Hi all,
I'd like to ask: if I'm programming a continuous trajectory for a SCARA robot in the task space is it necessary this trajectory be a continuous trajectory also in the joint space?
for more understanding I'm uploading the trajectory for axes in the task space (yellow) and the same trajectory after inverse kinematics in the joint space (violet)
Ahmad,
I have never worked directly with a SCARA but did design and program a 6-axis manipulator. If I understand your plot correctly, you show 2 joint axes having 2Pi (360 deg) position discontinuities. I strongly suspect the position references provided to your controller must be continuous. Also, the references for each joint must be within the joint's allowable range of motion. The solution to your problem might be as simple as modifying your inputs by removing the 2pi step discontinuity. My logic is that a 2Pi step in a joint's position will cause the joint to return to the same physical rotational position as before the step.
I have no mathematical proof, but it should result in continuous trajectory unless you are passing by singularities of the Jacobian matrix. Please provide your controller, it is Jacobian based ?
The way I do it : X_dot = JAC*q_dot, where X_dot is the task space coordinates and q_dot for the joint space coordinates, then using the pseudoinverse you get q_dot, if you are working near singular configuration (at which the determinant of the JAcobian matrix JAC is equal to zero) you can use the damped pseudoinvers.
Thank, but this we expected. I found the problem which is the new version of MATLAB 2015a.
- Badges
- Science topic
- Similar topics
- Correction