasty napkin drawings and some stern words I managed to convince a few people of this, and got invited to spend some weeks at MIT in August doing some tutorial lectures for the robotics group, and testing out our ideas on some of their robots. I will possibly also visit Cornell.

I think it will be great fun, and as a bonus I'll get a chance to catch up with the following illustrious characters in Boston or New York: KC, KT+Mat, Marjukka, and David K.

Between now and then, I'll be back in Korea for a week, and then I'll have two weeks vacation in Slovenia and Croatia with Paul.