[copying from Robohub]
Two of the most challenging problems tackled with quadrocopters so far are balancing an inverted pendulum and juggling balls. My colleagues at ETH Zurich’s Flying Machine Arena have now combined the two.
As part of his Master thesis Dario Brescianini, student at ETH Zurich’s Institute for Dynamic Systems and Control, has developed algorithms that allow quadrocopters to juggle an inverted pendulum. If you are not sure what that means (or how that is even possible), have a look at his video “Quadrocopter Pole Acrobatics”:
You can read the full article with much more details and other photos here:
@Krysztof It doesn't even have to be that high-res. You could couple a panoramic mirror with a decent processor to do object & position recognition and measure optical flow. Most of the technology is probably available today -- the hard part is writing the software for it.
@Kal This is interesting. I have seen recently ARGU-IS by DARPA video. Maybe a smaller version of it would be feasible in EU?
@Krysztof I'm not sure that we're too far off from putting this kind of image processing equipment on board.
You assume that they do something that has a chain of consequences and followers. To my knowledge in Lausanne they did something that worked better in that respect: pix4d and swinglet cam. From what happens in ETH Zurich I saw no implementations, but I would like to. Maybe somebody can share some info.
At the same time their research in applied maths was quite basic at Lausanne. This is why I would like to know what is the flagship outcome of ETH Zurich (Solar Flyer maybe?). I believe there should be somethign I have overlooked. Just teasing them to tell the story, all my research always falls back into quadcopter domain.
I think that 90% of the usefulness of this group is coming up with these out-of-the-box concepts and sharing them with us. It gives me ideas, it gives others ideas, and for the most part they document and share how they accomplished it. It's also good because it is a school and it gets students involved with cutting edge research! And yes, this is a baby step, a stepping stone, to outdoor tests.
I work at a medical research Institute, and 99% of the work they're doing here would be considered "useless" to 99% of the general population, but the research and experiments done here lead to better understanding (through isolated lab environment) and lay the framework for REALLY important breakthroughs.
Of course, but it is merely scratching the surface. Moreover if they scratch the surface with the same quadocopter layout, in the same hall, for years, only by changing control algortihm and scenario wihtout ever testing it with outside conditions, by scratching the surface at the same point they only create elegantly looking brushed steel but are not going any deeper.
Let's not make a mistake, I am pointing out the weakest point in the larger picture, the research group itself is well focused and generates perfect results like the one presented.
Krzysztof, I refer you to John Birkelands explanation. Laboratory testing WITHOUT those external factors is 100% necessary.
'You gotta test, experiment and develop in the laboratory first.'
No-no. If you ignore the wind, turbulence, changing temperature and humidity, you ignore 95% of aerodynamics on this planet. Then throw in ultra resolution positioning and you also ignore real life accuracy problems. Settle in the same testing hall for 5 years and you completely miss repeatibility and deployment issues.
You gotta test, experiment and develop in the laboratory first. The hardest thing they're doing is the algorithms and creating concepts (thinking WAY outside the box). Then later they can move outdoors... GPS is actually way more accurate than what we're allowed to use, its been degraded down to a resolution of 1 to 3 feet for us but the military is using much more accurate numbers. There are also other systems being developed that would be 100x more accurate than GPS... and come close to these indoor systems they're using. It'll happen, and Skynet will win, lol!
Krzysztof, I agree this would fall apart in the real world. But that's part of the point. To get real world applications, you first have to figure out the principle of operation and the mechanics behind them in a controlled lab environment with no external factors interfering. Only then can you realistically know where to start adding robustness to handle real world scenarios later on.