Abstract
For vision-based controlled flexible arm, the accuracy of the finite dimensional dynamic models and stability become very important for controller design. This paper designs suitable controller for a flexible arm with vision-based control. Input of the controller is the measured position error of the beam tip, which can be directly obtained by camera. In spite of the observed differences of results between actual system and simulated system, the results are good enough to validate the arm model. Simulation and experimental results show that the PID controller is acceptable. It is also stressed that two limiting factors, which are the camera sampling rate and a dead band in the velocity command, will influence the properties of the control system for vision-based controlled flexible arm.
