Back
[04:05:47] <deshipu> anniepoo: convert them to metric \o/
[09:32:24] <rue_house> it looks like the body is still being positioned by the legs tho
[09:32:56] <deshipu> what should it be positioned by?
[09:33:14] <deshipu> should it be an independent entity gently floating over them?
[09:33:27] <rue_house> the body should be positioned by the brain, the legs should be positioned by the body
[09:33:51] <deshipu> I'm afraid that only works in computer game graphics
[09:34:01] <rue_house> the position of the legs should be such to suit where the body is and where its going
[09:35:13] <rue_house> that way you dont get jerky motion
[09:36:36] <rue_house> it also makes the system able to travel at whatever yaw/pitch/roll it wants to
[09:36:51] <rue_house> because the legs are moving to suit where the body is
[09:44:15] <deshipu> actually, with your system, you don't need legs at all!
[09:44:28] <deshipu> the body can just move around on its own
[09:44:29] <rue_house> its too bad you dont get it
[09:45:11] <jhylands> rue_house, that
[09:45:20] <jhylands> that's pretty much how IK works
[09:45:35] <deshipu> and pretty much how we do it
[09:45:45] <rue_house> I'v seen _2_ people do it the right way around
[09:45:52] <deshipu> however, it's not enough once you move out of the range of your legs and need to start making steps
[09:46:08] <rue_house> and they can move the body around however they want and the legs move to keep it up
[09:46:22] <jhylands> rue_house, this is my quad walking with IK:
https://www.youtube.com/watch?v=1q4nsocmLWU
[09:46:22] <rue_house> so the motion range is great, and the bodys motion is completely smooth
[09:46:26] <jhylands> not sure if you've seen it
[09:47:12] <rue_house> but everyone tries to make motion patterns of legs to move a body around, and the bodys motion is jerky as all hell, they cant orient the body in any way.
[09:47:45] <deshipu> rue_house: I think you have no idea what we are doing
[09:48:04] <deshipu> rue_house: you are just making strawman assumptions and give "good advice" based on that
[09:48:15] <rue_house> ugh
[09:48:18] <deshipu> it's annoing
[09:52:18] <rue_house> https://www.youtube.com/watch?v=ryO03bQk3m4
[09:52:34] <rue_house> pretty sure that was one of the robots where he was doing it right, see how it can strave?
[09:52:59] <jhylands> mine can do all that also - I just haven't programmed it to do that
[09:53:16] <jhylands> I'm more interesting in actual walking and navigation than cute party tricks
[09:53:35] <rue_house> were not talking about what the robot can physically do, were talking about how its programmed
[09:53:49] <jhylands> I can adjust the pitch, roll, and yaw of the body independently of the legs
[09:53:51] <jhylands> I just don't
[09:54:27] <jhylands> Mine is programmed to be able to do that - those three values (plus the height and left-right/back-front position) are just variables
[09:54:39] <jhylands> I just happen to keep all six of them as constant values
[09:56:56] <deshipu> rue_house: he does absolutely no planning of where the robots puts the legs
[10:02:33] <jhylands> rue_house, this is the code where five of those six values are set:
https://github.com/JonHylands/roz/blob/master/Nuke.py#L48-L53
[10:03:32] <deshipu> practically everybody do IK, except for the most basic 2-dof-per-leg toys
[10:04:30] <deshipu> the problem we were discussion is the choice of which leg to move next and where to move it when the body is getting out of the range of motions it can do with the legs staying on the ground
[10:04:42] <deshipu> and unfortunately that requires a little bit of anticipation
[10:04:56] <jhylands> the last of the six values is set here:
https://github.com/JonHylands/roz/blob/master/Nuke.py#L97
[10:06:07] <deshipu> or you just hardcode it and accept that you have a suboptimal solution
[10:20:01] <veverak> everybody does it wrong except for me!
[10:20:08] <veverak> I don't yet know how I do it :/
[10:24:25] <deshipu> the spring model sounds reasonable at least for determining which leg to move next
[10:24:49] <deshipu> or just choose the one that is the furthest away from the body
[10:25:37] <deshipu> moving it along the current direction of body movement, with some margin for unexpected changes, also seems reasonable
[10:26:42] <deshipu> although you might want to add a limit on the minimum area of support to that, so that you don't put all legs on one line accidentally
[10:28:35] * veverak wants to make it modular by design maybe
[10:29:03] <veverak> not exactly implemented modular, but with design that allows easy rewrite
[10:30:44] <deshipu> always a good thing
[10:34:59] <veverak> NO NO NO NO
[10:35:06] * veverak wrote something nad "lyx" freezed
[10:35:08] <veverak> damn it
[10:35:58] <deshipu> lyx is a prosthesis
[10:36:34] <deshipu> write in vim as Saint Knuth intended
[15:18:55] <RustyShackleford> is there a channel for mechanical engineering?
[15:19:50] <robopal> ask here first
[15:20:29] <robopal> people do mech stuff here
[15:22:03] <deshipu> they are not real people
[15:24:22] <RustyShackleford> wait is this #robots or #robotics? :p
[15:24:49] <RustyShackleford> well I'm trying to design a sturdy adjustable height desk
[15:25:03] <RustyShackleford> the mechanism for adjusting the height is where i'm stuck hah
[15:26:45] <RustyShackleford> I'm thinking I could anchor two metal tracks to a wall, then find a way to attach some heavy duty shelf brackets to that track in a way that I can securely lock it into place, but also move it freely when I want to adjust the height
[15:27:14] <robopal> https://www.myupdesk.com/
[15:27:22] <robopal> copy them
[15:27:33] <robopal> oh you want it wall mounted?
[15:28:00] <robopal> well thats ok too
[15:28:15] <RustyShackleford> freestanding is better. I'm thinking wall mounted would be easier to fabricate the parts I need
[15:28:37] <RustyShackleford> those updesks have hydraulics?
[15:28:59] <RustyShackleford> or is that like an electric motor? Either way, that looks harder to make myself on the cheap
[15:29:19] <robopal> electric lift mechanism
[15:31:19] <RustyShackleford> can I diy one of those?
[15:36:41] <RustyShackleford> so for the track design, a naive way to set the height would be to use pins. but that would be awkward to adjust, you could drop the desk
[15:37:46] <RustyShackleford> how could I create intervals where the desk would click into place?
[15:38:11] <RustyShackleford> also might want to use some bearings somehow to allow for smooth up and down movement
[15:40:10] <robopal> what is the travel? how much up and down do you want to move it?
[15:41:40] <RustyShackleford> 2-3 ft
[15:41:56] <RustyShackleford> 30" or so for sitting and um 50" or so for standing
[15:42:09] <RustyShackleford> damn i'm short lol.
[15:43:02] <RustyShackleford> more like 45" for an standing keyboard height
[15:45:07] <robopal> so it is just one setting, the normal position (down) and the up position
[15:51:23] <robopal> https://woodgears.ca/laptop_table/wheely_desk.jpg
[15:53:13] <robopal> ikea sells a sit/stand desk i think
[15:53:30] <robopal> http://www.ikea.com/us/en/catalog/products/S19022530/
[16:04:29] <RustyShackleford> i keep meaning to look at ikea's desks
[16:05:26] <RustyShackleford> I'm thinking more than just those two settings. Anywhere on the track ideally
[16:05:43] <RustyShackleford> two settings would be simpler though
