Enhanced control of a flexure-jointed micromanipulation system using a vision-based servoing approach

8th TSME-International Conference on Mechanical Engineering

Chuthai, T.^1,2 ; Cole, M.O.T.^1,2 ; Wongratanaphisan, T.^1,2 ; Puangmali, P.^1,2

Department of Mechanical Engineering, Faculty of Engineering, Chiang Mai University, Chiang Mai

50200, Thailand^1

Center for Mechatronic Systems and Innovation, Chiang Mai University, Chiang Mai

50200, Thailand^2

关键词: Controller gain tuning;    Fast dynamic response;    Flexure-jointed mechanisms;    High-precision motion;    Micro manipulation systems;    Parallel kinematic mechanisms;    Pseudo-rigid-body models;    Vision-based measurements;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/297/1/012046/pdf
DOI  :  10.1088/1757-899X/297/1/012046

来源: IOP

【 摘 要 】

This paper describes a high-precision motion control implementation for a flexure-jointed micromanipulator. A desktop experimental motion platform has been created based on a 3RUU parallel kinematic mechanism, driven by rotary voice coil actuators. The three arms supporting the platform have rigid links with compact flexure joints as integrated parts and are made by single-process 3D printing. The mechanism overall size is approximately 250x250x100 mm. The workspace is relatively large for a flexure-jointed mechanism, being approximately 20x20x6 mm. A servo-control implementation based on pseudo-rigid-body models (PRBM) of kinematic behavior combined with nonlinear-PID control has been developed. This is shown to achieve fast response with good noise-rejection and platform stability. However, large errors in absolute positioning occur due to deficiencies in the PRBM kinematics, which cannot accurately capture flexure compliance behavior. To overcome this problem, visual servoing is employed, where a digital microscopy system is used to directly measure the platform position by image processing. By adopting nonlinear PID feedback of measured angles for the actuated joints as inner control loops, combined with auxiliary feedback of vision-based measurements, the absolute positioning error can be eliminated. With controller gain tuning, fast dynamic response and low residual vibration of the end platform can be achieved with absolute positioning accuracy within ±1 micron.