idea for Inverse Kiematics of Robotic arm

[Himanshu]

I would like to know of any fresh idea for a robotic arm Inverse Kinematics.I have given the details in my last posting.

[Greg_McKaskle2]

> I would like to know of any fresh idea for a robotic arm Inverse
> Kinematics.I have given the details in my last posting.

I haven't dealt with this since college, but it is very common for there
to be an infinite number of ways for a robotic arm with several degrees
of freedom to place its actuator at a given point in space. Typically,
the solutions are further limited by considering not only the x,y,z
placement of the actuator, but also the orientation of it.

In other words, if the arm can move in three dimensions, then where is
the actuator pointing when it is at that point in space. You also have
to take into account that the motors and or joints often have limits
placed on their range, which will eliminate some solutions, and the arm
segments cannot intersect with one another, which will eliminate others.
This often leaves you with a smaller, but still infinite set of
solutions to choose from.

To choose between these, it is often useful to know the other properties
of the motors. Sometimes the motors will have a calibrated angle they
can move to that will have the least amount of error. Let's say it is
zero degrees. By alternately adding conditions such that the various
joint positions are set to zero, we can add additional constraints until
you come up with a finite set of solutions. In almost all cases, you
will have more than one solution, if no other reason than you have the
mirror image of the arm. Choosing between these is often just a matter
of choosing the solution that is closest -- getting the arm there the
quickest.

As for moving along a path, you can choose your solutions based upon
moving the minimum number of segments, speed, accurracy, torque, or
other qualitative solutions that you don't currently have equations for.

In otherwords, when the robot arm begins to move, things get much more
complex and the dynamics come into play.

If you want my suggestion, try setting all angles to a fixed angle where
the motor positioning is most accurate and looking for a solution, then
remove a constraint at a time until you find a solution to the set of
equations. Ordering which constraints you relax first should be based
upon the arm's motor characteristics. When you find a solution, there
are almost certainly others that are mirror images of the arm. You can
then try to add constraints back to the set to select which solution you
are interested in. Or use optimization techniques to maximize or
minimize some characteristic such as speed, or accurracy.

Greg McKaskle

[Zvezdana S.]

Details are in this posting:
http://exchange.ni.com/servlet/ProcessRequest?RHIVEID=101&RPAGEID=135&HOID=5065000000080000009C5F0000&UCATEGORY_0=_49_%24_6_&UCATEGORY_S=0

If it is the same issue, you can use the same post in the future.

Zvezdana S.