Closed-loop Control System Based on Visual Measurement for LAMOST Optical Fiber Positioning

With the development of optical fiber positioning, the LAMOST (Large Sky Area Multi-Object Fiber Spectroscopy Telescope) system requires higher positioning accuracy. Despite the great success of open-loop control mode in previous techniques, it would be hard to meet the strict requirements of the next-generation optical fiber positioning techniques. In contrast to the open-loop system, in this paper, we focus on the closed-loop control of fiber positioning based on visual measurement. In our proposed framework, by measuring the actual position of the optical fiber positioner and feeding it back to the control system, we control the positioner to approach the target position multiple times to improve the accuracy of optical fiber positioning. Nevertheless, the challenges such as cross-border phenomenon and collision of the fiber positioner still hinder the accuracy of our designed system. To tackle the former challenge, we normalize the central axis angle of the positioner according to the algorithm we proposed, which alleviates the problem that the positioner cannot move successfully due to the cross-border phenomenon near the mechanical zero position in the central axis. Extensive experimental results show that closed-loop control consistently improves the positioning accuracy of the unit. After ignoring the lens positioning error, 95% fiber unit positioning error under our proposed closed-loop control method is less than 40um after three steps of approximation, 98% fiber unit positioning error is less than 40um after four steps of approximation, which meets the requirements of high precision fiber positioning. It is worth noting that our experiments also exhibit the insensitivity of the closed-loop positioning scheme to the initial unit parameters, verifying the satisfactory robustness of our proposed method.