#robotics Logs
Jun 10 2015
#robotics Calendar
01:34 rue_house that tiem of year eh?
01:35 rue_house you still not sent me the 6mo seperated images of the same star field
01:35 orlock Hmm - actually, i'll be able to do that soon
01:36 orlock just need to drill 3 holes in some concrete
01:36 orlock well, pavers
01:36 orlock http://en.crypt.net.au/Astro/rho-oph_by_ozzzy.png
01:38 orlock rue: i set pavers into the ground, now i need to drill holes to set the legs into to make it all repeatable
01:39 orlock now next clear night i need to test the homebrew autoguider
01:39 orlock software all works
01:39 orlock even put it on github
01:39 orlock well
01:39 orlock "works" as in i can fool both INDI and ASCOM into thinking "Hey! Its an LX200 mount!"
01:40 rue_bed pff, that image is millions of years old
01:41 orlock need to make electric blanket for lense
01:41 orlock rue: what sort of FoV you want the image to have?
01:41 orlock i guess i'll make it as big as i can
01:41 orlock without getting the mountain in shot
01:42 orlock i should figure out what that is anyway
01:43 orlock either 55mm or 135mm i guess, forget what sort of FoV that is
02:10 rue_bed dont care, as long as its the same field with a 6mo difference
02:32 LiohAu JT-Shop: still here?
02:33 LiohAu i'm looking for a way to mirror a spline in a 3D sketch
02:42 mrdata LiohAu, inkscape should do that
02:42 LiohAu yes but i'm in solidworks :d
02:43 mrdata scream bloody murder at them, then
02:43 LiohAu to who?
05:43 veverak hmm
05:43 veverak any tips for DIY linear scales?
05:43 veverak I found this things quite expensive
05:51 veverak nah
05:51 veverak rotary encoders could do fine too
05:51 deshipu how about a linear pot?
05:52 veverak 300mm long?
05:52 veverak :D
05:53 veverak nah
05:53 verak would really prefer closed loop stepper contro
05:54 veverak linear scales are what CNC machines uses to determine postiion
05:54 veverak but that is expensive
05:54 veverak maybe rotary encoder would do just fine
05:54 deshipu why do you need an ecoder with a stepper?
05:54 deshipu I thought that you can just count the steps
05:54 veverak that's ideal condition
05:55 veverak practically just because send 100 steps
05:55 veverak it doesn't mean 100 steps will be done
05:55 deshipu why not?
05:55 veverak like for example when printing head on 3d printer is blocked for a while on z axis
05:56 veverak layers on the 3d model than shifts
05:56 deshipu but wouldn't dropped steps make the stepper motor get out of synch with the controller and stop working?
05:56 veverak why it should? :D
05:56 deshipu it's not like each step is the same
05:56 veverak controller just powers coils
05:56 deshipu you need to change them in order
05:56 veverak that's all
05:56 veverak :D
05:56 deshipu but the position of the rotor wouldn't match the next step you send
05:56 deshipu and it wouldn't rotate
05:56 veverak would
05:57 veverak or it does
05:57 veverak you can argue why, but it definetly does
05:57 veverak just blocks over something, miss few steps and than continues
05:57 deshipu you would miss all the steps until you get back to the one you missed
05:57 veverak not one that you missed
05:57 veverak you got two pair of coils
05:58 deshipu so 4 steps
05:58 veverak yeah, not much
05:58 veverak but, it makes your day bad
05:58 veverak so, anyway, I am researching if I can just make it like any CNC-like machine should do it
05:58 veverak measure where it is :)
05:59 veverak instead of relying on stepper, that all steps that were sent were done
06:00 veverak it's pretty useles for Z axis
06:00 veverak but on X/Y axis it would be nice
06:06 veverak linear scales are pretty expensive.. but rotary encoders could do just fine
06:14 SquirrelCZECH well
06:14 SquirrelCZECH http://hackaday.com/2015/01/20/closed-loop-control-for-3d-printers/
06:14 SquirrelCZECH this is what I am talking about :D
06:31 hypodyne howdy
09:34 JT-Shop LiohAu, I'm here for a bit
09:35 LiohAu_ JT-Shop: nevermind I found how to mirror a spline from a 3d sketch :)
09:35 LiohAu_ a little bit ugly to use a surface for that, but it works ^^
09:35 JT-Shop lol, as long as it worked that's all that counts
09:37 LiohAu_ now I have another issue. My original mesh is half of a sphere (flat on bottom so), and on this 1st part I have 4 "half" of sphere
09:37 LiohAu_ I have no idea of how I can make the spline of these 4 half sphere follow the 1st half sphere
09:39 LiohAu_ basically, I would like 4 half sphere like the green one : http://imgur.com/y9j9dmP (making a cross), and the original half sphere would be in the middle
09:45 Triffid_Hunter LiohAu_: those look like 1/8th spheres
09:45 LiohAu_ they are not even real spheres
09:46 Triffid_Hunter LiohAu_: if you want spheres rather than splines, I'd use openscad to generate the model
09:46 JT-Shop is that a single part or an assembly?
09:47 LiohAu_ a single part
09:48 JT-Shop lol I thought you were talking about splines on a shaft not spline lines
09:48 LiohAu_ JT-Shop: http://imgur.com/R9shbD0
09:49 LiohAu_ this is the original model I'm trying to replicate on SW
09:49 JT-Shop can you import it?
09:49 LiohAu_ the yellow loft is ok. The big red half sphere too. But the 4 "spherical" shit are a nightmare
09:49 LiohAu_ Yes I could with ScanTo3D
09:50 JT-Shop might save you some time to import unless you want to change something then import won't work
09:50 LiohAu_ problem is that I need to change things in SW :(
09:51 LiohAu_ I can't keep the mesh like this
09:51 JT-Shop yea your stuck creating it from scratch
09:52 LiohAu_ yes, but even from scratch, I have no idea of how I can make a spline follow the "big" sphere
09:54 LiohAu_ at the moment the 4 half sphere I made (actually 1 half of the half, mirrored, and then mirrored again 3 times), are not in contact with the "big" one everywhere
09:54 JT-Shop I'd have to ponder it a bit, I normally don't do that type of design work
09:55 LiohAu_ don't know if I was really clear, so look at the picture : http://imgur.com/ZPqONjq
09:56 JT-Shop yea, I see what your talking about
09:56 JT-Shop I'll ponder this while I'm at a customers place and I'll be back after while
09:57 LiohAu_ ok thanks :)
12:42 data is watching http://www.ustream.tv/channel/nasa-msfc<
12:43 mrdata army of angry robots just did their run
13:04 mrdata south hadley is setting up now
13:26 deshipu mrdata: thanks!
13:26 mrdata running!
13:27 deshipu mrdata: what's that?
13:29 mrdata deshipu, south hadley robot is running the course now
13:30 mrdata the task is to find the right white object and pick it up, methinks
13:42 mrdata the robot just turned away from rather than toward the target object
13:42 mrdata running out of time; 2 mnutes to go
13:44 deshipu but what's the contest?
13:45 mrdata i know nearly nothing about it, except it's yearly; and hosted by nasa
13:45 deshipu I see, thanks
13:47 mrdata but on livestream at the link i shared, above
13:55 mrdata next competitor is setting up now
14:01 mrdata goal is sample retrieve. the robot must find, pick up, and return to home
14:01 mrdata this is level 1, involves a single sample to return; tomorrow is multiple samples to return
14:02 mrdata there is an additional requirement to maintain sample integrity; the samples cannot touch each other
14:09 mrdata "NASA Centennial Challenge program's Sample Return Robot Challenge [...] will take place June 8-13, 2015. The autonomous robot competition, which carries a prize purse of $1.5 million, will be held at Worcester Polytechnic Institute in Worcester, Massachusetts, which has hosted the event since 2012."
14:09 mrdata -- http://www.nasa.gov/press/2014/september/nasas-2015-sample-return-robot-challenge-open-for-registration/#.VXiFclKInzI
14:10 mrdata MIT is setting up now
14:10 mrdata http://www.nasa.gov/centennial_challenges/sample_return_robot/meet-the-20-teams-competing-in-the-2015-sample-return-robot-challenge.html
14:22 deshipu mrdata: nice
14:22 deshipu I wonder if at least one of these will have legs ;)
14:27 mrdata idk
14:28 mrdata the MIT team cant run today
14:28 mrdata the bot got soaked in sa torrential downpour during a "dry run" yesterday lol
14:29 mrdata they have a blown motherboard or power supply
14:29 mrdata yet, the competition is supposed to proceed in case of light rain, so competitors are supposed to tolerate moisture
14:29 mrdata they will try to run friday
14:30 mrdata that was today's last team; coverage is concluded until tomorrow
14:31 mrdata follow @nasaprize on twitter for details
14:37 SpeedEvil Is 'light rain' defined?
14:37 deshipu on unrelated news, a friend ran those small SG90 for me through the osciloscope
14:38 deshipu https://hackaday.io/project/6050-tote-affordable-spider-robot/log/18992-proprioception-on-the-cheap (at the end)
14:38 deshipu this is the voltage on the motor for when the servo is going one way, then the other
14:39 deshipu I wonder what is happening on the last picture
15:07 wolfmanjm deshipu: can you tell what is wrong here it is doing the creep gait... https://flic.kr/p/upPjWJ
15:08 wolfmanjm I think I need to reverse the hip joint for the bottom two
16:11 deshipu wolfmanjm: loading...
16:12 deshipu wolfmanjm: hmm, which way is front?
16:13 wolfmanjm the leg on the left is the front left then going counter clockwise left rear etc
16:14 wolfmanjm I also think I need to increase the distance the gait uses I just used the same creep parameters as in you code
16:14 deshipu so it seems the two hind legs have reversed hips
16:14 wolfmanjm yea ok I thought that too....
16:15 deshipu ankles are hard to tell, really
16:15 deshipu I usually have my robot do some shifting od the body without walking
16:15 deshipu to figure out that
16:16 wolfmanjm I think the ankle and knee are ok I tested them with a simple raise command and they seem to all go up and slight in, funnily enough no reversing was needed on any of those joints
16:17 wolfmanjm I couldn't figure out how to make it just shift withot walkign :) it seems to want to walk regardless of what I tell it to do :)
16:17 deshipu well, replace the main loop
16:17 deshipu with your own code...
16:17 deshipu that just does shifting
16:17 deshipu as in, moves all the legs by the same amount in the same direction
16:18 wolfmanjm like just call _shift_body_by() ?
16:19 deshipu wolfmanjm: yeah, that would work too :)
16:19 wolfmanjm ok thanks
16:20 wolfmanjm I mean it looks like it is almost working ;) but it doesn't actually move other than shake around :)
16:21 wolfmanjm it bobs up and down and kinda gyrates but doesn;t actually move in any direction (that was after reversing the two hip joints)
16:23 deshipu :/
16:29 deshipu wolfmanjm: did you change the dimensions of the robot?
16:29 deshipu wolfmanjm: everything else is the same, right?
16:29 wolfmanjm I changed the TIBIA FEMUR COXA and BASE
16:30 wolfmanjm the rest is the same
16:30 shipu scratches his
16:30 deshipu logically thinking, you shouldn't have to reverse any servos
16:30 deshipu or you should reverse all of them
16:31 deshipu at least all of them at a given joint
16:31 wolfmanjm the BASE setting is unclear if that is the diagonal length from the center to the hip joint or the X distance from the cnetrer line to the hips
16:31 deshipu x distance
16:31 wolfmanjm oh ok I got that wrong then
16:31 deshipu I think, let me check
16:31 deshipu but that doesn't affect anything except rotation
16:32 deshipu yup, x distance
16:32 deshipu and y, they are asumed to be the same
16:32 wolfmanjm yea the ankle jpints are all reversed the knee jpints are al not reversed and the two fron hip are normal and the two back hip joints re reversed
16:33 deshipu and the legs are in the same order?
16:33 deshipu starting with front left, etc?
16:34 wolfmanjm not sure what order you have them in but mine are 0 is front left, 1 is back left, 2 is back right, 3 is front right
16:34 deshipu it's the same then
16:34 deshipu there may be one more thing
16:34 wolfmanjm (you may want to specify that in your build document :)
16:35 deshipu the legs will move funny (often in reversed direction) if you try to move them outside their range
16:35 deshipu well, it's specified by the PCB
16:35 deshipu and since you have different dimensions, your ranges are all different too
16:36 deshipu so you might want to try to make the constants in creep.c smaller
16:36 deshipu and see what happens
16:36 deshipu I have to got to bed, good luck
16:40 wolfmanjm ok thank you :)
18:26 veverak my attempt to make dual extrusion, ersion one
18:26 veverak http://i.imgur.com/SepiB7E.png
18:27 veverak this should allow me to move up/down each motor separatedly
18:27 veverak and because I need only "down" and "up" position, servo with "springed pulling" should do just fine
18:27 veverak and on the motors, I will use these extruders http://www.thingiverse.com/make:48057
18:39 wolfmanjm using 1.75mm filament?
21:18 rue_house I got bearings from ebay(?) that I didn't order.... (?)
21:19 rue_house Z629
21:22 wolfmanjm oh those are mine ;)
22:01 rue_shop3 http://www.aliexpress.com/item/Motor-R130-Micro-DC-Motor-3V-13000RPM-Four-Wheel-Drive-Motor-Experiment-Medium-Motor-10PCS-Lot/32331396133.html
22:01 rue_shop3 FOUND THEM
22:04 rue_shop3 http://www.aliexpress.com/item/Round-DV-Motor-for-Toys-DIY-Made-3V-to-6V-High-Speed-Motor-10PCS-Lot/32331150480.html
22:04 rue_shop3 eat your heart out rue