#robotics Logs

Jan 01 2015

01:06 GuShH YOU
01:06 GuShH you too Tom_itx
01:06 ShH fr
01:06 GuShH and maybe armyofevilrobots
01:11 Jak_o_Shadows hi
01:11 Jak_o_Shadows happy new year
01:12 GuShH aww you too
01:12 ShH points out it's 4AM "h
01:12 GuShH slightly drunk, listening to music
01:13 Jak_o_Shadows I'm slightly tired, having had new years17ish hours ago
01:13 GuShH :o
01:13 GuShH y u in da futures?
01:14 Jak_o_Shadows I don't feel like i'm in the fiture. I feel like sleep.
01:15 GuShH better than feeling like sheep
01:15 Jak_o_Shadows True.
01:58 rue_bed happy new years to all time zones I missed so far!
02:03 orlock happy new years
02:03 orlock http://dark.crypt.net.au/Prusai3/20141230_215531.jpg
02:03 orlock no retraction set
02:04 orlock .2mm layers
02:14 GuShH rue_bed: it's been 5 hours!
02:15 rue_bed hah
02:15 rue_bed my judgement lapsed!
02:15 rue_bed its another few for me
02:15 GuShH and I'm actually 50% drunk, but I met an incredible gal
02:15 GuShH orlock: that's nice
02:15 ShH tries to f
02:15 GuShH what would you say caused the banding
02:17 GuShH I can't remember the words... halp
02:17 GuShH backlash?
02:18 GuShH is one of the Y belts slipping
02:19 rue_bed oh lets see
02:20 rue_bed happynewyears
02:20 rue_bed I think someone just set off a pole transformer!
02:21 rue_bed it was loud enough!
02:21 rue_bed happynewyears!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
02:22 rue_bed GuShH, I think its a platic temperature or size inconsistancy
02:23 rue_bed will it jam if you dial down the temp a bit?
02:26 GuShH it could
02:26 GuShH or you can have adhesion problems
02:26 GuShH but dont really take me serious right now I've had 6-7 shots of scotch, 3 beers and whatnot.
02:27 GuShH I can type since I used to drink a lot and I'm still quite used to it, but that doesnt' mean I'm not a bit drunk. I do try to avoid it though.
02:30 rue_bed try a layter height at 0.8x the tip size
02:37 GuShH rue_bed: but when the error isn't uniform it can be slippage or other issues
02:37 GuShH that's why I asked
02:38 rue_bed yea
02:38 rue_bed I think my feeder may have an issue, I plan to do a rebuild on it tho
02:40 GuShH oh is that pic not from orlock
03:00 rue_bed well its a new year
03:00 rue_bed time to make new resolutions
03:00 rue_bed I'm done with trying to find a mate, on to something worth while
04:19 Jak_o_Shadows Mine is to become more flexible
04:27 Mr_Sheesh IDK, Jak_o_Shadows, dissolving your bones sounds sorta dysfunctional...
05:19 rue_more e> happy new year!
05:20 orlock 5 minutes of filing extruder motor housing didnt work
05:20 orlock i will try another 5
05:20 orlock or maybe .5 seconds with dremel
05:20 rue_more why are you filing the housing?
05:20 orlock by housing i mean the bolt holes
05:21 orlock that the motor bolts to
05:21 rue_more your filing the motor?
05:21 orlock mounting holes
05:21 orlock no
05:21 rue_more oh
05:21 orlock the printed carrier holes
05:21 rue_more ooooh:/
05:21 rue_more is the print out!?
05:21 orlock it fits together
05:21 orlock like, it fits well
05:21 orlock the print seems pretty good
05:21 rue_more k
05:21 orlock and these holes are slotted
05:22 orlock but i just need like, .5 of a mil more adjustment
05:22 orlock theres a little bit of slop between the extruder gears
05:22 orlock i just need to move them together that little bit
05:22 orlock take up the last bit of slack
05:23 orlock i just adjusted the hobbed bolt spacing by adding a washer (too much adjustment) and replacing a flanged washer with normal (sat .5mm closer in), now its juust right
05:24 orlock filament hole lines up smack bang with the deepst part of the notches
05:25 orlock going to try using an optical sensor from an inkjet
05:25 orlock 3 wire, hope i kept the loom
05:25 orlock tiny plug
05:25 orlock using that for Z axis endstop
05:26 orlock i have some wide enough that an M3 bolt end will fit, to make precise adjustment easy
05:27 Jak_o_Shadows dude, flexibility is not bones
05:27 Jak_o_Shadows normally
05:28 rue_more a bit of slop between the extruder gears is nothing
05:28 rue_more other things will be your larges nemisis
05:29 rue_more orlock, did you print that thing already tho?
05:50 orlock rue_more: except when it comes to retraction it can actually add up to a small percentage
05:50 rue_more you set the retraction distance
05:51 rue_more its just some value
05:52 orlock yeah, and its often pretty small, and its the only time the extruder reverses that i can think of
05:56 orlock there are metal dual extruders that are not disgustingly expensive
08:35 Veverak hmm
08:35 verak won
08:37 Veverak let's say I create part with intent to use specific way of making it at first (3d print)
08:38 Veverak and want the possibillity to cut it out of metal plate
08:38 Veverak but, problem is that it won't fit on the 3d print desk
08:39 Veverak A. should I made the mechanism in a way where it is made of multiple smaller parts? (small parts by design)
08:39 Veverak B. should I just make one big part, split it specifically for 3d printing into two segments with that libraries with "connect" joints
08:53 Veverak https://cloud.veverak.org/public.php?service=files&t=4b10593034e1ea95d99dae97d3e21a7e here, picture, speaking of that silver part :)
09:08 Tom_itx make the frame from wood
09:08 Tom_itx first article anyway
09:19 Veverak 0so make it from more parts by design?
09:19 Veverak ok
09:19 Veverak why? D:
09:22 Tom_itx because it won't fit under your printer
09:22 Tom_itx poor design criteria
09:23 Veverak ...
09:24 Veverak I know that it won't fit in this state
09:24 Veverak but
09:24 Veverak points, make it from smaller parts by design or print that big thing as multiple parts with 3d prining tools used to that
09:25 Veverak first means that If I would do it from metal/wood after... it would be from small pieces
09:25 Veverak seconds means that it won't have to be
09:54 Tom_itx whichever way, the center needs to be strong. making multiple pieces would weaken it i think
10:08 Veverak Tom_itx: yeah, thinking about making another layer on top of it
10:08 Veverak maybe... 10mm above
10:08 Veverak connected with normal plastic collumns you can buy almost anywhere
10:08 Veverak would made it really strong I suppose
10:10 Tom_itx or print a flange on the halves and screw it together
10:10 Veverak yeah, thought about that
10:11 Veverak http://www.thingiverse.com/thing:35834
10:11 Veverak nah, not screws :/
10:11 Veverak Tom_itx: I will maybe think about it more for a while
10:13 Tom_itx still leaves room for wobble
10:13 Tom_itx ever try to pick up a jig saw puzzle?
10:13 Veverak yeah, so it's bad idea, I realize now :)
10:14 Tom_itx L flange or overlapping joint
10:14 Tom_itx _-
10:14 Tom_itx one on top of the other
10:14 Veverak I udnerstand
11:19 LiohAu anybody has some experience with the Dagu t-rex controller?
11:20 LiohAu I'm controlling it with a raspberry pi using i2c, but after few orders the board does not answers anymore. I don't really know what's happening, but the "write" calls on the i2c device are failing
11:29 deshipu LiohAu: I would ask on the letsmakerobots forums, the author is there
11:31 deshipu LiohAu: make sure you provide all the relevant details
13:27 rue_bed probably malformed i2c
13:28 rue_bed its easy to get the states wrong and lock something up
13:29 rue_bed its generally true that if you have a problmem with a program you wrote, your gonna have to post the source code so people can look for problems in it
13:36 LiohAu rue_bed: program code : http://pastebin.com/J2VW03kv and program output : http://pastebin.com/XF9k4dQW
16:22 ty1234 hey guys
16:23 ty1234 does any one know what microprocer or microcontroller does big dog uses?
16:23 ty1234 *boston dynamics
16:29 ty1234 anyone?
17:10 wolfmanjm anyone used a 5" RA8875 based LCD from buydisplay.com?
17:23 ace4016 wait...wtf?
17:23 ace4016 -> ace4016x (745d8526@gateway/web/cgi-irc/kiwiirc.com/ip.116.93.133.38) Quit
17:24 ace4016 oh...i see
17:24 deshipu what?
17:24 ace4016 i think that was ty1234's attempt at getting attention
17:26 deshipu works every time, eh? ;)
17:27 deshipu but about his question, I'd guess they have a normal PC in there
17:27 ace4016 probably
17:27 deshipu because then they can use all the tools that the mathematicians are used to
17:27 deshipu like Mathematica
17:27 ace4016 well...not really
17:28 Loshki They're a bit shy about the details, which is odd because with darpa funding, I'd have thought it would be fully PD.
17:28 deshipu Mathematica can generate C++ code, IIRC, but it's platform-specific
17:29 deshipu I know that the Atlas robot uses that
17:29 ace4016 they're using a normal PC cpu because it's simple and does the job
17:29 deshipu ace4016: Intel CPUs are anything but simple
17:29 ace4016 they're also probably using some more embedded type MCUs as well for control of the acuators and sensors and such
17:30 deshipu that's for sure
17:30 ace4016 deshipu, from the end user's point of view, with operating systems and applications and environments already up and running...it's simple
17:30 ace4016 if you think otherwise, you'd be a fool :P
17:30 deshipu ace4016: I've build several robots, two of them have operating systems on board, it doesn't make things easier, on the contrary
17:30 deshipu ace4016: you basically get a perambulating datacenter
17:30 deshipu that you have to manage
17:31 ace4016 much simpler than creating a custom OS and frameworks and compilers and so on and so forth
17:32 ace4016 basically, there's a plethora of pre-built tooling
17:32 ace4016 and when you're trying to solve robotics problems, it's nice if you can worry about the unsolved problems, rather than recreating the wheel
17:32 ace4016 also, faster results
17:33 deshipu my point is that you get a lot of problems that are not immediately related to what you are trying to do
17:33 deshipu and I'm not advocating writing your own operating system
17:34 deshipu on the contrary
17:34 ace4016 there's a cost to everything :P
17:35 deshipu I understand that at the scale of complexity of the Big Dog it may be a good choice to go for an operating system, even if you have to do some customizations to it
17:35 ace4016 with anything you select, you inheret the good, the bad, and the ugly (that goes for hardware and software)
17:35 deshipu but at the scale of complexity of hobby projects, I don't think it pays
17:35 ace4016 depends
17:35 ace4016 are you creating a line follower? no, an OS doesn't make much sense
17:36 deshipu for a line follower, a CPU doesn't make sense
17:36 deshipu you just need an opamp :)
17:37 ace4016 creating a robot with multiple tasks, most of which are complex (even at the hobbyist level); you might consider a larger chip and an OS to handle some of the things that come with it
17:37 rue_more use a 555!
17:38 ace4016 and big dog lays in that realm
17:38 deshipu ace4016: do you have anything in particular in mind that would be easier to handle with an operating system?
17:38 deshipu big dog is not in the hobby projects realm
17:38 ace4016 it's not, but it's in the realm where letting an OS handle hardware things is useful
17:39 ace4016 resource allocation and management (CPU time and memory) are the main things
17:39 ace4016 depending on architecture, you might create different applications to handle different aspects of the robot
17:39 deshipu now, that's also not that simple, since big dog is going to have a lot of custom stuff in it, so you need to write all the drivers yourself anyways
17:40 ace4016 that's fine
17:40 deshipu ace4016: well, you don't want dynamic resource allocation in a real time system such as the big dog robot
17:40 ace4016 not all OSes are full blown linux or windows :P
17:40 ace4016 i wasn't talking about dynamic resource allocation
17:40 deshipu ace4016: you don't want the balancing submodule to suddenly run out of memory because the picture recognition module kicked in
17:40 ace4016 there exist RTOSes
17:41 ace4016 oh trust me, i know all about that
17:41 ace4016 i work in the aerospace industry :P
17:41 deshipu well, if it's not dunamic, then the management gets much easier :)
17:41 deshipu dynamic*
17:41 ace4016 it does, somewhat
17:41 ace4016 scheduling becomes more difficult
17:41 ace4016 tit for tat
17:42 ace4016 there's a cost for everything :P
17:42 deshipu do you think it's realistic to attempt a hobby project at a complexity that would actually warrant having an operating system?
17:43 ace4016 yes
17:43 deshipu assuming it's not your day job
17:43 ace4016 you can go for a monolithic application that handles various aspects of your project
17:43 ace4016 but that's one solution to a problem with many solutions
17:44 ace4016 and sometimes you end up creating an OS by accident :P
17:44 deshipu that's the worst possibility
17:45 ace4016 it happens
17:45 ace4016 but some people are more adventurious in thier hobby projects than just "lol quadcopter" or "lol hexabot"
17:46 tws2172 I've made a ROS based robot that uses a tiny desktop computer. It works very well. It also uses SLAM to navigate around rooms on its own. I thin this type of project warents an OS
17:46 tws2172 think*
17:46 ace4016 and even those could benefit from an OS if they do more than just fly/crawl (i haven't seen too many that performed an interesting task autonomously)
17:47 ace4016 not underplaying the complexity of a motion controller, but that's all they generally do. they tend to be R/C or pre-programmed motions
17:47 rue_more the large hexapod I'm making
17:47 rue_more I have 5 atmega32 boards that control the 18 servo loops
17:48 Tom_itx why not mega238?
17:48 Tom_itx 328
17:48 rue_more and I have a linux SBC doing the realtime math and dishing out the position updates
17:48 Tom_itx i thought you did your code for the 328
17:48 rue_more the big hexapod was long before the 328