Dynamixel backlash
Dynamixel backlash
by HyperCube
Fri Mar 02, 2012 7:20 am
Does anyone know which dynamixels have the least backlash?
I'm considering upgrading axis 1 of a manipulator, it's currently an AX-18F which doesn't have quite enough torque. I have an EX-106 I could use, but it has some pretty hardcore backlash. I really need the positioning to be accurate (0.3º resolution is OK, but something like 0.1º would be better). The arm has both TTL and 485 buses running to it so it can use either bus.
Do the RX-64's have just as much backlash as the EX-106's? The new MX-28's look promising with a resolution of 0.088º. If these have very little backlash this would be perfect...
I'm considering upgrading axis 1 of a manipulator, it's currently an AX-18F which doesn't have quite enough torque. I have an EX-106 I could use, but it has some pretty hardcore backlash. I really need the positioning to be accurate (0.3º resolution is OK, but something like 0.1º would be better). The arm has both TTL and 485 buses running to it so it can use either bus.
Do the RX-64's have just as much backlash as the EX-106's? The new MX-28's look promising with a resolution of 0.088º. If these have very little backlash this would be perfect...
by HyperCube
Sat Mar 17, 2012 2:30 am
I just noticed that some of the bolts in the EX-106's were slightly loose, which was contributing to some (but not all) of the play. What a rookie mistake, lol! I think the remaining cause of play in the servos is backlash, without opening them up tho it's hard to tell.
I'll be really keen to hear about how much play there is in the latest line of Dynamixels, I guess people haven't had much of a tinker with them just yet to let me know.
I hope my post above hasn't scared anyone off from buying high-end Dynamixels. Overall, these servos really are something special.
I'll be really keen to hear about how much play there is in the latest line of Dynamixels, I guess people haven't had much of a tinker with them just yet to let me know.
I hope my post above hasn't scared anyone off from buying high-end Dynamixels. Overall, these servos really are something special.
Backlash in walking
by Compadre
Sat Mar 17, 2012 8:30 am
Good night.
Backlash is a hardcore problem for the issue of humanoid walking. When passing from double support (that might be viewed as an parallel robot, which augments accuracy) to single support I get near 6 - 12 Deg of play in the final element (pelvis).
This is for slow motion. When moving fast something happen that it is possible to compensato only with around 3 Deg.
I do this just by inmeddiately adding an extra amount of Degs in ankle knee and hip when passing from double to single support (parallel to not parallel robot).
Then I visually check if the motion looks like a the one that one can expect for REAL rigid robot (perfectly inelastic joint).
Even if elastic joint modelling is a treated subject, backlash shall be aswell and the issues that arises from play in robots motors (joints) are, in my oppinion manneageable.
Backlash is a hardcore problem for the issue of humanoid walking. When passing from double support (that might be viewed as an parallel robot, which augments accuracy) to single support I get near 6 - 12 Deg of play in the final element (pelvis).
This is for slow motion. When moving fast something happen that it is possible to compensato only with around 3 Deg.
I do this just by inmeddiately adding an extra amount of Degs in ankle knee and hip when passing from double to single support (parallel to not parallel robot).
Then I visually check if the motion looks like a the one that one can expect for REAL rigid robot (perfectly inelastic joint).
Even if elastic joint modelling is a treated subject, backlash shall be aswell and the issues that arises from play in robots motors (joints) are, in my oppinion manneageable.