Back
[01:22:44] <Anniepoo_> worked on the remote comtrol this evening
[06:16:37] <Jak_o_Shadows> why do I have so many 220 Ohm resistors loose?
[06:16:45] <Jak_o_Shadows> the 470s I understad, but not the 220s
[10:20:32] <synja> jak, b/c there's less resistance holding them in place, silly
[11:03:17] <veverak> hmm
[11:03:24] <veverak> wanted to think again about moving arm
[11:03:28] <veverak> (or leg, in mine case)
[11:03:38] <veverak> I mean, I have 3DOF, so it's easy, but what if I had more?
[11:03:51] <veverak> and how to add effeciency optimalization, or speed optimlization
[11:04:32] <veverak> somehow, I ended up with big fucking graph of states the leg can be in and dijskra to move from state A to state B (where B does not have to be ocncerete state, can just be state that fits in requiremens)
[11:09:15] <z64555> IK be hard, yo
[11:10:40] <z64555> You could generatlize the IK so that the leg behaves like a 3DOF limb
[11:10:43] <veverak> each node is tuple of angles for all joints, each edge contains information how long it takes from one node to another OR how "costly" in terms of energy it is
[11:11:10] <z64555> azimuth and altitude (degrees) and radius
[11:11:15] <veverak> nah
[11:11:22] <veverak> z64555: that's not a problem or solution I would want ;)
[11:11:33] <z64555> no, for the hole leg
[11:11:34] <z64555> *whole
[11:11:41] <z64555> not for the individual links
[11:12:01] <z64555> This would be the interface for the driver which is operating on "go to this point"
[11:12:32] <veverak> ...
[11:12:35] <veverak> z64555: I know IK
[11:12:40] <veverak> and ITS NOT THE PROBLEM I AM SOLVING
[11:12:42] <veverak> ;)
[11:13:33] <veverak> oh, yeah
[11:13:43] <veverak> and each node in graph represents coordinates in (x,y,z)
[11:13:48] <z64555> there's not a whole lot of optimizations you can do, unless you make it under-actuated in terms of the controller
[11:14:00] <veverak> there is a lot you can do
[11:14:33] <veverak> "I need to move from (x,y,z) to (a,b,c), make sure that Z is > 'value' during the movement"
[11:14:40] <veverak> you can move it really fast, or in effecient way
[11:14:56] <veverak> and than with more DOFS there are even more possibilities in which you can do it
[11:15:08] <veverak> but only one of them should be the "fastest" or "most effecient"
[11:16:14] <z64555> k
[11:16:29] <veverak> graph + dijskra should work perfectly
[11:16:42] <veverak> but, that is big amount of calculation and you need precalcualted graph
[11:18:37] <z64555> what if you had the ability to control force output of the servos, in addition to angle?
[11:18:49] <z64555> That way, you can "relax" certain joints as you're moving
[21:42:26] <delinquentme> double micros_between_samples = 1.00 / 333.00 / 150.00;
[21:42:34] <delinquentme> anyone know why in ardino thats equal to 0.00 ?
[21:45:49] <z64555> could be underflow
[21:47:28] <z64555> Try using parens to isolate those divisions in the order that you think they should be
[21:47:45] <z64555> such as (1.00 / 333.0) / 150.00
[22:03:14] <stanford_ai> synja, are you actually working with multicopters?