#robotics Logs

Mar 18 2018

12:50 AM rue_: ok I need to think
12:53 AM rue_: but I think they burned out my brain
07:07 AM z64555: acetone dissolve sounds like a bad time
01:06 PM Bernhard is now known as Hamilton
02:08 PM Bernhard is now known as Hamilton
02:10 PM mikah: hey rue
03:38 PM rue_: mikah_, hey
03:38 PM rue_: you got he computer online,
03:38 PM rue_: I just got back from sechelt, have to go out again tho
05:19 PM ferdna: what speed should i use to cut 1/8" thick aluminum when using a wood blade on a band saw? I understand that with a metal blade you have to go minimum speed...
05:29 PM z64555: push until you hear the saw change speeds, then back off a tiny bit
05:31 PM z64555: I'm not sure if this will damage the blade or not, nothing that I could tell from a quick inspection, at least
06:02 PM ferdna: thanks!
06:06 PM SpeedEvil: if the aluminium is heavily oxidised, sanding off the oxide in the area of teh cut will reduce saw wear
06:12 PM z64555: hmmm. no. quaternions *dont* have a singularity at 0, euler vectors and maybe also rotation matrices do
06:29 PM ace4016: you can use dual ones for position info too! http://gamedevs.org/uploads/beginners-guide-to-dual-quaternions.pdf
06:30 PM ace4016: though you're on your own for implementation, unless some folks have created a nice library
06:30 PM z64555: no, I don't want to deal with octernions yet, thank you
06:31 PM z64555: I have enough trouble with 4 elements right now :P
06:31 PM ace4016: lol
06:32 PM Rez is now known as LoRez
06:35 PM z64555: Now, the real question arises.
06:36 PM z64555: Given two orientations in quaternion form, how Do I create a delta so that a PD controller or similar can operate on them?
06:36 PM z64555: Is it a simple difference, or do I need to create another quaternion (like with cross product) to rotate the actual to the target?
06:46 PM ace4016: when you say difference, you mean angle difference?
06:50 PM ace4016: iirc it's usually something like q' = inv(q1)*q2. then you can convert the q' quat into an axis-angle representation to look at its angle, which is the difference, assuming you meant angle difference
06:51 PM z64555: well yes, angle difference
06:52 PM z64555: I don't really want to convert into euler angles, though
06:52 PM z64555: trying to keep the computation overhead at a minimum
06:52 PM ace4016: hrm, you can parameterize the "angle" portion of the quat then
06:53 PM ace4016: rather than converting it
06:53 PM z64555: yeah, that's what I was thinking of doing
06:53 PM z64555: have controllers on the three imaginary parts of the quaternion and just ignore the real part
06:54 PM ace4016: but iirc, getting the angle is theta = 2*arccos(q'_0) (assuming q'_0 is the angle bit)
06:55 PM z64555: that's if you're doing the angle between the two
06:55 PM z64555: hm.
06:55 PM ace4016: that's the angle difference, no? :P
06:56 PM z64555: Well, yes and no
06:56 PM z64555: The controllers need to be in the reference frame, like x, y, z
06:57 PM ace4016: hrm, tricky thing with quats iirc. i think i need to brush up on quats
06:57 PM z64555: ...which tbh I need to do a conversion into said frame, anyway
06:58 PM z64555: yeah, quaternions and rotation vectors are normal to the points/group that's being rotated. They point along the axis of rotation
07:34 PM z64555: I think
07:34 PM z64555: What I'll have to do
07:35 PM z64555: Is q_d / q_a, where _d is the desired orientation, and _a is the actual
07:36 PM z64555: Then convert the resulting q into tait-bryan angles so that the controllers can do their work
07:36 PM z64555: (tait-bryan are aviation angles, {yaw, pitch, roll})
07:37 PM z64555: other convention I could use is the z-x-z convention, although I'm not sure what that would actually involve
08:00 PM z64555: or I might actually be able to just use the i,j,k components directly
08:01 PM z64555: need to verify that scaling a rotation q maintains the axis of rotation, however...
08:30 PM rue_: yaw, pitch, and roll, applied in which order?
08:31 PM rue_: as I understand it, without the quaternions, your facing gimbal lock
08:33 PM z64555: two seperate issues. When applying the rotation, its "backwards" to the readout (so roll, pitch, then yaw)
08:34 PM z64555: this is intrinsic angles, what the object sees
08:34 PM z64555: gimbal lock occurs when the the underlying math doesn't use a quaternion or suitable rotation matrix
08:35 PM rue_: geez today was long
08:36 PM Tom_L: but what did you acomplish?
08:37 PM orlock: i attached a servo to a trigger-pump style spray bottle
08:37 PM orlock: going to wook up some sort of rangefinder/motion sensor
08:37 PM orlock: damn possums have been eating the succulents
08:37 PM rue_: I went and bought new gloves, went to the dump, got some metal for a project for work, dropped by a buddies (wasn't home), dropped off a charger to someone, got back, unloaded the metal, went to see ma and picked up a bunch of stuff from her place to bring back to her storage, unloaded it all at home, loaded up a bunch of scrap metal and took it to the local recycler, , got into building stuff
08:37 PM z64555: orlock: why not a roller pump
08:38 PM rue_: air based water gun
08:38 PM orlock: spray bottle , $5 at hardware shop
08:38 PM rue_: mmm sanry gun
08:38 PM * z64555 tries to find the actual name for it
08:38 PM rue_: sentry
08:39 PM rue_: yea, use the nozzel
08:39 PM z64555: I'm thinking of the thing you'd see on some automatic dishwashers
08:39 PM rue_: then take a pop bottle and fill it 10% with compressed air
08:39 PM Tom_L: i got nothing useful done
08:39 PM z64555: It's a pump with a rubber house inside that's squeezed by a rotating set of rollers
08:39 PM orlock: z64555: yeah, i know the sort
08:39 PM orlock: this doesnt need to be food safe, etc
08:39 PM rue_: yea
08:40 PM orlock: they use them in insulin pumps, etc
08:40 PM z64555: no, I'm thinking more along the lines of a more consistant stream
08:40 PM rue_: but air proppelled water is awesome
08:40 PM z64555: hm, air propulsion may be easier to fabricate
08:41 PM orlock: peristaltic
08:41 PM rue_: pop bottle
08:41 PM orlock: rue_: they have those as well, i could have gotten one for about the same price
08:41 PM orlock: but i dont need to worry about having a plstic pressure vessle sitting ourside while the plastic weakens, etc
08:42 PM orlock: anyway, it's done, just needs the sensor added
08:42 PM orlock: but i was seriously considering air pressure
10:44 PM id4rk_ is now known as id4rk
10:52 PM rue_shop3: motion detector
10:53 PM rue_shop3: they have silly motion alarms at the dollar store
10:53 PM rue_shop3: wireless, trigger a buzzer or something silly
10:53 PM rue_shop3: or,.... was it all one peice....
10:54 PM orlock: yeah i have a few around, sonar or IR
10:54 PM rue_shop3: https://www.aliexpress.com/item/HC-SR505-Mini-sensing-module-body-sensing-mode-mini-body-sensor-switch-NEW/32334062992.html
10:54 PM rue_shop3: I got a thing like that, I'v not tried it out yet
10:55 PM rue_shop3: https://www.aliexpress.com/item/Pyroelectric-Infrared-sensor-Human-Body-Detecting-PIR-Motion-Sensor-Module-for-Arduino-MCU-023115-optical-lens/32704422261.html
10:55 PM rue_shop3: more like that
11:30 PM Damionwat54: Any cool robotics projects anyone?
11:48 PM rue_: not this weekend