0
$\begingroup$

I have a 6 DOF arm whose velocities I'm controlling as a function of force applied to the end effector. The software for the robot allows me to input either the desired end effector velocity or the desired joint angular velocities, which I know can be found using the inverse Jacobian.

Are there any benefits of using one scheme over the other? Would, for example, one help avoid singularities better? Does one lead to more accurate control than the other does?

$\endgroup$
  • $\begingroup$ Do you mean "the software... allows me to input either the desired end effector velocity or the desired joint angular velocities"? $\endgroup$ – Brian Lynch Oct 30 '15 at 19:21
  • $\begingroup$ One reason you might want to control within the end-effector space instead of joint-space is that it makes things more intuitive when you are planning your motions. Although this is probably more applicable to position control instead of velocity control, as well as being more important for pre-planned trajectories instead of reactive movements or open-loop control. $\endgroup$ – Brian Lynch Oct 31 '15 at 22:15
1
$\begingroup$

In the end, end effector speed is going to be determined by joint speed + kinematics, so I would choose whichever format you want that makes life easier on you.

Regarding singularities, there's no singularity caused by joint velocities, only by joint angular positions. So, if you're trying to avoid singularities but are only controlling based on velocity, then you'll probably need some separate controller or other method to "watch" the trajectory you're on to alter the specific combinations of joint velocities to avoid the singularity.

Regarding accuracy of control, again your end effector is located based on the combination of joint angles, so ultimately your joint angle positioning AND end effector positioning are both determined by your ability to encode, read, and control the joint angles. Again, I would go with whichever method is more convenient for you.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.