#robotics Logs

Mar 06 2017

03:57 AM Jak_o_Shadows: Hmmm. Starting to get an idea of how the legs are going to work.
04:14 AM Jak_o_Shadows: Hmmm. On a cat, the front leg, the shoulder joint, it can go close to 180 deg
04:16 AM deshipu: it can go full 360
04:16 AM deshipu: but only once
04:17 AM Jak_o_Shadows: ah. Well, normal range of motion
04:17 AM Jak_o_Shadows: Trying to think of what part of the gait that actually needs 180
04:17 AM deshipu: pouncng
04:17 AM Jak_o_Shadows: opening doors :Pp
04:22 AM Jak_o_Shadows: I suppose a full-on running motion would
04:22 AM deshipu: stretching
04:22 AM Jak_o_Shadows: Luckily, motors don't need to stretch :D
04:22 AM Jak_o_Shadows: Through, I wonder if that's going to be a concern with air muscles at some point
04:56 AM SpeedEvil: Distributing lubrication
04:57 AM deshipu: lubricating distribution
05:41 AM Jak_o_Shadows: Even just pre-stretching before the first use.
05:43 AM Jak_o_Shadows: ‪Like with balloons
06:04 AM * SpeedEvil ponders silly idea again.
06:04 AM SpeedEvil: Baba-yaga shed.
06:05 AM deshipu: AT-ST
06:09 AM SpeedEvil: at-at is probably more practical.
06:09 AM deshipu: for what
06:09 AM SpeedEvil: A shed.
06:10 AM deshipu: I think that for a shed, a shed beats everything else
06:10 AM SpeedEvil: yes, but who said you can't put legs on it.
06:10 AM deshipu: you can't and still claim it's practical
06:11 AM SpeedEvil: Well, it's admittedly not as practical as if you could do the LASER cannons properly.
06:11 AM deshipu: how about a gingerbread house shed?
06:13 AM SpeedEvil: Unfortunately, with todays health and safety culture, baking children is considered a bad thing nowadays.
06:15 AM deshipu: o tempora, o mores
06:15 AM deshipu: a beautiful tradition dies
08:15 AM [cnc] is now known as [cube]
11:58 AM Guest66078 is now known as CrazEd
11:58 AM CrazEd is now known as Guest90411
12:58 PM Guest90411 is now known as CrazEd
12:59 PM CrazEd is now known as Guest79035
01:42 PM Atomsk is now known as ace4016
01:59 PM Guest79035 is now known as CrazEd
01:59 PM CrazEd is now known as Guest12003
02:59 PM Guest12003 is now known as CrazEd
03:00 PM CrazEd is now known as Guest16998
03:03 PM [cnc] is now known as [cube]
04:01 PM Guest16998 is now known as CrazEd
04:01 PM CrazEd is now known as Guest68406
05:01 PM Guest68406 is now known as CrazEd
05:02 PM CrazEd is now known as Guest81277
06:02 PM Guest81277 is now known as CrazEd
06:02 PM CrazEd is now known as Guest64882
07:02 PM Guest64882 is now known as CrazEd
07:03 PM CrazEd is now known as Guest9651
08:03 PM Guest9651 is now known as CrazEd
08:04 PM CrazEd is now known as Guest19207
09:09 PM Guest19207 is now known as CrazEd
09:10 PM CrazEd is now known as Guest57336
09:34 PM justan0theruser: can I expect problems with high torque applications if I mount a pulley to a motor where the bearing is bigger than the shaft? https://imgur.com/a/La2xJ
09:36 PM justan0theruser: shaft is 30mm, pulley is https://www.mcmaster.com/mva/library/20161031/1326n520_at10%20series%20timing%20belt%20pulleyl.gif
09:36 PM justan0theruser: maybe I should mount it the reverse way so the load is on the center of the shaft :
09:38 PM rue_shop3: is this for the 'gonna sheer something' arm?
09:38 PM rue_shop3: isn't that the motor with the coupler stuck to it?
09:38 PM justan0theruser: no
09:39 PM justan0theruser: what do you mean gonna sheer something arm
09:40 PM rue_shop3: there was a guy trying to make a robot arm with some crazy specs and some undersized hardware
09:40 PM justan0theruser: thats probably me
09:40 PM justan0theruser: what undersized hardware
09:40 PM rue_shop3: pans a thru c
09:40 PM rue_shop3: plans..
09:41 PM justan0theruser: what
09:41 PM justan0theruser: thru c?
09:41 PM z64555: "everything"
09:41 PM justan0theruser: never heard that expression
09:42 PM rue_shop3: in answer to your question, everything should fit right
09:42 PM rue_shop3: have you ever heard the term 'plan A' ?
09:42 PM justan0theruser: sure, but does it reduce my torque rating
09:42 PM z64556: somebody messin with my router? :(
09:42 PM justan0theruser: rue_shop3: yes, it's what you do before you use emergency contraceptive
09:43 PM rue_shop3: not fitting things right causes stuff to not connect properly and come loose, resulting in something that falls apart
09:44 PM z64556: I say just let him build and learn
09:44 PM z64556: if it breaks, it breaks
09:44 PM justan0theruser: I don't understand this philosophy
09:44 PM justan0theruser: if I can minimize the number of things that break, that is a good thing
09:45 PM rue_shop3: justan0theruser, you have the ora of someone who needs to learn lots about making things
09:45 PM rue_shop3: 'what you can get away with' is kinda an individual thing
09:46 PM z64556 is now known as z64555
09:46 PM z64555: justan0theruser: yes, it is a good thing, but it can hold back the learning process if you have to stop at every detail
09:46 PM justan0theruser: I suppose
09:46 PM justan0theruser: z64555: I mostly come here when I have stopping time and I'm desiging another part
09:46 PM rue_shop3: I made a cnc machine in like 2-3weeks cause I just did it
09:47 PM justan0theruser: surprisingly little has broken, though I did order the wrong bearing size because I went by diameter not width
09:48 PM justan0theruser: rue_shop3: did you mock it in a CAD program
10:10 PM Guest57336 is now known as CrazEd
10:11 PM CrazEd is now known as Guest70630
10:14 PM rue_house: no I just BUILT IT
10:27 PM deltaModulator: I have designed a robotic arm in OpenSCAD and I would like to hear perspectives on the design. https://postimg.org/image/xetw2ajur/
10:41 PM synja: are the gears really needed?
10:54 PM deltaModulator: how so?
10:55 PM deltaModulator: there must be a way to drive one segment from the preceding one.
10:55 PM deltaModulator: i have an animation of a closeup of the gears if that would help.
11:11 PM Guest70630 is now known as CrazEd
11:12 PM CrazEd is now known as Guest58575
11:13 PM branjb: anyone have much experience with IMUs? if I have an IMU mounted on a static base, and an IMU on a moving joint, can I get accurate yaw from comparing the two IMUs?
11:15 PM SpeedEvil: branjb: No.
11:16 PM SpeedEvil: Unless you only care about fractions of a second to several seconds.
11:16 PM SpeedEvil: depending on sizes and accuracies
11:16 PM branjb: I want to know the reference frame of the actuator in relation to the base
11:16 PM SpeedEvil: Gyros are somewhat better.
11:16 PM SpeedEvil: Gyros may drift a few degrees an hour, which can be fine.
11:18 PM SpeedEvil: For example - the MPU6050.
11:19 PM SpeedEvil: The total RMS noise is 0.05 degrees a second RMS
11:19 PM branjb: that's what I'm using right now; I can get accurate tilt angles, but on a cold start I don't know my yaw in relation to the base
11:19 PM SpeedEvil: run to a stop
11:20 PM branjb: yeah, that's ideal; trying to come up with a way to have orientation on a cold start without using absolute encoders
11:20 PM SpeedEvil: In principle, you could use a magnetometer.
11:20 PM branjb: or calibrating to known positions
11:20 PM SpeedEvil: And get better than a degree.
11:20 PM SpeedEvil: But that relies on nearby lack of metal and magnetism.
11:20 PM branjb: I only need to be accurate within +/- 1 degree
11:20 PM SpeedEvil: And almost certainly won't work in practice
11:20 PM branjb: and motors are probably in too close proximity to use a magnemoter, though I haven't tested it
11:22 PM deltaModulator: how about a magnetic reed switch and a magnet?
11:23 PM branjb: it would probably work, but we want to know orientation without going through a calibration routine
11:25 PM SpeedEvil: Add an absolute encoder of some form.
11:25 PM deltaModulator: use the accelerometer in combination with the gyro to refine the reading?
11:25 PM SpeedEvil: deltaModulator: not for yaw - rotation around the vertical
11:26 PM branjb: http://www.olliw.eu/storm32bgc-v1-wiki/Using_a_2nd_IMU seems to lead me to believe I can find yaw using a static IMU on the base, and one on the actuator
11:26 PM branjb: I only care about the relation from the base to actuator, not in the earth reference frame
11:27 PM SpeedEvil: branjb: It's flat wrong.
11:27 PM SpeedEvil: Your IMU will drift at kilometers an hour
11:28 PM branjb: would it be accurate within the first minute or two?
11:28 PM deltaModulator: at vertical angle, accelerometer does not help, true, but at any other angle, would the tilt sensing of the accelerometer not work? or am i picturing it wrong?
11:28 PM branjb: only need to know where it is at start up
11:28 PM branjb: not hours later
11:28 PM SpeedEvil: It needs to be continually running in order to work at all as an IMU
11:29 PM SpeedEvil: If you switch it on, you have no idea of the relative positions at all
11:29 PM branjb: howso? I currently have an IMU on the actuator and as soon as I turn it on I know the pitch angle
11:30 PM SpeedEvil: yes, but you can't know rotation about the vertical
11:30 PM SpeedEvil: Unless you use a magnetometer
11:30 PM SpeedEvil: (and knowledge of the magnetic field)
11:30 PM branjb: at that point I don't care if it drifts, because I know it's pointing at say 45 degrees, and I know the steps per degree of my motor. so I can just keep track of steps
11:30 PM deltaModulator: stick a magnet on it to increase the magnetic field to the point that it dominates the noise from the motor?
11:30 PM SpeedEvil: When you turn it on it does not zero. There is no infoirmation at all about the position.
11:31 PM branjb: howso? if I put my IMU at 45 degrees, turn it on and start reading from it, it says 45 degrees
11:31 PM SpeedEvil: branjb: Because it is reading the angle.
11:31 PM SpeedEvil: Of gravity with respect to the device.
11:32 PM SpeedEvil: For yaw, this is obviously always zero and you can't measure it
11:32 PM branjb: yeah
11:32 PM SpeedEvil: (unless you can move the whole robot)
11:32 PM branjb: I actually can move the whole robot
11:32 PM branjb: does that help?
11:32 PM SpeedEvil: If you drive forward, you now know the relative yaws because of the differences in accelleration
11:33 PM SpeedEvil: as you know that it should be moving parallel
11:33 PM branjb: ah! that's what I was kind of getting at
11:33 PM branjb: that would be comparing the IMU on the static robot base with the IMU on the actuator, right?
11:33 PM SpeedEvil: yes.
11:33 PM branjb: so, for that to work, it would require slightly moving the robot
11:33 PM SpeedEvil: yes.
11:34 PM branjb: ok, that helps a lot
11:34 PM SpeedEvil: Even vibration may be enough
11:34 PM SpeedEvil: If you know the response well enough
11:36 PM branjb: i'll be experimenting with that then, as well as the possibility of mounting a strong magnet and just using a single 9 DOF IMU
11:36 PM branjb: I imagine the magnet would have to be pretty close
11:36 PM SpeedEvil: What is the robot?
11:37 PM branjb: the motors are also mounted at fixed positions exactly parallel with the IMU, so that could possibly used as well
11:37 PM branjb: it's an underwater ROV, my team is designing a pan/tilt camera for it
11:38 PM SpeedEvil: Interesting.
11:38 PM branjb: during power cycles the payload platform can potentially drift, and we're looking for solutions to know it's orientation at startup without having to move to hard stops
11:38 PM SpeedEvil: have the camera look at something
11:38 PM branjb: haha, that's one idea
11:39 PM branjb: we're exploring that with a fiducial, an IMU, and absolute encoders
11:39 PM SpeedEvil: though that requires light
11:39 PM SpeedEvil: (and turbidity)
11:39 PM branjb: it has a light mounted with the camera