#robotics Logs

Feb 18 2015

00:17 rue_more so close...
00:18 rue_more PATH='/sbin:/bin:/usr/bin'
00:18 katsmeow-afk not really, if you can get the compression up on a gasoline engine, instead of feeding it gasoline, you feed it normal city air as fuel
00:20 rue_more how about if you could bring the temp of the air in a gas engine cylinder from 20c to 1500c without gas?
00:21 katsmeow-afk timing might matter
01:52 theBear heh, how about if you could do that... you'd revolutionise the smelting industry for a start
01:53 theBear on that note, you probly wouldn't want one of those modern ally block engines if you were doing that kinda thing, pretty sure they'd be liquid long before 1500c
01:55 Jak_o_Shadows What do people think of the dirt cheap ebay air compressors. the ones around $15
01:56 katsmeow-afk 12v ?
02:00 Jak_o_Shadows mmm, yeah.
02:02 katsmeow-afk good for intermittant use, there is no oilingof the moving parts, and no allowance for fatigue failure
02:04 katsmeow-afk in some cases, there is no bottom to the cylinder, the piston is visably free to suck dirt up into the cylinder on the compression stroke
02:05 theBear i think they sound suspicious
02:06 theBear tho i've seen a few of those little "4wd tyre pump" kinda ones thug thug along for years happily
02:19 Jak_o_Shadows Hmmm
02:20 katsmeow-afk i carry one in the car , i'd never build a robot with one
02:21 katsmeow-afk i don't carry a $18 hipressure piston water pump in the car, but i would build a robot with one
02:22 katsmeow-afk not that i would use water in the robot
02:37 Jak_o_Shadows I was thinking of water rockets tbh
02:52 katsmeow-afk the bottles blow up if you go over 130psi
02:53 Jak_o_Shadows Well, hence teh pump, it's REALLY painful doing it by hand
02:56 katsmeow-afk standup a section of water pipe, with air in it, fill with water, whenthe air at the top reaches your desired psi, shut off the water pump
02:56 Jak_o_Shadows that isn't a bad idea
02:57 katsmeow-afk read the ratings on the pipe, do not exceed, allow a safety margin
02:58 Jak_o_Shadows a decent safety margin.
02:59 katsmeow-afk iirc, 10ft of 3" pipe is a ~4 gallons, so a 2gpm pump gives you worst case a 2 minute shot rate, if you use all the air in one shot
02:59 Jak_o_Shadows Mmm, I think I can do those sorts of calculations
03:04 deshipu katsmeow-afk: man, why can't you people use normal units?
03:04 Jak_o_Shadows (hence why I could do my own calcs :P)
03:06 katsmeow-afk what's metric for a minute?
03:09 deshipu 60 seconds
03:09 katsmeow-afk not very 10's based, is it?
03:10 deshipu katsmeow-afk: well, you have kiloseconds, megaseconds, etc.
03:10 deshipu katsmeow-afk: they just don't translate directly to your weird 60-based system
03:20 Jak_o_Shadows dude, time is one thing.
04:00 SquirrelCZECH katsmeow-afk: yeah
04:00 SquirrelCZECH why that units?
04:00 SquirrelCZECH ;D
06:16 Martindoo Hi
06:17 deshipu hello Martindoo
12:05 veverak Triffid_Hunter: any tips for fast way to get centroid of polyhedron ?
12:05 verak got 20k poitns and his method took too much
12:06 codepython777 veverak: you want the centroid of the surface? the volume?
12:06 veverak ehmm
12:06 veverak need point to calculate volume
12:06 codepython777 so you just need a point inside?
12:06 veverak based on points which mades the surface of the object
12:07 veverak but wrongly selected point messes up the calculations
12:07 veverak codepython777: I calculae the volume of object by making tetrahedron with 'center' point AND each triangle
12:08 codepython777 you are interested in volume? your polyhedron is convex and triangulated?
12:08 veverak convex?
12:08 veverak ANY polyhedron :)
12:09 veverak made of triangles
12:09 veverak it works, perfeclty
12:10 veverak but my fancy way to get the center point is slow
12:23 veverak because actually
12:24 veverak it calculates for 20k points 187550028 times distance from two points
12:24 veverak distance = size_of_vecotr(sum_vectors(A, neg_vector(B))
12:24 veverak yay
15:09 deep HI
15:11 deep I want to measure the orientation of a walking human leg and feed the data to a motor which drives a link so that the link never deviates from the position of leg more than some given degrees.
15:11 deep Is it possible?
15:13 deshipu everything is possible
15:13 deshipu some things are just prohibitely expensive
15:14 deep Budget is not a problem.
15:14 deshipu famous last words :P
15:15 deep I just want to know if it is possible and how it can be done.
15:15 deshipu there is a million ways, the easiest is to just do it mechanically, with a potentiometer attached to the leg
15:17 deep deshipu: Are there any electronic sensors for measuring tilt or orientation that can measure the position of leg and link with respect to vertical?
15:17 veverak deshipu: btw: any alternatives for the openvm?
15:17 veverak https://www.kickstarter.com/projects/botthoughts/openmv-cam-embedded-machine-vision
15:18 deep What I thought was to measure the orientation of the leg and the link wrt absolute vertical and give the error signal to the motor.
15:18 deshipu veverak: nope, the guy made it precisely because all the alternatives suck :)
15:18 deshipu deep: sure, you have to use PID
15:18 veverak deshipu: thought so
15:18 veverak well
15:19 deep deshipu: What sensors can I use? And what about their accuracy? I mean how accurately can they measure the angles wrt the absolute vertical?
15:20 deshipu deep: there are accelerometers and gyroscopes, but they give rather noisy output and you have to filter them, which gives you lag and not much accuracy
15:20 veverak deshipu: I suppose I will wait for wi-fi anyway
15:20 deshipu deep: mechanical sensors are going to be the cheapest and the most accurate
15:20 deshipu deep: then you can use optical ones
15:21 deshipu deep: even full motion capture rig
15:21 deshipu veverak: I think the wifi shields are going to be sold separately
15:21 deshipu veverak: they do have prototypes for them, though, so I'm sure they will be available
15:22 veverak yeah
15:22 veverak but it's useless for me without it
15:22 veverak ...
15:22 veverak ;)
15:22 veverak and I don't need it today, or tomorrow
15:22 deshipu yeah, and the kickstarter is funded already
15:23 deep deshipu: What I want to do actually is to attach a link on a human leg and move it with human leg as the leg moves.
15:23 deep deshipu: Another thing that I thought was to attach a force sensor between leg and the link. The force sensor will sense the force between the human leg and the link and if the force crosses a certain level, the link should move.
15:24 deshipu deep: yeah, good luck
15:24 deep So that's a bad idea.
15:25 deep So it's better to use optical or mechanical sensors?
15:25 wizkidsafo Greeting, If I'm doing a solar pannel installation and I want to record the data which computer 'robot should i use. I know Ill be using a mppt tracker to track the resistance and voltage but am unsure of what 'robot ' to use... any inputs
15:26 wizkidsafo the system im creating is both pv and thermal
15:27 deep deshipu: Ok...can you please tell something more about the optical and mechanical sensors that you talked about?
15:28 deshipu deep: https://en.wikipedia.org/wiki/Motion_capture
15:29 deep deshipu: But this is not feasible for what what I am doing. It needs a camera to capture the motion at all times which is not possible.
15:31 deshipu deep: there are many methods listed there
15:31 deshipu deep: motion capture is basically what you want to do, just read that article to know how it can be done
15:42 deshipu in particular, you have section on non-optical systems there