RoboSavvy Forum

Re: Inverse kinematics

2008-06-17T10:29:27+01:00

StuartL wrote:
Out of curiosity...
How many people here have developed (or are developing) inverse kinematics for the Bioloid kits? How far did you go? Are you using an alternative to the Cm-5?

At the University of Plymouth we have developed forward kinematics for the robot and some simplified inverse kinematics for the legs.
We are trying to keep the motion editor and flash pages but also integrate
kinematics and the HUV-IMU unit.

Here is a table of the forward kinematics

Denavit-Hartenberg Table for the Bioloid Robot

Code:: Link α d θ x M7 π/2 0 0 30 M8 π/2 0 0 30 M9 -π/2 0 π/2 0 M10 π/2 0 π/2 0 M11 -π 75 0 0 M12 -π 75 0 0 M13 0 75 0 0 M14 0 75 0 0 M15 π/2 0 0 0 M16 -π/2 0 0 0 M17 π 32 0 0 M18 π 32 0 0

M18 is the link corresponding to floor-to-servo 18
Post multiply M18,M16,M14,M12,M10,M8 to
get the transformation from floor to hip.
Dimentions are in radians and millimeters

Kinematics of an industrial robot often described with homogenous 4x4 transformation matrices. If you lost, what this table means look up "Denavit-Hartenberg" in google.

regards,
Joerg

Statistics: Posted by joerg.wolf — Tue Jun 17, 2008 10:29 am

Inverse kinematics

2008-05-17T14:53:59+01:00

I've developed inverse kinematics for a (slightly customized) humanoid and a custom-built hexapod. Since I do a lot of floating point calculations requiring inverse kinematics (like interpolation, motion planning, ...) and a whole bunch of other complex stuff (like forward kinematics, visualization, scripting language interpretation, ...) I've chosen to implement the inverse kinematics in the pc program (which communicates with a cm-5 controller). The cm-5 therefore has only to do simple calculations (like linear interpolation).

It's all included in my project mentioned in this post: http://robosavvy.com/forum/viewtopic.php?p=15449.

Statistics: Posted by cosa — Sat May 17, 2008 2:53 pm

Inverse kinematics

2008-05-16T20:34:39+01:00

Out of curiosity...

How many people here have developed (or are developing) inverse kinematics for the Bioloid kits? How far did you go? Are you using an alternative to the Cm-5?

I'm most of the way there, now, and I'm genuinely surprised with how little processing power it's taking. It seems that waiting for even one dynamixel to reply to a get position packet (even with the return delay being set to one) takes longer than doing all the maths on the stock CM5.

I, of course, know of Jon's work and Rook's Pawn III. I'm curious who else is going beyond pre-programmed sequences.

Statistics: Posted by StuartL — Fri May 16, 2008 8:34 pm