Recent Comments:
WiiBot cuts other Wiimote hacks down to size {Engadget}
Jan 25th 2007 10:35PM while I just pointed out the difficulty in doing real time control due to technical limitations, what you're saying about only one point of articulation is actualy... not true.
Robots (Fanuc's anyway) like this one have a variety of coordinate systems you can put them in, including one center on the end of the robot. There, you can give it transational (up, down, forword, sideways) movement and rotational movement commands. You tell the robot where to move the 'hand' and it figures out how to move all six joints to acomplish this. I've used all six axis of movement on the wiimote. It detects acceleration and rotation in all six axis. Exactly what the robot needs.
WiiBot cuts other Wiimote hacks down to size {Engadget}
Jan 25th 2007 10:29PM Fairly standard industrial robot. Looks like a Kuka? Coloration could be an ABB, but I'm used to their welders with the springs... So for the heavier lifters, you're talking apx $100,000 US for the higher end. Yeah, it's not chump change, but in the scheme of industrial things? And if anyone wants to talk about hourly costs due to wear on the robot... 10 year life time on them in a production, continous enviroment moving serious loads. The racquet and sword are both well under the rated load on that robot.
Second, if you look, its not like they stole the robot. It's a lab. Tinkering with a robot like that is part of their (and my) job. If I did that, I will tell you my boss would find it cool, as long as I didn't bill them for the time spent doing it.
As for your second comment... at my company, we typicaly fuse our robots at 30 amps @ 480 volts. That is a peak draw, when you're multiple servos on the robot, fast, while carrying a load. "Jacking up the companies power bill" is a bit of an overstatement.
Now... as for the people talking about variety of movement... not really. To say it outright for everyone, think the sword work in Twilight Princess. Same basic thing.
Industrial robots like this are basicaly set up to run a list of programs, and then have an external device/system/controller tell them what job to run, when. The interfacing and everything is already there, since it's what they're set up to do. However, giving free-form instructions as to position and orrientation is normaly done through the robot's own controller. For various reasons (safety being one of them), it's rather hard to get into that portion of it. You'd need to do a fair amount of custom coding, probably semi-special hardware as well. (Okay, this is a bit of a reach on my part - I've never seen a robot being given more than basic offsets by anything other than a Teach Pendant... and that has multiple safety circuits to emulate if you try to replace it...)
SAE Convergence 2006: The difficulty in implementing effective engine management systems {Autoblog}
Oct 20th 2006 9:54PM Although I was sadly not at the confrence, I think you're dismissing it out-of-hand. From what I'm reading, he's advocating taking everything to a feedback based control.
That is a heck of a lot more than Peak Preasure Point. The article outright talks about it being a high-dimensional system.
I won't say that using feedback based control is somehow a new idea. Taking system of equations, linearize, state and input weighting matrixs, ricotti equations, system solved.
Actualy... price of computing power? Assuming you can get a decent model for the behavior of the car, the cost is trivial. I've done some fairly complexe simulations on the MPC555, that at least a few years ago was not uncomon as a chip in core car systems. And in bulk they were... $20?
And although I ramble... at some point, there's something to be said for being the one to come forword and say it's time to do something.
Wish I had notes or an abstract of the speach though. Without that, I can't really say much.
