Back
[04:47:44] * z64555 attempts to code something for the kalman filter
[04:50:38] <deshipu> z64555: A/B filter works almost as well :P
[04:56:19] <anonnumberanon> I have a phone interview today :)
[04:56:29] <anonnumberanon> Hopefully I don't make an ass of myself.
[04:57:35] <z64555> deshipu: yeah, but I've been putting this off for too long
[04:57:38] <deshipu> making an ass of yourself is your basic human right
[04:58:08] <z64555> anonnumberanon: I hear wearing what you'd wear to a normal interview helps
[04:58:51] <deshipu> but my hotdog costume is in the laundry :(
[04:59:19] <z64555> bummer, I guess a ketchup-stained shirt will have to do
[04:59:44] <z64555> er, wait a minute
[05:02:43] <z64555> re: kalman filter. Although the A/B filter would do good enough for craft stabelization, more precision will be needed when I get around to inertial navigation
[05:03:03] <deshipu> it's also super-fast
[05:03:06] <deshipu> compared to kalman
[05:03:10] <z64555> oh definitely
[05:05:22] <z64555> I've used an A/B filter before with success, the quad video I linked a few days ago demostrated that
[06:13:11] <z64555> duh, σ is a random variable
[06:32:40] <theBear> it's a funny looking o from here
[06:32:50] <theBear> o with a fancy hairdo like peewee herman
[06:33:01] <theBear> or umm,tintin
[06:48:34] <joga> σ__σ
[07:06:38] <z64555> ah, shoot, I can't tell by inspection whether some of these are matrices or vectors
[07:15:24] <jhylands> z64555, what language are you programming in
[07:16:03] <z64555> C
[07:16:09] <jhylands> oh well
[07:18:12] <z64555> C++ has some nice features in it, but some freeware IDE's put a size limit on how much memory you can code
[07:18:19] <z64555> which... tends to be quite small
[07:18:56] <jhylands> I wrote a nice GUI in python that allows you to visually inspect & explore objects
[07:19:06] <jhylands> but of course it only works in python
[07:19:19] <z64555> oh, sorry. I meant by inspecting the formula on paper
[07:19:24] <z64555> not by code
[07:19:28] <jhylands> heh, ahhhh
[07:20:14] <z64555> going back the wiki on kalman filters:
https://en.wikipedia.org/wiki/Kalman_filter#Details
[07:20:21] <z64555> s/back/by
[07:20:47] <z64555> I'll have to look for some primary sources to see if they have more detail on it
[07:21:19] <jhylands> I gave up on trying to understand Kalman filters a long time ago
[07:21:34] <jhylands> decided from now on, I only use IMU modules that do their own filtering
[07:21:54] <z64555> Yeah, I had to spend a few days mulling this over
[07:22:45] <z64555> also a good choice, but without an kalman filter on-board you'll have difficulty doing stuff like navigation
[07:23:06] <z64555> *automated navigation
[07:23:28] <z64555> at least, that's what I currently believe
[07:23:40] <jhylands> you're almost certainly right
[07:25:01] <z64555> learning is an iterative process :)
[07:50:56] <anonnumberanon> but some freeware IDE's put a size limit on how much memory you can code
[07:51:00] <anonnumberanon> what the actual fuck
[07:55:29] <z64555> sorry, not freeware, free licensed
[07:55:46] <z64555> Freescale's Codewarrior is the example I have in mind.
[07:56:01] <z64555> you have to buy a license in order to get rid of the C++ limit
[07:58:06] <z64555> re: my earlier connundrum, I think they're all matrices
[07:58:48] <z64555> it's a mathematical convention to have matrices and transforms as capital letters
[08:00:10] <z64555> the wikipedia example has H as a vector, but that's probably a simplification of {{ 1, 0 }, {0, 0}} since the multiplication result is the same as {1, 0}
[09:36:37] <rue_house> thanks, I'll remember to not use freescale
[10:01:52] <anonnumberanon> Even more importantly, keep your memory usage low :)
[10:19:23] <theBear> no problems there, took measures almost 2 decades ago :)
[11:10:43] <z64555> anonnumberanon: yeah, but the C++ limit was rather pathetic, like 10KB or so
[11:11:03] <z64555> compared to the entire flash memory of 256KB
[11:55:47] <anonnumberanon> z64555, Yeah that's weak if you need to make a program for naviguation which would take a lot more. Man even the program for PID stabilization and flight only for my quadcopter was 12 kB.
[11:58:26] <deshipu> yawn
[11:58:40] <deshipu> good evening robots
[11:59:30] <deshipu> what's cooking?
[12:00:08] <veverak> well
[12:00:15] <veverak> it's sad that only few people get this reference :/
[12:00:18] <veverak> :D
[12:00:37] <deshipu> really?
[12:01:05] <veverak> I know only you
[12:01:08] <veverak> :D
[12:01:15] <deshipu> I was raised on wabbits
[12:27:01] <anonnumberanon> deshipu, Spaghetti Bolognese with bison meat, you?
[12:38:19] <deshipu> anonnumberanon: no, but we have żubrówka
[12:41:26] <anonnumberanon> That name has so many accents that it already sounds delicious. Can you cook it?
[12:46:05] <anonnumberanon> >google it >find out it's booze >kill self
[13:11:19] <deshipu> anonnumberanon: it's booze made with bison grass
[15:37:34] * z64555 noms food
[15:37:38] <z64555> ok back to work doing nothing
[15:38:00] <deshipu> z64555: no nothing more efficient!
[15:38:28] <deshipu> waste time faster
[15:55:08] <z64555> :)
[16:19:53] <z64555> ok, got that settled. now i'm wondering what the effect of a control vector would be
[16:20:46] <z64555> I'm guessing this is the output from a controller before its passed through the transform gains needed to move the acutators of a system
[16:21:55] <z64555> spelling os down, i repeat, spellings is down
[16:25:29] <z64555> yeah, pretty sure that's what it does. Since the control vector is added in the a priori state
[16:25:37] <z64555> or phase, or whatever they call it
[17:30:18] <z64555> good, i don't need another 2d array to hold the transpose
[17:30:27] * z64555 checks his matrix math
[17:54:01] <mintyant> hi... I would love to wire together some motors and wheels to make them spin. But i'm honest to god not sure what motor, what wheel, and what shit in between I need just to get the spinning wheel
[17:54:24] <mintyant> I'm looking at gearbox motors and ball bearings, what's next?
[17:55:17] <z64555> :)
[17:55:39] <z64555> If it's just a wheel that you want to spin, then anything that can fit on the shaft of the motor will do
[17:56:12] <z64555> Pendantry aside, What's attached to your wheels, other than the motor/gears?
[17:56:54] <mintyant> Honestly, I was just like "I want to control some wheels with a raspberry pi"
[17:57:11] <mintyant> but i've never done work with any sort of motor, or RC cars in general..
[17:58:37] <z64555> so, a sumobot
[17:58:55] <orlock> mintyant: You generally wont be interfacing thewheels directly to the rpi
[17:59:15] <orlock> mintyant you'll want some sort of motor driver, most common is a H bridge
[17:59:52] <z64555> right, the motor driver supplies the power to the motor, and the driver is controlled by your µC (in this case the raspi)
[18:00:11] <z64555> The driver you need is dependant on the motor you want
[18:00:32] <z64555> and the motor you want is dependant on your wheels and your chassis
[18:00:40] <orlock> The one type you might be able to control is a simple servo
[18:01:04] <orlock> but thats not ideal for driving wheels
[18:01:10] <orlock> but it lets you move smething
[18:01:15] <z64555> yes
[18:01:26] <orlock> or with some resistors, you could turn lights on and off
[18:01:41] <orlock> without resistors, you make things burn out
[18:01:55] <z64555> heh, especially LED's
[18:02:17] <z64555> ye old lamp bulbs don't need a resistor, because they are the resistor.
[18:02:55] <mintyant> Okay... thank you, that's good to know. So I need a gearbox, which connects to a motor driver, which connects to my pi
[18:03:05] <orlock> yup
[18:03:21] <orlock> servos include the driver and gearbox, but only rotate 180 degrees or so
[18:03:39] <z64555> eh, a gearbox is the mechanical gearing, motor not included
[18:03:39] <orlock> you can get motors with integrated gearbox
[18:03:55] <z64555> yes
[18:04:39] <orlock> Tamiya makes a pile of parts, or you can get them cheap from ali or dx
[18:04:45] <orlock> or tear apart toys
[18:04:51] <mintyant> Okay, I found a couple guides that seem to do this. Looks like these two components are probably sufficient:
http://pastebin.com/kHY9XuMd
[18:05:04] <mintyant> though, I still don't quite have wheels or how to connect them
[18:05:51] <orlock> Tamiya will sell something to suit
[18:06:08] <orlock> those tracks are fiddly though
[18:06:17] <orlock> hard to tension right
[18:06:56] <orlock> but yeah, that should work
[21:28:47] <z64555> oh, that's an interesting thought
[21:29:32] <z64555> with a kalman filter, the sensors can have a different dt from each other
[21:30:05] <DagoRed> They may.
[21:30:39] <z64555> makes a mess of things, but it is possible
[21:30:55] <DagoRed> It's not that bad.
[21:32:36] <z64555> letsee... the H matrix would be different per sample, essentially turning on and off the senor inputs
[21:32:59] <z64555> the gains would be different, as would the Q covariance
[21:33:11] <z64555> since Q would have to be recalculated
[21:34:20] * DagoRed nods
[21:35:55] <z64555> the R measurement covariance would also be different, but it may be possible to apply H to it and shut off the noise on sides that aren't getting updated
[21:36:36] <z64555> think that's everything
[21:37:54] <z64555> not applicable to the current attitude angles stabelization, but it will likely play a key factor when I get around to waypoint navigation, where the GPS signals are slow-ish
[21:39:16] * DagoRed has done this before.
[21:40:02] <DagoRed> You are fine. If you want.. you can use a priori estimator on the GPS with the IMU data that allows for a greater confidence of measurements.
[21:42:50] <z64555> that's the plan :)
[21:43:12] <theBear> i'm super-fine ! like the sugar
[21:43:52] <z64555> powdered sugar, and not the stuff from cocoa leaves?
[21:43:59] <z64555> just making sure...
[21:44:15] <DagoRed> Nose candy?
[21:44:51] <z64555> blargh, #multirotors's raunchiness is growing on me
[21:45:34] <theBear> nah, i got a scan of an old packet somewhere, standard single-serve sugar kinda rectangle from a cafe kinda deal, it clearly says "SUPERFINE" in big letters....
[21:45:54] <theBear> https://www.youtube.com/watch?v=q6tfMVtYs4w there, that'll set it all clear in your mind
[21:46:07] <z64555> note how he leaves
[21:46:11] <z64555> oh wb
[21:46:17] <theBear> woops, sneaky cursor-focus
[21:46:32] <DagoRed> it happens.
[21:46:44] <theBear> usually it's itchy-mouse-trigger-finger and stupid default middle-click-on-channel-list , but not this time, oh no sir :)
[21:46:47] <z64555> "this video contains content from columbia" what?
[21:47:10] <theBear> sidenote: theBear NEVER leaves unless he accidentally upsets the nicotine-coated connections to the modem in the next room :)
[21:47:24] <theBear> seriously ? sorry maan, usually if that happens to anyone it would be me
[21:47:27] <theBear> but wait, i got a trick maybe
[21:48:11] <theBear> https://www.youtube.com/embed/q6tfMVtYs4w try thatversion
[21:48:31] <theBear> obvious trick a buddy showed me last night, definately works to avoid stupid login-to-prove-age rejections
[21:48:36] <z64555> shouldn't that be "Colombia?"
[21:48:48] <z64555> unless somebody is trolling youtube?
[21:49:36] <z64555> theBear: also blocked. is there a text version of it
[21:49:45] <theBear> google troll themselves almost at microsoft kidna levels the last year or two
[21:50:06] <theBear> umm, it's a clip from an old old tv show, the text just wouldn't do it justice if it did exist
[21:50:23] <z64555> not monty python, is it
[21:50:51] <theBear> and i can't believe that it would still fall under copyright, letalone be tracked/blocked... it was long-forgotten and many-reruns-passed long before teh web existed
[21:51:08] <theBear> nah, monkey magic clip "All Forms are Emptiness and Emptiness are in all Forms" it titles
[21:51:11] <theBear> titled
[21:51:24] <z64555> so, somewhere in the late 60's early 70's
[21:51:59] <theBear> it was into the double-digit repeat zone by the time i was in primary school mid 80s i think
[21:52:38] <theBear> but nice memories, and that clip not only covers the major characters and sums them up reasonably, but has some nice light-philosophy/buddhism kinda stuff sprinkled in
[21:53:04] <theBear> https://www.youtube.com/watch?v=LAbE36IBTZY try that (intro sequence) if ya not familiar, or want some nostalgia
[21:53:12] <z64555> ahh, so it's propaganda or at least somebody think it is. lol
[21:54:41] <z64555> ah, one of the original stone monkey stories.
[21:55:29] <z64555> see that crop up from time to time, dragon ball, and forbidden kingdom are examples
[21:55:50] <z64555> stone monkey/ monkey king
[21:57:11] <z64555> hm... maybe its something from hong kong?
[21:58:54] <z64555> no, it's actually from japan. BBC did english dubs in 79
[21:59:07] <z64555> https://en.wikipedia.org/wiki/Monkey_%28TV_series%29
[22:00:22] <z64555> neat
[22:02:37] <z64555> ohhh dangit. now I've got ideas for microdrones
[22:02:44] <z64555> one project at a time, z64!
[22:06:12] <theBear> be confident, think mega-drones
[23:18:35] <z64555> mega drones is like, on the 10hp and up scale
[23:19:13] <z64555> although, I do remember a heavy drone, about 2Kg or so, that was an octoquad
[23:19:25] <z64555> hm, either an 8 or 12, actually
[23:19:36] <z64555> University of Adelaide made it a few years ago
[23:19:45] <z64555> had some neat prop hubs on it
[23:19:54] <orlock> eh bullshit
[23:19:59] <orlock> nothing worthwhile comes from adelaide
[23:20:01] <orlock> except coopers
[23:23:01] <theBear> adelaide, that's the middle bottom one innit ?
[23:23:18] <theBear> or the island one ?
[23:23:30] <z64555> dunno
[23:24:01] <flyback> THE NATIONAL WEATHER SERVICE HAS ISSUES A "YOU GO TO HELL!.............YOU GO TO HELL AND YOU DIE!" ..........FOR YOUR AREA
[23:24:55] <z64555> ?
[23:25:09] <flyback> BAD bad wind and rain
[23:25:28] <z64555> any flying cows?
[23:25:37] <flyback> not yet
[23:26:04] <z64555> sooo. probably not good weather to stress test a quad
[23:26:18] <z64555> gotta wait until the cows fly home
[23:26:42] <flyback> no
[23:26:47] <flyback> quads suck ass
[23:27:02] <flyback> in terms of bad conditions vs fail
[23:27:10] <flyback> you lose a motor, etc and its a brick
[23:27:15] <DagoRed> flyback: aren't you in Maryland?
[23:27:18] <flyback> pa\
[23:27:21] <flyback> close enough
[23:27:23] <DagoRed> which part?
[23:27:27] <z64555> I blame the crappy ESC's
[23:27:32] <DagoRed> Pittsburgh or Philly?
[23:27:33] <flyback> 45 mins from pittsburghm beaver falls
[23:27:43] <flyback> z64555, no dude it's physics
[23:27:44] <DagoRed> .weather pittsburgh, pa
[23:27:44] <makepi> DagoRed: Pittsburgh, PA: Overcast, 42.3 F (5.7 C), 29.60(+), Feels Like: 36 F (2 C), Humidity 85%, WindFrom the West at 12.3 MPH Gusting to 32.4 MPH - Last Updated on March 2, 12:02 AM EST-
http://j.mp/1oXzEQt
[23:27:53] <flyback> quads are totally dependent on balance
[23:27:58] <flyback> that's one thing sucky abotu them
[23:28:04] <z64555> well, yeah
[23:28:40] <flyback> they almost should put one way clutches on them so a prop could auto rotate if a motor failed
[23:28:41] <z64555> I was talking about flight stability testing, not so much motor limits...
[23:28:49] * z64555 gets himself confused again
[23:28:54] <DagoRed> z64555: If you go collective pitch with flybar assemblies you're better off in the wind with quads... but 16 DoF is a bitch to design.
[23:29:06] <z64555> neat
[23:29:10] <DagoRed> I've done it... don't.
[23:29:16] <z64555> lol
[23:29:48] <DagoRed> Cool part, we were able to hover sideways. We created a device to do under bridge inspection.
[23:30:17] <z64555> hover sideways? real neat
[23:30:27] <DagoRed> Well... 20 degrees from center.
[23:30:31] <flyback> DagoRed, assuming you don't get patent lawsuits
[23:30:33] <flyback> there is a money maker
[23:30:38] <flyback> espically in pa
[23:30:42] <flyback> we got a lot of bridges ready to drop
[23:30:44] <z64555> ah, strong winds
[23:30:55] <flyback> they finally closed fallston bridge last year
[23:30:56] <flyback> sadly
[23:30:57] <DagoRed> Yeah... I live on the other side of the state. Shit's falling apart.
[23:31:36] <DagoRed> We're closing a major one in Bethlehen... new street bridge hasn't had a proper sidewalk in 5 years. They put up a concrete barricade and force pedestrains to walk on the street.
[23:33:31] <flyback> I got pics of 10th street bridge
[23:33:41] <flyback> closed in 1978 used to see this bridge on way to hs
[23:33:47] <flyback> go under it fishing with dad
[23:33:51] <flyback> never put 2 and 2 otgether
[23:33:59] <flyback> they finally took it down early 2000's
[23:34:32] <DagoRed> Your side of the state is an infrastructure nightmare.
[23:34:45] <flyback> yep
[23:34:56] <flyback> don't forget centrailia
[23:35:02] <DagoRed> I'm not to far from there.
[23:35:07] * DagoRed wants to check that out soon.
[23:35:16] <DagoRed> 45 minutes to an hour away.
[23:35:22] <flyback> they shoudl just setup a power plant or something to take advantage of the heat
[23:35:32] <DagoRed> Centralia is actually... getting better. The fire moved from the heart of the town.
[23:35:45] <DagoRed> The heat is deeper in the mine.
[23:35:56] <DagoRed> I'm surprised they couldn't snuff out that fire.
[23:36:17] <flyback> small nuke
[23:36:24] <DagoRed> Right...
[23:36:34] <flyback> why not
[23:36:38] <flyback> deep underground
[23:36:42] <flyback> vaporize a patch
[23:36:45] <flyback> nothing left for the fire
[23:36:53] <DagoRed> Because the fallout and heat would cause more fires.
[23:37:17] <DagoRed> It wouldn't vaporize the patch, plus underground nuke testing has been banned for at least 3 decades.
[23:50:18] <z64555> what about a fuel-air bomb
[23:53:53] <theBear> what about a fuel air mixture, maybe in a straight glass with a fancy straw
[23:57:23] <z64555> just mix in a bit of water to retard the combustion a bit
[23:57:41] <z64555> bar tenders do it all the time
[23:58:22] <theBear> my combustion chamber is already more than retarded enough, i might skip the water