How to avoid this kind of isolation in controller?

[jc-bao]

After increasing the

[jc-bao]

Control objective: 1rad/s for 1s and -1 rad/s for 1s.

Only control with Kd: cannot deal with object torque

Control with PI (KP=4e2,KI=2e4): performant normal in the beginning but isolated later.

Smaller KI

Choose a smaller KI -> does not help :question:

KI*=0.3	KI*=0.1

Larger KP

Choose a larger KP (*=3.0)

Make both KI KP smaller

Choose a smaller KP(x=0.3) and a smaller KI (x=0.3):

Larger KD

Choosing a larger KD =0.3 :arrow_forward: does not help to reduce isolation.

:grey_question:

• Why would isolation happen in the second step command?
• Is this the fundamental limitation of the PID controller?

[jc-bao]

Final solution

This problem is caused by the high-frequency isolation caused by the spring-based rope. By changing the spring coefficient to a smaller value to make sure the control frequency is higher than the rope isolation frequency, the problem is resolved.

This comes at the cost of a softer rope.