#robotics Logs

Nov 09 2015

06:22 knob Might be an interesting "local navigation" option
07:22 LiohAu knob: so basically he found a way to use manyears without the 1K$ sound card required to connect 8 synchronized microphones ?
08:48 knob LiohAu, I... not sure man. Just throwing that out there. I found it interesting.
08:48 knob Googling manyears now...
08:48 LiohAu well this is a project from a canadian university
08:48 LiohAu they made hardware + software
08:49 LiohAu because they need 8 synchronized micrphones
08:49 LiohAu I have been interested in their project, but can't find a cheap soundcard with 8 inputs...
08:50 LiohAu and they sell 1k$ the one they made (and looking at the number of components, they probably don't make a benefit ^^)
08:50 knob manyEars... haha... I thought it was "manYears"
08:51 knob Found this on them: http://sourceforge.net/p/manyears/wiki/Main_Page/
17:37 veverak deshipu: where did you get buttons for your keyboard?
17:37 veverak let's say I would like 4x4 matrix of proper keys
17:37 veverak :)
17:37 veverak but add mousewheel instead of one position
17:58 Anniepoo a thing I've noticed a few times and thought would be a great tool for a robotic or process type application
17:59 Anniepoo is a keyboard made for flight sim enthusiasts - it's got a surface and individual keys you can stick on it with adhesive
18:00 veverak http://gallery.techarena.in/data/513/Special_Gaming_Keyboards_Images.jpg
18:00 veverak I was just wondering how hard it would be to DIY something like this
18:01 Anniepoo pretty trivial.
18:02 Anniepoo take apart a cheap ordinary PC-104 keyboard, you have all the electronics.
18:04 Anniepoo much of the mechanical could be made from caprolactone
18:06 veverak well
18:06 veverak sparkfun got neat joystick https://www.sparkfun.com/products/9426
18:06 veverak wheel could be get from mouse
18:09 Anniepoo java.Joystick is a really simple way to get the data off such devices
18:09 Anniepoo I've struggles with RSI, and have built various adaptive systems for myself
18:09 Anniepoo mostly using these kind of game input devices
18:09 Anniepoo java.Joystick + java.Robot was a good way to do it
18:13 verak fucking hates
18:13 veverak ;)
18:22 Anniepoo 8cD yah, me too, but it worked for that
18:23 Anniepoo I try not to confuse religion with engineering. If Java or Javascript or VBA or something is the best tool for the job, I use it
18:24 veverak yeah
18:24 verak can't really justify that for DIY robotics th
18:24 veverak see no point to swithc from soemthing liek python
18:24 veverak P.S: pygame is friend for this kind of stuff
18:26 Anniepoo I discovered a nice set of low level interfaces for this kind of bodging thing together - Snap!, the 'blocks' language, has a bunch of real world controllers for it that are all talked to via http - so you can be anything on the other end
18:27 Anniepoo and if you think Java's heavy and ugly, try ROS =8cO
18:29 verak got ROS in
18:29 veverak :)
18:29 veverak as Robot Operating System?
18:29 Anniepoo yup
18:29 veverak next project is going to be ROS based
18:29 veverak but not sure if I want python/C++ nodes
18:30 verak is skilled in python, but more or less new to
18:35 veverak Anniepoo: it looks like it could be worth it, if it's done properly
18:44 Anniepoo if what's done properly?
18:51 veverak proper usage of ROS?
18:52 Anniepoo well, I didn't say it was useless, I said it's heavy and ugly
18:53 Anniepoo a logging truck is heavy and ugly, and not what you want for running to the grocery store
18:53 Anniepoo but it's just the thing if you need to haul 5 million board feet of raw timber out of the Blue Mountains
18:54 Anniepoo the question is, are you going to be paid back for the hell of getting ROS installed and running?
18:54 Anniepoo also, you're wedding yourself to ROS in a way that could kill you
18:55 veverak well
18:56 verak tried to write his own thing for robot con
18:56 veverak and eventually, I've got everything EXCEPT
18:56 veverak entire app being splitted in multiple separated apps
18:56 veverak and than I found ROS...
18:56 veverak with things such like running 'simualtion of robot' alongside real usage
18:56 veverak etc etc etc, many fancy things ;)
18:57 veverak so now I am at getting my stuf, splitting it into ROS thing
19:04 Anniepoo might be worth it then...
19:09 veverak it's overkill for the project
19:09 veverak but it's just assumption that learning ROS will allow me to do bigger projects
19:09 veverak :)
19:43 niepoo would prefer a ROS like that didn't require drinking so much kool
19:44 niepoo waves at recent immigrants from ##prolog
19:44 niepoo reminds dmiles of the 20% rule r
19:45 dmiles_afk hehe
19:46 Anniepoo swipl would be a great candidate for a ROS alternative
19:47 Anniepoo since ROS's problem is just 'heavy'
19:49 zar001 thought that rvirding was a member of this hcha
19:49 eazar001 that's the suspciion he was trying to confirm, but his wild guess was wrong =[
19:50 eazar001 well you can't win em' all =P
19:50 eazar001 https://www.youtube.com/watch?v=HdeYwObD-j4
19:51 dmiles_afk ROS also though promises to run the whole OS for you i tthought.. meaning the whole linux isntall is about having lots of batch files
19:52 dmiles_afk i dont mean promises but is supposed to not be heavier than a single install of linux
19:52 dmiles_afk (or lighter)
19:52 Anniepoo well, it's atop linux
19:53 dmiles_afk becasue of this it cant really buckle under the weigth of its tech stack
19:54 Anniepoo unlike hmmppph... mummmph...grrnnph...
19:54 Anniepoo it seems to be headed down that same sinful path
19:55 dmiles_afk ok the 20%.. its hard not to.. one mmeting at work we needed to design a web UI for some peoiple
19:55 dmiles_afk oh no.. it was we needed tpo play vioeos
19:55 dmiles_afk so i suggested installing appache on the robot
19:55 dmiles_afk gawd.. no .. that was a big.. omg .. we cant have yet another dependancy
19:56 dmiles_afk yet in contract to all the other instane depancy hells
19:56 Anniepoo lol
19:57 dmiles_afk like a 1.5 gb .jar files
19:57 Anniepoo so instead you wrote a whole RTSP server?
19:57 Anniepoo LOL
19:57 dmiles_afk yea which required to mirror mvnrepository.com on disk
19:58 Anniepoo LOL
20:01 dmiles_afk we have to face the fact everything is going to be a kitchen sink
20:01 dmiles_afk OpenCog, ROS etc
20:02 dmiles_afk LogicMOO, Cogbot, CYC
20:02 dmiles_afk may as well expand to the size of an OS
20:02 Anniepoo well, *we* don't, but *dmiles* is a diff story
20:02 dmiles_afk sucks slightly
20:03 dmiles_afk for isntance there is two really good ROS projects going on .. one in all Lisp things with cram* in the name and there there is rosprolog
20:04 ace4016 is there a reason a robot needs a robot specific OS?
20:04 dmiles_afk yet they are both on fruitfull tracks.. i dont know if they will converge
20:04 dmiles_afk ace4016: i dont think it is required no
20:05 dmiles_afk i think with ROS its was about setting scope of the kitchen/sink
20:05 ace4016 heh
20:06 dmiles_afk however sure theuir is liek a few pottential /dev/whatnots devices all batch scripts can wank arround
20:06 dmiles_afk like /dev/battery is usefull
20:07 dmiles_afk and the fact physical components are better if you can maintain them from the OS -level
20:08 Anniepoo ace - yes, just as we don't want to have to deal with inodes and insanity every time we write to file, a robot needs certain abstractions
20:09 Anniepoo message queues, hardware abstraction
20:10 Anniepoo dmiles, are you involved with rosprolog?
20:10 ace4016 hrm
20:10 dmiles_afk i studied it pretty heavy.. i follow and read their discussions
20:12 dmiles_afk they are seduced slightly by the semweb .. like it might come and make AI for them.. but i can tell they actualyl will become the sematic web since they are acutally more interested in using it for AI that the w3c
20:12 dmiles_afk which is refreshing
20:13 Anniepoo I think we need another 20% rule about RDF ;-)
20:14 dmiles_afk for isntance rosporlog loads http://knowrob.org/kb/ias_semantic_map.owl
20:14 Anniepoo dmiles, maybe we should have a robot hack weekend at The Elgin Works
20:15 Anniepoo or some more productive sort of event 8cD
20:16 Anniepoo the normalized geometric data is scary
20:16 Anniepoo we didn't even consider doing that
20:16 Anniepoo our lt persist is SVG path strings
20:17 zar001 makes a feeble attempt at raising his voice in attempts to establish rapport at #robotics ..... "ummm, i errr, ummm, *cough*, *cough*; I like ..., I -uh like robots. *COUGH COUGH COU
20:18 eazar001 =P, i know next to nothing about robots, but i did once show a lot of interest at the roomba open vac project.
20:18 eazar001 i thought that was kind of neat.
20:18 niepoo hands eazar a screwdr
20:18 zar001 fumbles and drops screwdriver
20:18 eazar001 what, is this ... this some kind of robot tuning device?
20:19 ace4016 i'll have to look at what some ROSes are offering
20:23 rue_house ros is stupid,
20:24 Anniepoo dmiles, how do you feel about the whole package/filesystem/node thing
20:24 rue_house just write a program to do what you want done
20:24 niepoo gently guides eazar away from the l
20:25 ace4016 lol
20:29 dmiles_afk Anniepoo, i think its ok.. though i admit it makes it seem more complex that it would need to be
20:30 Anniepoo yes, I never got why they needed to be a package manager as well as a message queue
20:30 dmiles_afk pacvkage manager makes sense to me.. as in how to i get rosprolog to have the right files
20:32 dmiles_afk message queue i assume is mostly like a debugger
20:32 dmiles_afk i mean the same way the linux dmsg is
20:33 Anniepoo no, the message queue makes sense to me - it's a good way to decouple these processes
20:33 dmiles_afk erm syslogd i meant
20:33 Anniepoo and yes, for a robot, being able to replay is of course important
20:34 dmiles_afk right yeah it maqkes sense for that for sure.. it just trying to get them to have the same topics that gets wierd
20:34 dmiles_afk i mena the rosprolgo and roslisp should be thinking about the same messages one day
20:34 Anniepoo yes, definitely
21:31 warlorck Hi, I'm currently making an Hexapod with 3DOF per legs. It's my first robotic project and I don't know where to start. I understand Inverse Kinematics but I don't know how to make a proper gait. If you have any tips, I will be glad to hear it :-)
21:37 rue_shop3 what is the meaning of life, the universe, and everything?
21:37 rue_shop3 warlorck, you gonna simulate or make it?
21:38 norias 42
21:49 warlorck rue_shop3: both if possible
21:49 warlorck I think of using V-Rep to simulate it
21:50 warlorck And the hexapod is already built
21:55 rue_shop3 ok
21:55 rue_shop3 you need a gait suggestion?
21:56 ace4016 find a spider, mocap the spider, release the spider, use in bot :P
21:56 rue_shop3 na
21:57 rue_shop3 define areas for the legs, an ideal position an area that the leg will need to be moved from before the next gait cycle and an area were the leg must be moved from immediatly
21:58 rue_shop3 as the body moves, use an update order to take each leg back to its 'comfort' position
21:58 rue_shop3 make sure you DO NOT move the legs to change the body position
21:58 rue_shop3 move the body, use the kinematics to figure out where the legs need to go to make it happen
21:59 rue_shop3 but most importantly, ignore me so that I come out with the best walker whenever the hell I get my platform going
22:02 warlorck ace4016: Sorry to contradict you but a spider have 8 legs :-p
22:02 warlorck But I appreciate
22:02 eadthem warlorck ther are some with 10
22:03 warlorck My bad !
22:03 ace4016 more training data!
22:03 warlorck An ant will do
22:04 eadthem its funny, i did inverse kinematics in a atmel once
22:04 eadthem takes .1 seconds per a cycle for a delta to convert X Y Z to motor A B C
22:04 rue_shop3 I really need to write a page on ik, people make way too big a deal of it
22:04 eadthem think it was pepole in here that helped me do that
22:05 warlorck rue_shop3: What do you mean by comfort zone ?
22:05 eadthem i did do it with floats so
22:05 rue_shop3 warlorck, the best place to put the foot to keep it out of the red zone by the end of the update cycle (of all legs)
22:07 warlorck My problem is that I don't know where to put my legs to make it follow the right direction
22:08 rue_shop3 your wrong
22:08 rue_shop3 its the wrong problem
22:08 rue_shop3 put the co-ordinate system on the world
22:09 rue_shop3 move the body over the world
22:09 eadthem http://final.servegame.com/public/inverseKinematics/ In case it helps anyone, the GCC folder is what ran. looks like it was rue and ace4016 that helped me.
22:09 ace4016 o.0
22:09 rue_shop3 each leg must change its position to make the points where it joins the body match the ground its touching on the world
22:09 eadthem iirc the clavdelt.pdf had BASIC and the GCC folder contains the C code i made it in to.
22:09 warlorck Okay, I was thinking that the coordinate was on the CG of the bot
22:10 ace4016 were you making a delta machine?
22:10 rue_shop3 the only legs that break that are the ones flying
22:10 eadthem i was for a while
22:10 ace4016 ah, i might have helped with some calculations or something
22:11 eadthem i thought kat might of also helped.
22:11 ace4016 i remember doing some equations by hand to find a somewhat closed-form solution
22:12 warlorck And the red zones are the zone that is impossible to reach ?
22:13 rue_shop3 yea
22:13 rue_shop3 thats the zone your too late on
22:14 rue_shop3 usually the leg is fully extended out and dragging
22:14 rue_shop3 I should post a video
22:14 rue_shop3 there are so many things I should do
22:14 rue_shop3 if you get the kinamatics right, you walk in any direction, pitch, roll, and yaw
22:15 rue_shop3 without dancing like the boston dynamics robots do
22:15 e_shop3 grumbles about boston dyna
22:16 warlorck Boston dynamics bots walk unlike mine :-(
22:17 rue_shop3 hopefully you do better
22:17 rue_shop3 afk
22:19 eadthem https://www.youtube.com/watch?v=cNZPRsrwumQ
22:19 eadthem guess the imperial walkers must of been made by boston dynamics
22:22 rue_shop3 I hate bigdog, its the stupidest thing ever
22:27 eadthem https://www.youtube.com/watch?v=chPanW0QWhA
22:27 eadthem i would think if they told the legs to be a bit further from the middle, it might work better
22:31 eadthem https://www.youtube.com/watch?v=_luhn7TLfWU Finaly MIT has a use for there gynmasium
22:37 rue_shop3 cut all the legs off, put it in a car crusher, forget it was ever made
22:42 warlorck I think I will try to implement a tripod
22:45 rue_shop3 dont be crazy, start with a monopod
22:45 rue_shop3 it totally removes the problem of which leg to position
22:46 eadthem its intersting, but that could work in theory
22:47 warlorck It seems more realistic, thnk for the advice
22:47 eadthem https://www.youtube.com/watch?v=kRDrKg5J9jk
22:51 eadthem ya i wonder if it could balance that well if it had no motors
22:52 rue_shop3 nobody can tell when I'm being carcastic
22:55 rue_shop3 ugh, they are so not doing that right
22:56 rue_shop3 moving on
23:07 warlorck Do you have resources to get started with the monopod ?
23:31 rue_house no, its was a stupid idea, I was just being sarcastic
23:31 rue_house make a hexapod, they can be done to garuntee it staying up when any one leg is lifted
23:32 rue_house dont raise 6 legs or it'll fall down
23:34 zhanx Octopod
23:34 rue_house decapod
23:34 zhanx Raise 6 and balance it
23:42 warlorck rue_house: I jumped right headlong into the monopod idea x). I will try the tripod.
23:43 warlorck But I have no idea where to put the flying legs