Got to see!

[airmcn_3]

Just check it out.

http://heli.stanford.edu/

[jetblackaircra]

I wonder if they have a special airworthiness cert.... hope the FAA doesn't go bust them. lol

[CenTexFlyer]

Nah, they have a DoD blessing.... can you imagine how hard it would be to hit that little S*B with small arms fire!?!?!?

The next big deal on the horizon is a UAV "killer" bot. Many already in the works.

[bmw330i]

Old news, but still interesting. Its' all done in the ground station if I understood it correctly. That Heli is just a big flying sensor sending it all to the GS then receiving the commands from the GS.
I'm curious about how it "learns". They don't ever really explain that. Does the software simply read the PPM signals while a pilot does the maneuver the first time? Then it just reproduces that? Seems the only explanation. I mean they don't set one on the ground to watch the other one and it learns it that way ;]
-David

[rbeall]

If they leave it vague then they don't have to prove anything. I'm sorry this just seems fake. You can't just "copy" a pilots commands and flip a switch to replicate it. I don't know if you guys can do any of the moves on a heli but I can tell you that they are never and I mean never "exactly" the same entry piont or exit points. Any journal articles or sensor sweets nothing....really??

[HELModels]

Whether this heli actually is computer controlled is unknown, but what BMW guesses seems possible to me. A Kalman filter needs an underlying math model/function. If each maneuver is a seperate model that the filter must try to follow and smooth, then my guess is that the idea is possible.

Buuuut...I have trouble with simple wiring.

[airmcn_3]

If they leave it vague then they don't have to prove anything. I'm sorry this just seems fake. You can't just "copy" a pilots commands and flip a switch to replicate it. I don't know if you guys can do any of the moves on a heli but I can tell you that they are never and I mean never "exactly" the same entry piont or exit points. Any journal articles or sensor sweets nothing....really??


Ya your probably right these 4 guys must have made it up in order to get there PHD....... :rolleyes:

Pieter Abbeel
Adam Coates
Timothy Hunter
Morgan Quigley

Have a look at Adam's award winning paper, not exactly something you get by making it up.

[dmgoedde]

Nonsense... this is real. I am in awe of their IMU skills. They have huge cahunas. What could be more difficult than this??? Makes airplane autopilots seem like childs play in comparison.

[HELModels]

I see, not different models for each maneuver, but different trajectories and alot of statistical linear algebra mumbo jumbo that a KF smooths out. The initial pilot's maneuvers and all the data generated by those maneuvers Are the training data.

I'll stick to the simple stuff...like I really have a choice

[airmcn_3]

I see, not different models for each maneuver, but different trajectories and alot of statistical linear algebra mumbo jumbo that a KF smooths out. The initial pilot's maneuvers and all the data generated by those maneuvers Are the training data.

I'll stick to the simple stuff...like I really have a choice


Your spot on!