PixHawk competing at IMAV 2010

Our Swiss friends continue to impress with a new, lighter version of their PixHawk quad with optical navigation, which is competing at IMAV 2010 this week. From the team's update:

"We participated in the Indoor Autonomy and Indoor Dynamics missions. The Indoor Dynamics did start with some technical difficulties, we could not transmit the start signal any more to the helicopter because of heavy interferences on the 2.4 GHz link (which was fine during training). However as our approach is purely based on onboard computation, we changed the strategy for the second mission (Indoor Dynamics) and just automatically started after turning on the rotors.

That mission went better - we could fly autonomously and demonstrated for the first time on the International Micro Air Vehicle competition an onboard computer vision system. Our localization is based on articial markers. The flight inside the competition times was not perfect, but shortly afterwards we captured the video below, showing a fully autonomous flight in a figure eight at the IMAV 2010 Indoor Dynamics setup.

Tomorrow is the IMAV 2010 Outdoor Competition day and we'll enjoy the full day spectating the outdoor teams competing. We'll also bring our equipment, maybe we can help somebody out with a screwdriver or tweezers."

> Not to minimize any one's work here I'm sure there is more to this than it looks

You have no idea how hard this is!!! I can not state how impressive this demo is to me and how much processing power they must have on board to achieve it.

> No it looks like they are differentiating between images on the paper and using that to navigate.

Exactly what they are doing and it is hard and extremely complex to perform. I can understand that it does not look flashy but it is technically as cool as it gets and right at the edge of computer vision as it is all done on-board.

It is simple a case of "those with the eyes to see" I think. It is so easy to make a video look awesome when really it is a hacked together mess that only works in some very certain circumstances, yet you can lead people on to let them think it works like that all the time. I have seen it here over and over, simple problems being blown up to be some huge advance. This video is the opposite, a simple clear video with no hype that so some seriously smart people doing some advanced work.

You need to be a member of diydrones to add comments!

Join diydrones

Comments

Tumba July 10, 2010 at 4:52am

Leon:
There are some articles (in German) and fresh forum dicussions (in German).
http://www.heise.de/newsticker/meldung/IMAV-2010-Roboterschmetterli...
http://www.heise.de/newsticker/meldung/Micro-Air-Vehicles-Flugdynam...

I reached 3 titles. :-)

"Outdoor Autonomy Competition - Fixed Wing" 3rd place Full Autonomous -
GPS - Aircraft Lerche 500
http://diydrones.com/forum/topics/presentation-30-x-40-mm
http://forum.mikrokopter.de/topic-14900.html

"Indoor Flight Dynamics Competition - Rotor Wing" 3rd place - Remote
Controlled - Quadcopter
http://www.vimeo.com/12925133
http://forum.mikrokopter.de/topic-16412.html

"Outdoor Flight Dynamics Competition - Fixed Wing" 2nd place Remote
Controlled - Aircraft Lerche 400
http://gallery.mikrokopter.de/main.php/v/MKBilder/EMAV2008.jpg.html
+ Proceedings of the European Micro Air Vehicle Conference (EMAV08): Flight
experiments of different MAV concepts with a wingspan of 0.4 m

Best regards Tumba
Leon July 10, 2010 at 1:34am

This MAV has more computing power than my PC! It is very surprising to find a so powerfull PC on a MAV!

About the IMAV 2010 competition, do somebody have some news about it?

Leon.
Bill Porter July 9, 2010 at 3:13pm

If this is all true, then this is a very poor video. This could be done with edge detection and a state machine, and the same video, a copter circling a pattern, could be filmed.

Re-reading Matthew's first comment, I guess this is what he meant.

I don't want to degrade any advance is video processing, but they need to film something that can't be achieved by a simpler method.
Tumba July 9, 2010 at 1:57pm

I was on the IMAV 2010. It was only 150 km from my home. I have also seen this vehicle during the flight. All markers are different, the vehicle knows exactly the location. It was impressive.
David Ankers July 8, 2010 at 10:44pm

> Not to minimize any one's work here I'm sure there is more to this than it looks

You have no idea how hard this is!!! I can not state how impressive this demo is to me and how much processing power they must have on board to achieve it.

> No it looks like they are differentiating between images on the paper and using that to navigate.

Exactly what they are doing and it is hard and extremely complex to perform. I can understand that it does not look flashy but it is technically as cool as it gets and right at the edge of computer vision as it is all done on-board.

It is simple a case of "those with the eyes to see" I think. It is so easy to make a video look awesome when really it is a hacked together mess that only works in some very certain circumstances, yet you can lead people on to let them think it works like that all the time. I have seen it here over and over, simple problems being blown up to be some huge advance. This video is the opposite, a simple clear video with no hype that so some seriously smart people doing some advanced work.
Matthew M July 7, 2010 at 10:49pm

No it looks like they are differentiating between images on the paper and using that to navigate. My comment is what are they showing off the mems gyro stabilization, advances in battery technology, or processing ability? All of which are great advances in technology but had nothing to do with them. Any way it's all for fun and profit I'm sure. :D
Bill Porter July 7, 2010 at 7:42pm

I agree Matthew, this seems like a rather easy challenge to mount. It looks like it's just edge detecting the paper. That would be easy to implement in say Robo_Realm or maybe a CMU cam.
Matthew M July 7, 2010 at 6:13pm

Not to minimize any one's work here I'm sure there is more to this than it looks. Basically it looks like a line follower robot on a quadcopter. Instead of lines they are showing off the computational ability of modern processors by using pictures? I'd call this "on the backs of giants" but I'm not sure that would be appropriate. :D

This reply was deleted.