Back
[00:11:20] <anonnumberanon> what is "the plant"?
[00:12:16] <rue_house> its a system
[00:12:27] <anonnumberanon> what system?
[00:12:34] <rue_house> usually a feedback system
[00:12:43] <rue_house> its got all kinda responce attributes
[00:12:51] <anonnumberanon> I've never heard of that term.
[00:12:57] <rue_house> its common
[00:13:11] <anonnumberanon> Not in a bachelor of engineering degree.
[00:13:14] <anonnumberanon> :)
[00:13:27] <anonnumberanon> should have went for the master, I know
[00:13:43] <rue_house> thats cause the coruse is just a money grab
[00:13:56] <rue_house> its not meant to give you skill or employability
[00:14:13] <anonnumberanon> the plant?
[00:14:18] <anonnumberanon> or BAEE?
[00:14:19] <rue_house> just to make you pay and hopefully give you the impression your money wans't wasted
[00:14:36] <anonnumberanon> My money was far from wasted.
[00:14:48] <anonnumberanon> And it was well worth it.
[00:15:06] <anonnumberanon> Sure you can learn on your own but I needed school for motivation.
[00:15:11] <anonnumberanon> Grades, and stuff.
[00:15:14] <rue_house> they did a good job on you then, you aren't pissed off that they stole your money
[00:15:45] <rue_house> remember, there are only two ways to make money from a university session
[00:16:23] <orlock> We are continually hiring engineers here
[00:16:36] <rue_house> green ones?
[00:16:46] <anonnumberanon> orlock, cause you can't find good enough ones or cause you're growing?
[00:16:47] <orlock> just put on 3 graduates
[00:16:57] <orlock> and some experienced
[00:17:04] <rue_house> the university kids who use a 2Ghz processor to just a java app to flash a light?
[00:17:18] <anonnumberanon> lol that's a meme
[00:17:37] <orlock> nah heh
[00:17:39] <orlock> mechanical
[00:17:41] <orlock> mechatronics
[00:17:44] <orlock> electrical
[00:17:47] <anonnumberanon> based on a slight background of truth, but meme
[00:18:15] <rue_house> hmh, so lots of products designed to operate at 100% the rating on the face page of the datasheet?
[00:18:35] <anonnumberanon> depends on the school, students in my school didn't care and I did that's why our quadcopter was the only that flew
[00:19:00] <anonnumberanon> 50% of the robot projects were nice and the other 50% were junk
[00:19:16] <anonnumberanon> (I don't consider the quadcopter section robots.)
[00:19:25] <rue_house> (the irf1404 has an asterisk beside the current rating on it, that if you look REALLY hard, says that the rating is completely hypothetical and that the transistor would blow up at a current MUCH lower than it"
[00:19:27] <rue_house> )
[00:19:30] <anonnumberanon> by robots I mean they had to make servo robot arms, things on wheels
[00:19:35] <anonnumberanon> only a few were good
[00:20:23] <rue_house> I was stooping when I made the hobby servo based robot arms, and I appologize
[00:20:28] <anonnumberanon> orlock, so what were you expecting as far as level, from the mechatronics guy, if you don't mind me asking
[00:20:29] <anonnumberanon> ?
[00:21:37] <anonnumberanon> rue_house, it's kinda like about what you want to focus on, if you want to focus on other things than the mechanical part, user servos
[00:22:02] <rue_house> arm6 will be better
[00:22:27] <orlock> anonnumberanon: No idea, we have literally hundreds of them here
[00:22:28] <anonnumberanon> If you're a mechanical engineer though, I would expect you to be able to make robot structure parts and robot actuators out of school.
[00:23:04] <rue_house> http://ruemohr.org/~ircjunk/robots/arm6/p1070894.jpg
[00:23:08] <orlock> anonnumberanon: all we do is make fancy robots
[00:23:33] <anonnumberanon> orlock, That's my dream job.
[00:24:03] <orlock> mainly biotech
[00:24:07] <orlock> but some others
[00:24:13] <orlock> used a fancy touchscreen coke machine?
[00:24:22] <rue_house> wish I had more time to do better stuff
[00:24:30] <anonnumberanon> oh yeah you told me about that the other day
[00:24:40] <rue_house> you write part of that thing?
[00:24:42] <anonnumberanon> are you an engineering manager, project manager, engineer?
[00:24:52] <orlock> digital janitor
[00:24:55] <orlock> https://en.wikipedia.org/wiki/Coca-Cola_Freestyle
[00:25:04] <anonnumberanon> you clean data?
[00:25:38] <anonnumberanon> oh these ones that you use actually at fastfood places, I see
[00:25:50] <orlock> i unclog the tubes
[00:25:52] <anonnumberanon> i thought you meant some drinks and snacks dispensers
[00:26:06] <rue_house> a subway here got what I think is a newer one
[00:26:14] <rue_house> larger thouch screen
[00:26:17] <rue_house> portriat
[00:26:33] <orlock> they also did a bio-3d-printer
[00:27:51] <anonnumberanon> rue_house, define "better"
[00:28:39] <rue_house> better is like a wheel that is actaully, really round, and not square with pillows duct-taped to the flat sides like all the wheels you see currently used
[00:30:03] <anonnumberanon> i wish you did more large mecha stuff
[00:30:28] <rue_house> I'm trying
[00:30:34] <rue_house> I'm working on tools and skills
[00:30:51] <orlock> my desktop died
[00:30:53] <anonnumberanon> you "just" have to find a hack to be able to make somewhat lightweight, very powerful, fast actuators
[00:30:59] <rue_house> I got a welder, lathe, sandblaster, smelter for casting aluminum
[00:31:07] <rue_house> orlock, restart it
[00:31:15] <orlock> need to fix it so i can get back to CAD
[00:31:16] <orlock> i did
[00:31:16] <orlock> for 20 minutes
[00:31:18] <rue_house> anonnumberanon, air muscles
[00:31:25] <orlock> didnt even bring up POST
[00:31:30] <anonnumberanon> rue_house, yeah prioritize that
[00:31:37] <rue_house> oh, you mean DEAD
[00:31:49] <rue_house> I have those tire pressure sensors comming
[00:31:51] <anonnumberanon> find out if it makes sense as an actual block in your mecha's system
[00:31:54] <rue_house> that was a wickid idea
[00:31:58] <orlock> anonnumberanon:
http://www.danahermotion.com/website/com/eng/index.php http://dovermotion.com/
[00:32:07] <anonnumberanon> orlock, maybe time to check your power supply?
[00:32:53] <anonnumberanon> orlock, think, more powerful
[00:32:58] <orlock> anonnumberanon: Dragged it into work to do today
[00:33:01] <rue_house> orlock, you have to reboot the router before we can continue your service session ;)
[00:33:15] <anonnumberanon> as strong as the hardware that powers bobcats and stuff
[00:33:27] <rue_house> anonnumberanon, I'm confident in the air muscles
[00:33:30] <anonnumberanon> the actuators on those things, though, are too slow I think
[00:33:33] <orlock> anonnumberanon:
http://www.enermax.com/home.php?fn=eng/product_a1_1_1&lv0=1&lv1=54&no=5
[00:33:39] <rue_house> best power to weight ratio and cost point
[00:33:48] <rue_house> but the robot has to be 12' tall to pull it off
[00:33:52] <rue_house> AND NO SMALLER
[00:34:00] <anonnumberanon> that's perfect
[00:34:16] <anonnumberanon> it feels like a "minimum mech size"
[00:34:19] <rue_house> did you see the mockup?
[00:34:29] <anonnumberanon> i saw the welded tubes you made
[00:34:36] <anonnumberanon> but that one, that was a lot smaller
[00:34:53] <anonnumberanon> it does give kind of an impression that a larger one could actually happen
[00:35:12] <anonnumberanon> problem with these things, sky is the limit but the Japanese one costs 1 million to build
[00:35:14] <rue_house> http://ruemohr.org/~ircjunk/robots/sparrow1/p1040018.jpg
[00:35:27] <rue_house> yes, its called 'sparrow 1'
[00:35:42] <orlock> anonnumberanon: btw, those two other URL's are our sister companies
[00:35:43] <rue_house> its just a light duty, small utility mecha
[00:35:45] <anonnumberanon> oh I haven't seen that.;
[00:35:53] <rue_house> thats how big a 12' mecha is
[00:36:10] <anonnumberanon> yeah like the one they use in the movie Alien, which Sig uses to fignt the Alien queen
[00:36:22] <orlock> Aliens
[00:36:23] <orlock> _s_
[00:36:33] <anonnumberanon> hahaha
[00:36:36] <rue_house> its crotch clears my head
[00:36:44] <orlock> in Alient she used a grappel to blow it out the goddamn airlock
[00:36:49] <orlock> -t
[00:37:31] <anonnumberanon> how hard to gather enough air muscles for the entire thing, or at least to start, for the top like the arms, mounted on a wheeled-base (or tank tracks may be cheaper cause directly taken from a bobcat)
[00:37:33] <rue_house> it looks much smaller laying down
[00:37:35] <anonnumberanon> ??
[00:38:02] <rue_house> yea, I'm working that direction so i can use it to finish a brick wall
[00:38:10] <rue_house> you saw the finger?
[00:38:40] <rue_house> http://ruemohr.org/~ircjunk/robots/sparrow1/p1070331.jpg
[00:38:43] <anonnumberanon> no?
[00:39:05] <rue_house> I'm dropping the useless fingers to save complexity
[00:39:22] <anonnumberanon> lol. good start, if those muscles work you'll actually get something usable and neat
[00:40:08] <rue_house> awe putz, the forearm images aren't posted
[00:40:16] <rue_house> remind me to update that later
[00:40:33] <rue_house> I figured out how to do the forearm bearing
[00:41:18] <rue_house> working on it is exciting
[00:41:29] <rue_house> I'v recently worked out how finger ligaments go,
[00:41:40] <rue_house> but I might build a more robotic temporary gripper
[00:41:50] <rue_house> get it working and upgrade the hell out of it
[00:55:00] <Wetmelon> Jak_o_Shadows: 5th order
[00:56:13] <Wetmelon> Zeros at 1.298, 1.114, 2.424; poles at 0, 19.68+j10.9, 19.68-j10.9, 12.14
[00:56:20] <Wetmelon> Gain of 2020.775
[00:57:17] <Wetmelon> Sorry 4th order
[00:58:12] <Jak_o_Shadows> orlock, where are you though?
[00:59:16] <Jak_o_Shadows> Interesting.
[00:59:32] <Jak_o_Shadows> sorry, just read the scrollback
[00:59:50] <rue_bed2> those are the values from a simulator to make it work in a simulator
[01:02:02] <Jak_o_Shadows> well, yeah, but you gotta start somewhere
[01:02:41] <rue_bed2> I'v never seen a 555 simulator that even got me in the right ballpark
[01:05:12] <Jak_o_Shadows> How would you go about controlling something? Like, how would you get the gains for your control loop?
[01:05:35] <Jak_o_Shadows> PID*?
[01:07:20] <rue_bed2> miss and error
[01:07:54] <rue_bed2> usually you dial it all down, dial it up one param at a time, looking for particular behaviours
[01:07:59] <Jak_o_Shadows> Even ignoring simulators, isn't there empirical ways of getting better ballparks for gains?
[01:08:45] <Jak_o_Shadows> This is a field where I need to get more practical experience.
[01:08:46] <rue_bed2> this is why Idislike pid
[01:09:07] <rue_bed2> if you have a digital scope you can capture the step responce and tune it with that
[01:09:13] <Wetmelon> rue_bed2: Eh, we're going to test it on monday
[01:09:24] <Wetmelon> 555 simulator?
[01:09:36] <rue_bed2> simulator values are for simulators, not for real world
[01:09:53] <rue_bed2> LM555, its the oldest chip, you should look it up sometime
[01:10:00] <Wetmelon> Oh I see what you're saying. A simulator for a 555 chip
[01:10:11] <Wetmelon> I have a linear state space model of the plant
[01:10:28] <Wetmelon> we tested it with our previous, 2nd order controller and it worked great but was slow
[01:10:39] <rue_bed2> mhm
[01:10:45] <Wetmelon> So we added a 2nd order lead compensator
[01:12:57] <Wetmelon> ... well I suppose I didn't. I technically removed a pole and added a complex pole (i.e. low-pass filter), since the controller reacted very poorly to simulated noise
[01:13:22] <Wetmelon> Aaaanywho
[01:15:07] <Wetmelon> Jak_o_Shadows: Rue is right. If you can get the step response you can derive the transfer function from that. You *can* use classical techniques to derive the critical gains for PID control so you know you're in the right ballpark. But if you can hit the system with a step or impulse or sine wave sweep, you're much better off
[01:15:29] <Jak_o_Shadows> Yeah.
[01:15:41] <Jak_o_Shadows> I've never actually implemented control on anything, only looked at it on a computer
[01:18:59] <Wetmelon> I wish we had a lab for this class, but at least we have a decent project. It's basically one motor from a quadcopter, constrained to move vertically. We had to derive the ESC-to-Thrust model from step responses and then derive the equations of motion etc, develop a high performance controller, and test it. Fun stuff
[01:20:56] <Jak_o_Shadows> ah, cool
[01:21:23] <Wetmelon> rue_bed2: I should say we tested our previous controller on the actual hardware. Extremely stable but otherwise slow.
[01:21:40] <Wetmelon> More-or-less verified our model though.
[09:41:02] <rue_house> you would be really surprised how much the real world can varry from the results of a simulator
[10:20:50] <deshipu> verry
[12:43:10] <jason__> This class just opened up on Coursera:
https://www.coursera.org/learn/robotics-perception . I took a similar course at my school and it was good stuff.
[22:57:05] <anonnumberanon> jason__, not a single one of the lectures talks about hardware
[22:57:23] <anonnumberanon> no hardware, no robot perception
[22:57:57] <anonnumberanon> you have to have hardware labs at the same time or else you never see the point of what you're trying to study
[22:58:25] <anonnumberanon> >into the trash the class goes
