Abstract
In feedback control of mechanical systems, various kinds of sensors are used, and usually each sensor has its advantages and disadvantages at the same time. For example, accelerometers are suitable for measuring relatively high frequency motion, and displacement sensors are suitable for measuring relatively low frequency motion. In order to utilize only advantages of each sensor, sensor-fusion technique is used in many fields. However, when the sensor fusion is used for feedback control problems, it requires both signal filters and feedback controller. The signal filers are used to generate physical values from various sensor measurements, and then the feedback controller uses these generated values for control. This paper proposes a concept of "Controller Fusion". Controller fusion is the way to design controllers which works as both signal filters and feedback controllers. Therefore the control system design process becomes simpler than the sensor fusion, and also good motion control performance is obtained in wide frequency range. In this paper, first the concept of "Controller Fusion" is described, then the experimental results are shown for a 3-dof electric shaking table, and effectiveness of the controller fusion is evaluated. Finally some discussions for future practical systems are related.
