Up to now, most robotic applications on assembly works have been accomplished under static condition. In that case, since it consists of two separate phases; assembly and transfer, the assembly task requires rat her long insertion time. Therefore, in order to reduce the task time, "DRAS(Dynamic Robotic Assembly System)" is proposed in which assembly and transfer are performed simultaneously. One of the major problems in designing such a system is to minimize the relative position and velocity errors between the end effector of a robot and a moving part.
In this thesis, a part tracking controller for DRAS is synthesized and the tracking performance is investigated through a series of experiments and simulations. Also the motion characteristics of the major components of which the DRAS consists are measured and discussed for analysis of the experimental data and the tracking controller design. In measurement of motion characteristics, a laser interferrometer is used and the experimental results show that the motion of industrial robot contains some fluctuations around the nominal trajectory. As a tracking controller, an optimal feedback plus feedforward controller is adaopted. The optimal controller is used for gain tuning of feedback controller. And the feedforward controller used for faster synchronization adaopts an observer which estimates unknown target velocity. From the results of experiment and simulation, it can be concluded that the feedforward controller adopted is very effective for faster synchronization. It can also be concluded that in DRAS the tracking system must follow low frequency term of target motion as possible, while the high frequency term must be compensated by active or passive assembly tool.
