Alignment and Operation for Microassembly Based on Microscopic Vision and Force Information

Abstract

This thesis investigates the common key techniques in the field of microassembly to contribute to the development and practicality of microassemblies in a social industrial economy. With support from the National Natural Science Foundation of China, we study the entire assembly procedure for two thin annular components. We investigate the ways to fully utilize the pose information extracted from multi microscopic vision to estimate the relative pose differences between components in three-dimensional (3D) space. We also explore how to effectively align components in the Cartesian space in six degrees of freedom (DOFs), how to effectively and rapidly insert one component into another while ensuring assembly precision, and how to dispense fluid in nanoliter scale for component jointing. The main work and contributions of this thesis are as follows.
First, an automatic assembly system is developed to realize the high-precision assembly of two millimeter-sized components with interference fit in 3D space with six DOFs. The system consists of a manipulator, an adjusting platform, a sensing system, and a computer. The manipulator consists of three translation DOFs and is employed to align the components in position. The adjusting platform consists of three rotation DOFs and one translation DOF and is used to align the components in orientation and to insert one component into another. The sensing system includes three microscopes, a force sensor, and a laser range sensor. The three microscopes are approximately orthogonally mounted to observe components from different directions. The force sensor is introduced to detect the contact force in the assembly process. The laser range sensor is utilized to measure the fluid dot altitude.

Second, we propose a comprehensive approach to realize nanoliter-scale fluid dispensing, with a dispensing control accuracy of 120 pL. The proposed nanoliter fluid dispensing approach contributes much to the formation of a solid connection among small components exhibiting expected mechanical dynamics. Our method fully utilizes fluid dynamics, the guidance of microscopic vision, and the feedback of a high-resolution laser range sensor. In the dispensing task, fluid is dispensed into target blind holes with a diameter of 300 μm. Exactly 21 nL of fluid is needed in one dispensing task. To precisely control the fluid volume dispensed, we develop an accurate time-pressure dispensing model for a specific dispensing prototype. We also measure fluid dot altitude with a laser range sensor to reflect the dispensed fluid volume. Experiments and results demonstrate that with the proposed dispensing method and system, a dispensing accuracy of 120 pL is achieved, and the fluid dot altitude can be controlled with an error range of less than 2 μm.

Third, we develop an integrated assembly procedure and a corresponding control strategy for the automatic precision assembly of thin annular components in 3D space. We propose the alignment approach for the first time in precision assembly. This proposed alignment approach decouples the alignment of millimeter-sized components into orientation alignment and position alignment. We also develop a robust method to fuse the pose information extracted from three orthogonally mounted microscopic cameras to reliably align the components. The insertion of the components is based on a force-guided controller, which can reliably realize the insertion task. Experiment results demonstrate that the proposed assembly approach and the correspondingly established assembly system can finish one assembly task within one minute with a success rate of 38/40.

Fourth, we conduct comprehensive research aimed at improving the efficiency of the traditional insertion force-guided insertion control method for the precision assembly of cylindrical components. Our proposed approach improves efficiency by up to four times. It has two inherent advantages relative to existing insertion methods in the domain of precision assembly. First, the proposed approach does not need preparatory actions dedicated to horizontal forces before insertion actions.

Specifically, it integrates the preparations into the insertion step. Second, instead of inserting with a fixed incremental depth, the proposed approach estimates the insertion depth for every action to be as large as possible with a probabilistic approach. The insertion of components is modeled as a stochastic state transition process with uncertainty described by a Gaussian distribution. The state transition function is well defined based on the analysis of historical assembly data and the universal mechanical properties of the components. An assessment function is also elaborately designed to evaluate the performance of the state transition function and thereby stimulate the control strategy to behave progressively or conservatively. Finally, the action to be taken for the current state is decided with iterative calculations in N step estimation. Experiments on the insertion of thin annular components demonstrate that with the proposed insertion control method, the insertion process can be finished within 30 steps. This duration is a significant improvement from the 120 steps required in traditional insertion control methods.

Finally, we present a summary of the results of the research, along with future work.

Date of Award	20 Jun 2017
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	You Fu LI (Supervisor)