Research On Key Technologies Of Microassembly Robot System

The microscope visual servoing approach and its key technology of micro-assembly robot are very important to improve the efficiency of assembly and ensure the accuracy of assembly. The dissertation mainly aims to develop a robotic system for sub-millimeter size microassembly. Some key technologies of microassembly robot system have been investigated, including micro-assembly system platforms, the microgrippers, the multi-target classification and identification, target tracking with micro vision, visual delay analysis and microscopic vision control et al.Many kinds of microgrippers have been developed for handling and manipulating micro-sized objects in the fields of various applications. However, the micro-force sensing is still one of the most troublesome problems to improve the reliability of manipulation. To circumvent the problem, in this paper, two kinds of microgrippers, which are driven by two PZT bimorphs and the vacuum generator, are designed for the requirement of microassembly. In order to obtain the micro force, two kinds of sensors are introduced sensing the signal because of the strain of the cantilever, which are resistance strain gauge sensor and polyvinylidene difluoride (PVDF) sensor. The microgrippers can automatically pick and release the object by means of the PC monitoring system.Multi micro parts recognition is one of the important tasks for the assembly of multi micro objects in microassembly. We propose a polar coordinate based algorithm for the computation of Zernike moments, which improves the invariance properties dramatically. Due to the symmetry property of the Zernike basis functions, Zernike moments can be obtained by computing only one sixteenth of the Zernike basis functions, which means that the number of pixels involved in the computation of Zernike moments is only6.75%of previous method. This leads to significant reduction in the computational complexity requirements. To achieve multi micro parts recognition with higher performance, we present a support vector machine algorithm, which employs polar Zernike moments based on edge extraction to obtain feature attribute to identify and classify the targets. The obtained results show the superiority of the proposed method.Visual servoing has been around for decades, but the time delay is still one of the most troublesome problems to achieve target tracking. To circumvent the problem, in this paper, the Kalman filter is employed to estimate future position of the object. We present a current statistical model for moving target. An improved fuzzy adaptive Kalman filter, which is evolved from the Kalman filter, is put forward based on the current statistical model. The results show that the modified adaptive filter can improve the ability of maneuvering target tracking.In order to introduce the improved fuzzy adaptive Kalman filter, the accurate time delays, which include the processing lag and the motion lag, need to be obtained. Thus, the delays of the visual control servoing system are discussed, and a generic timing model for the system is provided. Based on the timing model, the improved fuzzy adaptive Kalman filter is employed to predict manipulator and target motion. We build the visual servoing system construction based on fuzzy adaptive Kalman filter. According to the characteristics of micro visual servoing, design the servoing path of XY-plane and YZ-plane and establish the control structure of three-dimensional visual servoing. Finally, a variable structure control law for micromanipulation is presented. The results show that the proposed adaptive Kalman filter can improve the ability of moving target tracking based on visual servoing.Finally, we analyze a typical structure of micromanipulation firstly, and describe the hardware system and software system of the microassembly system we have built. A series of experimental results show the reliability of the microgrippers, the polar Zernike moments based multi micro parts recognition system and the improved fuzzy adaptive Kalman filter based visual servoing system, which are mentioned above.