Back
[00:39:07] <rue_shop3> the max+1 of 12 bit is...
[00:39:16] <rue_shop3> 1024, 2048 4069
[00:39:18] <rue_shop3> 96
[00:39:54] <rue_shop3> you know what, having a cnc machine running all day in the background of the shop, is a sound thats gonna drive me crazy
[00:41:27] <rue_shop3> so half of 12 bits is about 2048
[00:51:46] <z64555> signed, unsigned?
[00:51:53] <rue_shop3> unsigned
[00:52:12] <rue_shop3> WHO USES SIGNED NUMBERS THESE DAYS GEEZ, talk about obsolete...
[00:52:30] <z64555> just about everybody, really
[00:53:35] <z64555> not sure what you mean "half of 12 bits" though
[00:56:26] <z64555> do mean the value you can use with 6 bits? or the max 12 bit value divided by 2 (which is 11 bits)
[00:56:42] <rue_shop3> no I need to send 50% pwm to a pwm chip
[00:56:46] <rue_shop3> its 12 bit pwm
[00:58:49] <z64555> ah, my mistake. It actually is 12 bits, just the 12th bit is 1 with the rest 0
[00:59:04] <z64555> which is a value of 2048, so you're correct
[01:04:40] <rue_shop3> gee thanks
[01:04:53] <rue_shop3> an alternate good answer would have been 2047
[01:07:42] <mrdata_> yeah, 2047 in negative logic
[01:08:07] <mrdata_> or -2048 in signed
[01:18:16] <rue_shop3> ok the irq is running
[01:58:08] <rue_shop3> yea, its like the data line is stuck high
[02:13:12] <mrdata_> thats bad
[02:13:22] <rue_shop3> its not stuck
[02:13:27] <mrdata_> ok, good
[02:13:36] <rue_shop3> you know, this circuit worked on a breadboard
[02:13:41] <mrdata_> uh oh
[02:13:52] <rue_shop3> the pwm chips getting hot, but everything seems to be right
[02:14:15] <rue_shop3> maybe I should add some nops to the serial data routine
[02:14:20] <mrdata_> slap heat sink on the pwm chip maybe?
[02:14:32] <rue_shop3> its not driving anything :)
[02:14:40] <mrdata_> blork
[02:14:40] <rue_shop3> its only midnight and I'm tired :/
[02:14:47] <rue_shop3> maybe I should take a nap
[02:14:51] <rue_shop3> ....8 hours or so
[02:15:02] <mrdata_> if it isnt driving anything, then why is it hot
[02:15:18] <rue_shop3> I dont think it likes the 4Mhz input clock
[02:15:30] <mrdata_> hrm
[02:16:00] <mrdata_> whats the partno
[02:20:19] <rue_shop3> TLC5940
[02:20:27] <rue_shop3> slowing down the data didn't do it
[02:20:43] <rue_shop3> I dont think 4Mhz was over its pwm count limit
[02:20:53] <rue_shop3> its an evil chip
[02:21:29] <rue_shop3> 30Mhz
[02:21:36] <rue_shop3> I'm at 4Mhz
[02:24:05] <mrdata_> yeah 4MHz shouldnt hurt it?
[02:25:13] <mrdata_> http://www.ti.com/lit/ds/symlink/tlc59401.pdf
[02:25:24] <rue_shop3> I have a datasheet
[02:38:57] <mrdata_> page 4, Icc (device supply current) max 60mA -- what is your supply voltage?
[02:39:26] <mrdata_> even so, 5V * .06 = 0.3 watt
[02:39:34] <mrdata_> not huge
[02:40:17] <mrdata_> packages are rated 2 to 4 watts
[02:49:52] <rue_shop3> 5V
[02:50:13] <rue_shop3> well, none of the 1k surface mount resistors are 200 ohms :)
[02:50:22] <rue_shop3> (102 vs 201)
[03:05:56] <rue_house> just mapping out what tools I need to have space for in the new shop
[03:06:04] <rue_house> 1200 square feet might not be enough...
[03:08:10] <rue_house> and I'm not even accounting for things I dont have
[03:11:23] <rue_house> Tom_L, on the new cnc I'm making, I'm gonna use a vacuum cleaner motor
[03:11:42] <rue_house> they can do 20k, I wont give it reduction
[03:14:33] <rue_house> guestimate the space:
[03:14:36] <rue_house> lathe
[03:14:39] <rue_house> milling machine
[03:14:43] <rue_house> cnc mill
[03:14:46] <rue_house> cnc router
[03:14:51] <rue_house> 3d printers (2 likeley)
[03:14:59] <rue_house> horiz band saw
[03:15:06] <rue_house> vert band saw
[03:15:08] <rue_house> scroll saw
[03:15:12] <rue_house> chop saw
[03:15:15] <rue_house> drillpress
[03:15:17] <rue_house> table router
[03:15:25] <rue_house> hydraulic press
[03:15:31] <rue_house> disc sander
[03:15:34] <rue_house> bench grinder
[03:15:43] <rue_house> table saw
[03:15:47] <rue_house> sand blaster booth
[03:15:52] <rue_house> THATS a lot of space
[03:16:01] <rue_house> and thats JUST the big tools
[03:16:53] <rue_house> a number of the tools need special environmental consideration
[03:16:59] <rue_house> for throwing sparks
[03:18:54] <rue_house> not including inside welding space
[03:18:58] <rue_house> thats gonna be outside
[03:19:07] <rue_house> hmmmmmmmmmmmmmmmmmmmmmmmmmmmmm
[03:21:26] <rue_shop3> AAAAAAAAAAARRRRRRRRRRRRRRRRRGGGGGGGGGGG I need space for large robots too
[03:26:02] <rue_shop3> ARG compressor
[03:26:09] <rue_shop3> it can go outside, somehow
[03:26:17] <rue_shop3> I'll need protected outdoor space
[03:28:07] <rue_shop3> you know, its important to have an auto-off for machines that take a LONG time to do their job
[03:30:38] <Jak_o_Shadows> You can stack hte 3D printers
[03:32:29] <rue_shop3> hahah, I suppose I can
[03:40:08] <rue_shop3> so, wait
[03:40:33] <rue_shop3> at 7.2V, a 4800KV motor can do...34560rpm!?
[04:09:08] <mrdata_> I need space for giant robots! like one of these...
https://www.youtube.com/watch?v=YVXJP8LuFsw
[04:30:40] <rue_bed> its amazing there are that many videos already
[04:34:53] <mrdata_> oops, that one was edited to remove the first few seconds, which were important. this one reveals:
https://www.youtube.com/watch?v=Q__bSi5rBlw
[04:39:22] <rue_bed> saw it
[04:40:12] <rue_bed> since midniht, 2 hours went by and I didn't even notice
[04:40:23] <rue_bed> it should be like 12:30 now
[06:12:30] <Celrenheit> Hi everyone
[06:26:41] <robopal> hi
[06:28:32] <Celrenheit> I was wondering if anyone has built or seen a delta robot on a moving vehicle ?
[06:48:31] <robopal> on a moving vehicle?
[06:48:39] <robopal> what do you have in mind?
[07:00:07] <Celrenheit> robopal: a moving car to grip/catch objects while moving
[07:00:59] <robopal> yes, I get that, do you have thought of a specific application?
[07:01:27] <Celrenheit> These are lightweight objects but their are many of them, this is why I thought about a delta
[07:03:08] <robopal> what objects?
[07:04:00] <Celrenheit> The goal is to collect small flowers
[07:04:47] <robopal> for example a robot driving on a crocus plantation
[07:05:04] <robopal> https://en.wikipedia.org/wiki/Crocus
[07:05:33] <Celrenheit> robopal: Yes something like that
[07:06:56] <robopal> well you could do it I suppose but you need to have an arm that exceeds the vehicle and supports the delta configuration, kinda beats the purpose
[07:07:20] <robopal> just have a typical arm with a claw
[07:07:39] <robopal> hmm unless!!!
[07:07:48] <robopal> I just thought of something let me draw it
[07:07:57] <Celrenheit> robopal: Ok
[07:09:25] <robopal> http://royhoo521.en.ecplaza.net/3.jpg
[07:09:45] <robopal> something like that with the delta mechanism in the center, much smaller scale of course
[07:11:41] <Celrenheit> robopal: Yes this is exactly what I was thinking of
[07:12:34] <robopal> Celrenheit, just found one :-(
[07:12:41] <robopal> the million dollar idea is taken!
[07:12:42] <robopal> :P
[07:12:57] <robopal> http://www.ubergizmo.com/2015/11/bonirob-weed-robot/
[07:14:24] <robopal> well yes, that is the future for some plantations that are still very labour demanding
[07:14:35] <robopal> like the crocus flower I said earlier
[07:14:38] <Celrenheit> Oh shit I really miss those millions :P
[07:17:50] <robopal> http://www.ubergizmo.com/2016/11/nokia-smartphone-return-2017/
[07:18:00] <robopal> WOW, now that's news
[07:18:17] <robopal> so nokia still exists? >_>
[07:21:33] <malcom2073_> Microsoft is trying to flog that horse
[07:25:34] <robopal> http://robohub.org/harvey-a-working-robot-for-container-crops/
[07:25:42] <robopal> poor robots, working so hard :D
[07:29:23] <robopal> https://www.youtube.com/watch?v=HIpelnM1NBE
[07:32:15] <Celrenheit> robopal: Which keywords did you use to find the first one (Bonirob) ? Before asking here, I searched a bit around delta but found nothing
[07:32:32] <robopal> hmm
[07:36:11] <robopal> not sure exactly but now that we know, you can search for "crop farming delta robot" see for images, I now see a lot
[07:36:51] <Celrenheit> Alright
[13:47:02] <veverak> 0hmm
[13:47:06] <veverak> question is
[13:47:12] <veverak> how to properly sync job over i2c
[13:47:20] <veverak> when slave needs to disable interrupts for the task ;)
[13:49:25] <SpeedEvil> have a counter running
[13:50:35] <z64555> or
[13:50:42] <z64555> Have the task in question within the ISR
[13:51:34] <z64555> It'll ignore any other interrupts while it's within an ISR, and immediately service another interrupt request when it's done
[13:52:05] <z64555> well...
[13:52:33] <z64555> I might be mis-remembering
[13:53:54] <SpeedEvil> Or use a multicore CPU, or ....
[13:53:58] <SpeedEvil> what processor?
[13:54:04] <veverak> arduino
[13:54:05] <veverak> ;)
[13:54:11] <veverak> esp on other side
[13:54:51] <SpeedEvil> use one of the timers, set it to a known value using I2C or whatever.
[13:55:57] <blib> metal chain vs neoprene belt for gearing?
[13:57:00] <z64555> blib: is timing or rotation precision needed?
[13:57:21] <SpeedEvil> Belts can transfer a hell of a lot of power, precisely and reliably done right
[13:57:35] <blib> z64555: yes
[13:58:23] <z64555> chain is the easy route, belts can slip if the torque is too high
[13:58:57] <blib> any good encoder recommendations ? upto 600rpm speed
[13:59:15] <SpeedEvil> z64555: chains can snap
[13:59:43] <SpeedEvil> Oh neat.
[13:59:44] <SpeedEvil> http://beta.ivc.no/wiki/index.php/Electric_Hoverboard_Monorover_R2_Teardown
[13:59:47] <z64555> so can belts
[13:59:54] <SpeedEvil> Teardown of the hoverboards I have 9 of.
[14:01:29] <z64555> Although, there is such a thing as timing belts
[14:01:54] <blib> that material is for timing belts
[14:02:51] <z64555> eh, is there specs on how much torque it can handle vs/ its width and thickness?
[14:02:54] <blib> http://www.robotshop.com/en/lynxmotion-qme-01-quadrature-encoder.html - cheaper version of this?
[14:03:06] <blib> z64555: not the one I am looking at - will have to try it out?
[14:06:15] <SpeedEvil> There are also power transmission belts.
[14:06:21] <SpeedEvil> The decent ones are really quite nice.
[14:06:35] <SpeedEvil> Carbon fibre core, for the tension element, for example.
[14:06:59] <SpeedEvil> http://www.gates.com/products/industrial/industrial-belts/synchronous-belts/poly-chain-gt-carbon-belts
[14:07:22] <z64555> that might be more than what's needed
[14:07:28] <SpeedEvil> sure
[14:07:35] <z64555> blib: this is to transfer power from the motor to the wheels, right?
[14:07:49] <blib> yes
[14:08:30] <z64555> ok, so figure out how much torque the wheels will start to slip
[14:10:20] <z64555> belt/chain should be strong enough to handle that
[14:11:07] * SpeedEvil has forgotten again what blib is doing.
[14:11:44] <z64555> he's got a differential drive robot. He's designing a control system that'll make it path to a point on a work surface
[14:11:54] <SpeedEvil> ah
[14:11:58] <blib> SpeedEvil: I am trying to make a robot with wheels
[14:12:12] * SpeedEvil is trying to work out how to use the above motors/... easily.
[14:12:42] <SpeedEvil> A four wheeled box, with the front two on a free piviot, and steering done differentially to make it piviot is looking interesting
[14:13:23] <blib> SpeedEvil: what are you making?
[14:13:37] <SpeedEvil> generalised robot base.
[14:13:37] <z64555> Sounds like something unstable
[14:13:43] <SpeedEvil> z64555: how?
[14:13:50] <z64555> unless the differential drive wheels are leading
[14:14:07] <SpeedEvil> z64555: Differntial used in the loose sense, not an actual differential
[14:14:12] <blib> SpeedEvil: are you thinking of using the hoverboard motors? I gave up on them, because I don't want to goto 36v
[14:14:23] <blib> SpeedEvil: are you thinking of using encoders?
[14:14:32] <SpeedEvil> blib: Well, I have the 36V motors, drivers, and encoders that are already in the wheels, so...
[14:14:51] <z64555> SpeedEvil: 2x4, 4x4?
[14:15:00] <blib> SpeedEvil: so the encoders are already in those hoverboard wheels? how good are they?
[14:15:02] <SpeedEvil> z64555: 4*4
[14:15:07] <z64555> interesting
[14:15:29] <SpeedEvil> z64555: the front two wheels are on a pivioting 'axle' (not really) and differentially drive in order to steer.
[14:16:11] <SpeedEvil> blib: I have not as yet measured them - but in order to not vibrate a lot in their intended use, they must be reasonable
[14:16:28] <blib> SpeedEvil: which battery are you planning to use?
[14:16:35] <SpeedEvil> The ones that came with them
[14:17:39] <SpeedEvil> I need to properly investigate the batteries to investigate if they do in fact have internal balancing, because if not, that would be bad
[14:21:52] <blib> SpeedEvil: which one did you buy?
[14:26:41] <blib> SpeedEvil: so you are trying to hack the controlboard?
[14:28:20] <blib> z64555: any recommendations for a good encoder which does not cost too much?
[14:28:30] <z64555> nope, sorry
[14:30:08] <z64555> If your motors are BLDC or steppars, you don't need an encoder
[14:30:19] <blib> http://www.kr4.us/rotary-encoder-1024-pr-quadrature.html - this looks nice, but expensive
[14:30:35] <blib> z64555: I was thinking of using a turnigy
[14:30:47] <blib> z64555: bldc - how do I get accurate encoding information from it?
[14:31:22] <Tom_L> get one from a copier etc
[14:31:39] <z64555> from the controller. There may be a positional controller instead of the usual ESC's, which are speed controllers
[14:31:48] <z64555> *positional controller for sale
[14:32:35] <blib> z64555: link?
[14:34:01] <z64555> Here's a paper on the concept:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3231115/
[14:34:15] <z64555> I am unaware of any commercial BLDC positional controllers for sale
[14:34:23] <Tom_L> blib what's this for?
[14:34:38] <Tom_L> http://tom-itx.no-ip.biz:81/~webpage/cnc/new_pulleys/spindle_enc1.jpg
[14:34:40] <blib> Tom_L: Just a ground 4 wheel rc robot
[14:34:45] <Tom_L> spindle encoder for my mill
[14:34:50] <Tom_L> surplus
[14:36:20] <blib> Tom_L: how accurate can one do bldc encoding information using hall effect?
[14:36:40] <Tom_L> you're gonna get wheel slip anyway
[14:36:53] <Tom_L> encoders are only part of the solution
[14:37:56] <Tom_L> http://www.ti.com/lit/an/sprabq6/sprabq6.pdf
[14:38:03] <Tom_L> read
[14:38:24] <z64555> You can get pretty accurate results on ground vehicles with dead reckoning, but you still need a reference point like a beacon or painted line
[14:39:02] <z64555> and it looks like I confused blib for branjb
[14:39:05] <z64555> welp
[14:40:39] <Tom_L> blib did you get that?
[14:41:31] <blib> Tom_L: yes. Looks complicated :)
[14:42:13] <blib> Tom_L: so back emf helps figure out encoder info?
[14:42:50] <Tom_L> http://robotsforroboticists.com/controlling-brushless-dc-motor-no-encoders/
[14:43:52] <z64555> not encoder info, but the position of the rotor
[14:44:32] <z64555> back-emf sensing is used in ESC's to detect speed of the rotor, as well as determine if the rotor is stalled
[14:45:04] <blib> Tom_L:
https://solenerotech1.wordpress.com/2013/09/09/tutorialhow-to-control-a-brushless-motor-with-raspberry-pi/ - in a circuit like this - how does one get position?
[14:46:03] <z64555> With back-emf sensing, you won't be able to determine the initial position. But you can determine the delta to initial from then on
[14:46:46] <blib> z64555: what hardware do I need? how do you modify that circuit to get back emf?
[14:47:15] <z64555> For a wheeled vehicle, the initial position of a drive wheel is irrelavent
[14:47:43] <blib> z64555: I just need delta
[14:48:52] <z64555> You'll need to be able to access the ESC's controller chip to get that information
[14:49:08] <z64555> off-the-shelf ESC's are not made for this
[14:49:37] <z64555> Some you can hack and solder on a pair of wires to access it via I2C
[14:50:47] <z64555> the controller chips do the back-emf sensing already
[14:51:12] <blib> z64555: are there other ways of getting precise position?
[14:51:29] <z64555> use an encoder, or use a stepper motor
[14:52:15] <Tom_L> servo
[14:52:49] <z64555> yeah, you could use a servo do, but you'd have to hack it to get position info from it
[14:52:59] <z64555> s.do.too
[14:53:23] <z64555> hm, no regex string substitution for me, I guess
[14:53:58] <Tom_L> i meant a 'real' servo
[14:54:37] <blib> servo is better than brushless for this application?
[14:55:05] <blib> price might be an issue with servos? 600rpm? good torque
[14:58:00] <blib> servos with these specs are more expensive than buying a great encoder?
[15:01:28] <blib> Tom_L: seems like servos are very expensive at that range?
[15:05:35] <SpeedEvil> The above devices have motors with three and five pin plugs
[15:05:47] <SpeedEvil> the three pin plug is simply motor wires for a brushless three phase
[15:06:00] <SpeedEvil> the five pin plug has red, black, and three other wires.
[15:06:38] <SpeedEvil> I would be moderately surprised if that is not ~200 steps per rev quadrature on two of the wires, and a index pulse.
[15:10:46] <SpeedEvil> Googling indicates I'm probably wrong, byut haven't investigated further
[15:13:18] <SpeedEvil> blib: also, the above indicates the battery has integral protecitons
[15:13:22] <SpeedEvil> so balances
[15:38:50] <blib> SpeedEvil: some of these have 3 + 5 wires
[15:39:14] <SpeedEvil> yes, as I said above
[15:40:09] <blib> SpeedEvil: where did you get your hoverboard?
[15:40:26] <SpeedEvil> ebay, guy selling a pile of broken ones
[15:41:04] <SpeedEvil> http://www.ebay.co.uk/sch/i.html?_from=R40&_sacat=0&_nkw=hoverboard&_dcat=47349&rt=nc&LH_ItemCondition=7000&_trksid=p2045573.m1684
[15:43:08] <peeps[lappy]> http://www.dailymail.co.uk/video/news/video-1003947/Hoveround-commercial-claims-wheel-Grand-Canyon.html
[16:29:55] <blib> SpeedEvil: thanks
[16:39:31] <blib> SpeedEvil: you there?
[16:52:09] <SpeedEvil> yes
[16:53:57] <blib> SpeedEvil: those 5 pins, any luck finding out what they report out?
[16:54:00] <blib> from the wheel?
[16:54:12] <blib> also, what is the diameter of the D shaft on the wheel? I don't have one, so can't measure it
[16:54:25] <blib> SpeedEvil: I just ordered a used one :)
[16:55:12] <SpeedEvil> about 15mm from memory
[16:55:24] <SpeedEvil> And no, I haven't done adequate research
[16:56:08] <blib> what are you planning to use to hold it for your platform?
[16:56:17] <blib> SpeedEvil: are you planning to use the entire metal enclosure?
[16:57:56] <SpeedEvil> ignore the metal bit probably
[16:58:13] <SpeedEvil> Or maybe lock it.
[16:58:20] <SpeedEvil> haven't looked at the details
[20:30:13] <rue_shop4> "And finally, if you have any concerns, please keep in mind the world in ending, and keep them to yourself."
[20:42:03] <z64555> sounds like something from Monty Python
[21:13:50] <pokmo> AFAIK, servos typically don't turn more than 360 degs. but how come the nema 17 can?
[21:14:10] <rue_shop4> a hobby servo is a system
[21:14:16] <rue_shop4> a stepper is just a motor
[21:14:18] <rue_shop4> not a system
[21:14:52] <rue_shop4> the hobby servo is limited to about 300 degrees because it uses a 'volume control' to sense the postiion, and that limits the turningness
[21:15:42] <pokmo> right
[21:19:30] <mun> anyone know of a cheap stepper that has at least 800g-cm torque?
[21:21:19] <rue_shop4> people keep comming in here asking for mtoors of specific torque and speed...
[21:22:10] <rue_shop4> mun, you can get steppers from multifunction printers. Hospitals get rid of big ones, go to the hospital and ask to talk to the computer dept, ask the geeks for an old printer for motors and parts
[21:22:16] <rue_shop4> if your nice, you'll win
[21:22:29] <rue_shop4> along with about 8 optical interrupters
[21:22:33] <rue_shop4> and stepper drivers
[21:22:52] <rue_shop4> you will have about 4 cubic feet of plastic to get rid of when done
[21:23:00] <rue_shop4> and a bunch of sheet metal
[21:28:23] <mun> rue_shop4, i might need 20-30, so i was wondering if there's any cheap, decent ones out there
[21:28:57] <z64555> hit up multiple hospitals, I guess
[21:29:04] <rue_shop4> thats going to take a lot of hospitals and printers, get a truck and a screwgun
[21:29:20] <rue_shop4> :)
[21:29:57] <rue_shop4> tell me what torque the motors turn out to be, I dont even know a ballpark on any motor
[21:30:33] <mun> i looked into 28BYJ - they're cheap, but they're very low torque
[21:31:36] <Jak_o_Shadows> 800 g-cm is only 0.0784 Nm
[21:31:40] <Jak_o_Shadows> which isn't a wohle heap
[21:33:08] <mun> i know
[21:33:33] <mun> the 24byj can only do 300g-cm
[21:33:44] <mun> pull in
[21:34:47] <rue_shop4> your talking holding torque?
[21:36:45] <mun> rue_shop4, well, pull-in torque
[21:59:01] <rue_shop4> hmm which way is pull in again?
[21:59:35] <rue_shop4> approximite normal speed?
[21:59:44] <rue_shop4> as it 'just some' lost steps?
[22:01:35] <z64555> pull-in, as opposed to pay-out
[22:02:28] <z64555> I think
[22:02:40] * z64555 pokes the mun
[22:05:56] <mun> z64555, hi
[22:06:03] <mun> sorry, went off to get some food
[22:06:16] <z64555> rue had asked a question
[22:06:20] * z64555 hides
[22:07:02] <mun> i'm trying to make sure if i got the terminology correct...
[22:08:25] <mun> i think i do. basically i'm trying to make sure the motor can pull up a 800g load at 1cm. so that should be pull-in
[22:09:51] <mun> rue_shop4, as with speed, hopefully at least 60revs/min
[22:18:57] <Jak_o_Shadows> Why not get a stronger motor and gear it up?
[22:25:18] <robotic> Hey, I'm trying to compute the kernel of the Jacobian matrix for a UR5 manipulator when the third joint causes a sing(when the 2nd, 3rd, and 4th joints all have twists that are parallel and colinear). I would like to do this in terms of the modified twists of the spatial Jacobian, using some knowledge of the structure of the Jacobian when this sin
[22:25:18] <robotic> gularity is occuring. Could anyone give me a hint/point me in the right direction?
[22:26:28] <robotic> I was able to calculate the Range of the Jacobian by observing that the 2nd, 3rd, and 4th joints had twists which were linearly dependent, but I can't find any such insight for the kernel
[22:32:36] <z64555> hm, not sure where our regular ME's may be, they should be able to help somewhat
[22:33:21] * DagoRed hasn't touched that stuff in years.
[22:34:20] <Jak_o_Shadows> Yeah, bit over my head
[22:34:39] <robotic> haha no worries, it's over my head too
[22:35:07] <DagoRed> robotic: ME? EM? AerE?
[22:35:08] <z64555> closest thing I've gotten to something like this is to use a train of rotation matrices
[22:35:27] <robotic> DagoRed: I'm an undergrad ME but this is a grad course in robotics
[22:35:37] <DagoRed> Understood.
[22:35:42] <Jak_o_Shadows> Yeah, I haven't gotten further than that z64555
[22:36:41] <z64555> singularity does remind me of differentiation, though
[22:37:49] <z64555> and using something like Newton's method
[22:38:22] <DagoRed> The only trick I have to answer the question makes the math messier before it gets better.
[22:38:32] <DagoRed> Quaterions!!
[22:38:47] <z64555> ...somehow I knew that was what you were going to suggest
[22:38:47] <robotic> Haha, yeah that's definitely not the approach my professor is looking for
[22:39:15] <DagoRed> robotic: You're right, because if you did use them... your professor would look stupid.
[22:39:15] <z64555> matrices are preferred for computers
[22:39:31] <z64555> where quaternions are perferred for hand calculation
[22:40:07] <DagoRed> .... this is a matrix... it just adds another dimension which allows you to have another degree of freedom for singularities.
[22:40:53] <z64555> in 3D graphics, I've seen a system use a 4x4 rotation matrix
[22:41:04] <robotic> so that's not just a rotation matrix but also translation
[22:41:09] <z64555> yes
[22:41:09] <robotic> called an Affine transformation
[22:41:59] <z64555> a 3x3 sub-block would be the rotation, and a 4x vector would be the translation
[22:42:14] <robotic> right, well 3 vector but yes
[22:42:30] <robotic> and the rest is just for structure so you can multiply them just like rotation matrices
[22:42:39] <z64555> yes
[22:43:42] <z64555> So we'd have an affine xform matrix per joint
[22:44:25] <z64555> and multiply the position of the first joint by its matrix representing its orientation (and maybe also translation)
[22:44:33] <Jak_o_Shadows> I've never used a 4x4. I really should next time I do stuff like this
[22:45:00] <z64555> then multiply another matrix for the next joint's orient and position relative to the first
[22:45:15] <z64555> and so on and so forth down the structure
[22:45:50] * DagoRed is doing HPC stuff right now.
[22:46:23] <DagoRed> I couldn't score a job doing embedded work... but someone got a job that occasionally playes with the fastest super computer in the US
[22:46:45] <z64555> don't tell me ENIAC is still kicking. :P
[22:47:10] <robotic> sort of, the most common and best way IMO is sort of a cool trick where you place the robot arm or system in a certain initial configuration, and then by writing down a bunch of twists (which are just rotations and translations about axes, like a screw motion about an axis) which describe how they move everything down the line from that specific jo
[22:47:10] <robotic> int, we can compose the whole system without really having to write down any of the transformations between joints
[22:47:16] <robotic> it's called product of exponentials
[22:47:40] <DagoRed> z64555: FUCK the ENIAC. No... I'm talking about Titan.
[22:48:09] <robotic> DagoRed: there should be some way to think about the linear dependence of all of the twists and use that to come up with a span of joint velocities which satisfy the linear dependence, but can't seem to work it through
[22:48:10] <DagoRed> robotic: that's cool
[22:48:50] <DagoRed> and yeah... my mind is shot and I'm about to head to bed.
[22:49:03] <robotic> haha no worries
[22:49:11] <DagoRed> robotic: is it a robotic arm with 4 axial joints spinning at the same rate or something?
[22:49:53] <robotic> not really, it's like a 6 axis revolute arm but in this singularity, three joints in the middle are all parallel and in the same plane
[22:50:18] <z64555> oh
[22:50:41] <DagoRed> that doesn't sound too terrible then... but still a mind fuck to get it right.
[22:50:44] <z64555> singularity occurs when the 3rd joint is 180 degrees
[22:50:58] <z64555> er, no
[22:51:18] <robotic> yeah that's right, depending on how you choose the angle conventions
[22:51:52] <robotic> but the question is really: what joint velocities in this situation won't give the end effector a velocity?
[22:51:57] <robotic> at least instantaneously
[22:54:42] <DagoRed> That explains the need for the jacobian. You're using it to do the derivatives of all position equations in the matrix.
[22:55:11] <DagoRed> robotic: are you familiar with state space models?
[22:55:29] <z64555> oh, now that I actually know something about
[22:55:32] <robotic> DagoRed: Yeah
[22:56:43] <z64555> You can draw a state space model which represents the influence of a joint on another
[22:57:10] <DagoRed> I'm thinking if you do something with a state space model where you only care about position and realitive angular velocities of each section may allow you to simplify what you're trying accomplish. From there you need to only state your range in time to let you figure out how to meet your conditionis of a plane.
[22:57:19] <DagoRed> z64555: exactly.
[22:57:35] <z64555> and whats more, If you provide paths to and from a joint
[22:57:56] <DagoRed> Yup.
[22:58:10] <z64555> You can do some fancy math to calculate a single path from a joint to a joint of interest
[22:58:18] <DagoRed> Also sets you up nicely for control algorithms later on.
[22:58:43] <z64555> state space is quite useful for control algorithms
[22:58:50] <DagoRed> robotic: Granted... I doubt this path is the one your professor is expecting.
[22:59:02] <robotic> DagoRed: haha yeah
[22:59:10] <DagoRed> z64555: for MISO and MIMO systems it is practically the only method.
[23:00:04] <z64555> oh boy, THOSE
[23:00:33] <z64555> I had the misfortune of handling a block diagram for one
[23:00:56] <DagoRed> That's ok... at work I'm doing some work to compare some mosaic method of sampling ocean currents and seeing if I can use an unscented kalman filter to help figure out the fluid dynamics equations with less computational resources needed in addition to using the implementation for increasing our weather model resolution.
[23:01:39] <z64555> Well, that's why you're getting paid. lol
[23:01:50] <DagoRed> z64555: I was writing software for a quad copter from scratch for a company I was working for. Needless to say, I didn't finish that in 3 months.
[23:02:05] <DagoRed> I'm getting paid worse at my current job than my last job :/
[23:02:12] <z64555> boo.
[23:02:14] <DagoRed> Er... less.
[23:02:37] <DagoRed> Yeah... boo is right. New college hires are making more money than me. THAT is a kick to the balls.
[23:02:53] <DagoRed> Plus side, I'm exclusively running linux all day everyday now.
[23:03:23] <DagoRed> And... working on campus at Princeton is pretty awesome.
[23:05:35] <DagoRed> .HaD 3
[23:05:39] <makepi> DagoRed: 0.) Commodore Home – Your Smart Home For 1983 -
http://j.mp/2gzIQeh
[23:05:40] <makepi> DagoRed: 1.) Hackaday Links: November 20, 2016 -
http://j.mp/2gbRaxo
[23:05:41] <makepi> DagoRed: 2.) MIDI Guitar Pedals -
http://j.mp/2gzHjF8
[23:05:58] <z64555> omg, the Commodore Home article still exists
[23:06:19] <DagoRed> apparently
[23:06:35] <DagoRed> Who or what is Guest93703?
[23:06:46] <z64555> My dad managed to pick up 2 c128's and a c-64
[23:06:51] <DagoRed> I keep seeing Kitteh in all by 2 channels I'm in.
[23:06:56] <DagoRed> Creeping me out!
[23:07:10] <z64555> I have the c-64 and at least 1 c128, along with some peripherals
[23:07:26] <DagoRed> I just have about 50 arm boards.
[23:08:29] <DagoRed> I want to know why arstech always brings my bought to it's knee's
[23:08:54] <DagoRed> .weather 08540
[23:08:55] <makepi> DagoRed: Princeton, NJ: Overcast, 39.6 F (4.2 C), 29.63(0), Feels Like: 34 F (1 C), Humidity 45%, WindFrom the NW at 7.6 MPH Gusting to 9.8 MPH - Last Updated on November 20, 11:38 PM EST-
http://j.mp/1WIrOrW
[23:09:17] <z64555> bought? you mean the makepi bot?
[23:10:07] <DagoRed> bot, yes
[23:12:17] <DagoRed> .arstech
[23:12:24] <makepi> DagoRed: 0.) Obama says he can’t pardon Snowden -
http://j.mp/2gbWeSv
[23:12:25] <DagoRed> This story makes me sad.