Robotic Gel Dispensing based on Visual Servoing for Fiber Threading

Abstract

Dispensing sessile droplets accurately on the top of the needle tip is important for applications such as fiber threading. In this paper, we propose an accurate online gel dispensing method to accurately dispense sessile droplets on top of the dispenser tip for highly repeatable fiber threading. We design a robotic gel dispenser based on positive-displacement piston dispensing that can accurately dispense sessile gel droplets of desired volume using visual servoing and adaptive model predictive control. An online gel volume estimation algorithm based on image processing is constructed to provide the estimation of the volume of the extruded gel droplet to the controller. To compensate for the nonlinear and time-varying process properties, the Adaptive Model Predictive Controller (MPC) adjusts its prediction model at runtime while satisfying a set of constraints. We employ the fiber threading technique based on impedance control with force tracking we developed recently to carry out the fiber threading experiments. To examine the benefits of the accurately dispensed droplet, multiple fibers threading experiments are conducted, where the repeatability of the fibers fabricated using the proposed methods are compared with two other methods: a) conventionally velocity regulation-based fiber fabrication, where the pulling force profile is not controlled, and b) fiber threading using impedance control with force tracking using a commercial time-pressure dispenser. The experimental results show that fibers fabricated using the proposed method have the highest repeatability based on the coefficient of variation of properties of the fabricated fibers, where the obtained coefficient of variation of the toughness, stiffness, elongation, and strength are 6.7%, 3.7%, 3.4%, 5.6% respectively.