#robotics Logs

Oct 21 2014

05:06 Jak_o_Shadows um, moment
05:07 Jak_o_Shadows wait, what? don't get it
05:09 woerr can i add a sum into an if statement
05:10 Jak_o_Shadows oh. Line breaks dude.
05:10 Jak_o_Shadows Um, so if (a-b>c) then c = c+5 ?
05:10 woerr yes
05:10 Jak_o_Shadows uh, sure? Why wouldn't you be able
05:11 woerr so i can work sums out in the statement
05:11 woerr ?
05:11 Jak_o_Shadows well, the {} mean that it's not in the comparison of the if statement, but in the body of it
05:12 woerr eg if (a+b+c > d) {d = 5;}
05:12 Jak_o_Shadows if (a+b+c > d) {
05:12 Jak_o_Shadows d = 5;
05:12 Jak_o_Shadows }
05:12 Jak_o_Shadows ?
05:13 woerr yes, is that valid
05:13 Jak_o_Shadows then yeah, sure? That'll work
05:13 woerr hmm
05:13 woerr mine aint working for some reason
05:21 woerr http://pastebin.com/nfA5bJz6
05:23 Jak_o_Shadows you're changing mot1pps, then checking it again? Maybe?
05:23 woerr the code is supposed to make gspps equal to mot1pps
05:24 woerr if the difference is more that 20 above or below it will increment by 15
05:25 woerr i found my problem
09:23 rue_house heh
09:23 rue_house servo controller lookup table?
09:23 rue_house usually proportionate is done differeently
13:40 deshipu https://www.youtube.com/watch?v=nHvaqpcGI1c#t=82
13:40 deshipu nice
14:09 wolfmanjm yea that is nice, I wish they would publish the code they use for that ;)
14:10 wolfmanjm is that 3DoF or 4?
14:11 deshipu I think it's 5
14:11 deshipu at least 4
14:15 deshipu it's 4, with the rotation of the whole leg at hip
14:21 wolfmanjm I still have not got mine to walk yet, there is a huge leap from getting the IK to work and getting coordinated leg movement
14:34 deshipu wolfmanjm: hmm, get it to shift the body sideways first
14:34 deshipu wolfmanjm: the rest is just making steps
14:34 wolfmanjm so just change the X for all 4 legs?
14:35 deshipu yeah
14:35 wolfmanjm or X and Z?
14:35 deshipu well, have (dx, dy)
14:35 deshipu for any direction of walking
14:36 wolfmanjm well with the 4 legfs at 45° to each other just movin gin X won;t move sideways
14:36 wolfmanjm well 45° to body
14:37 wolfmanjm I could make th elegs at right angles then X would move the body sideways I guess
14:38 deshipu well, that's why I had that "hip offset" thing there
14:39 deshipu but even without it, you just convert between the body reference frame and the leg reference frame coordinates
14:39 deshipu using a transformation matrix
14:39 deshipu just multiply the vector by the matrix
14:40 deshipu https://en.wikipedia.org/wiki/Transformation_matrix
14:41 deshipu wolfmanjm: especially, you want https://en.wikipedia.org/wiki/Transformation_matrix#Other_kinds_of_transformations
14:41 deshipu wolfmanjm: so you want a 4x4 matrix
14:41 deshipu wolfmanjm: and the vector to be [x, y, z, 1]
14:41 wolfmanjm yea I am very familiar with 3D transformations (I wrote the first full 3D arcade games at Atari ;)
14:42 deshipu so that should be easy for you
14:42 wolfmanjm although that was nearly 30years ago :)
14:42 deshipu (I still have to do it with my code)
14:43 deshipu wolfmanjm: when you can use the coordinates relative to the body frame, it should be all much easier
14:43 wolfmanjm BTW if W is always 1 then a simple 3x3 is only required for rotations and add the translation, it saves a few multiplies
14:43 deshipu later on you will want to have the virtual vehicle and world frames too
14:43 deshipu with their own matrices
15:00 deshipu https://www.youtube.com/watch?v=lb3Riq06fVI
15:00 deshipu wow, the guy is still working on it
20:41 rue_shop2 i really dont know why you guys are so married to using matrixies, they are really just a waste of time
22:47 rue_house matrix pansies
22:47 wolfmanjm math-o=phobe ;)
22:53 rue_house its just so easy to trig out
22:53 rue_house I'v not seen matrixes done in a way that is less processing intensive than trig
22:54 rue_house lots of extra shuffling numbers around that aren't doing anything
22:55 rue_house I dont get it, I'm doing things, but nothings getting done, that I can tel