Measurement and modeling of a flexible manipulator for vibration control using five-segment S-curve motion

User designed manipulators are widely used in industry as a part of automation. The design of lighter robotic arms is required for less energy consumption. Joints, structural features, and payload affect the dynamic behavior of manipulators. Even if the arms have sufficient structural rigidity, joints, or payloads further increase their flexibility. These factors should be considered at the design stage. Flexibility causes vibrations, and these vibrations negatively affect robot repeatability and processing speed. Reducing the vibration levels of flexible manipulators is an attractive issue for engineers and researchers. Accurate estimation of the mathematical model of flexible manipulators increases the success of vibration control. In this paper, the modeling and experiments for vibration control of a single-axis flexible curved manipulator with payload are considered. The experimental system is introduced to collect vibration responses synchronously at the tip of the curved manipulator for angular velocity input. The mathematical model of the manipulator is estimated using the continuous-time system identification (CTSI) method with a black-box model based on the experimental input/output (I/O) signals. A five-segment S-curve motion input based on the modal parameters is designed to suppress residual vibrations. Vibration control is successfully performed for different deceleration times of the designed S-curve motion input. The results showed that the residual vibrations from experiments and predicted models matched well for different cases depending on payload, angular position, and motion time.

Automated summary

This study focuses on vibration control of a flexible curved manipulator, estimating the mathematical model using payload, angular velocity, and vibration responses. An S-curve motion input is designed to suppress residual vibrations, successfully controlling vibrations for different deceleration times. Results show good matching between experimental and predicted models for various payload, angular position, and motion time scenarios.