Fast steering mirrors are used in many optical system applications, ranging from image tracking and scanning, to laser material processing, to jitter stabilization in optical communication. This paper presents a systematic approach to the modeling and feedback control design for a high performance fast steering mirror used for image scanning in a microscope. The overall approach consists of 1. definition of relevant image-based metric, 2. model identification, 3. simulation based controller optimization, and 4. experimental validation. The imaging application means that the performance metric is driven by the image quality rather than angular sensor output settling time. An output variance metric is shown to effectively capture the image quality and can be used in simulation for controller tuning. For controller development, a two-input/two-output mirror model is identified based on the small-input frequency response. In addition, amplifier nonlinearity and sensor and actuator noise models are also identified to ensure the overall model reflects the physical reality sufficiently closely to allow simulation-based controller optimization. Both Linear-Quadratic-Gaussian and Proportional-Integral-Derivative controller structures are considered. The controller parameters are tuned based on the simulated response and implemented in the experimental testbed to show the effectiveness of the proposed method.
6th IEEE Conf. on Automation Science and Engineering (CASE), Toronto, Aug, 2010.