With the development of spectral survey project, many large spectroscopic surveys are implementing closed-loop control fiber positioning system with a metrology camera system. The LAMOST improvement project plans to adopt a metrology camera system to update open-loop control to closed-loop control. To improve spectrum acquiring rate, the new generation of LAMOST being build is pursuing miniaturization and high-density of fiber positions, which takes more iteration time and is more likely to collide. Considering this situation, this paper presents a brand-new and real-time fiber positioning strategy with a fast fiber position recognition algorithm based on FPGA. A hardware acceleration platform is built on the main control board to process images of fiber positioners and calculate positions of each fiber every 17ms throughout the overall configuration. Then the main control board re-issues operation instructions to positioners based on the information, ensuring that positioners can update its running steps and speed in real time. When the offset of positioner’s actual position from target position is less than expected value (10um), the configuration is finished. This real-time control strategy can shorten configure time by positioning fibers in one time. And it can let two motors stop at the same time and avoid starting error, improving position accuracy. A prototype measurement platform was set up to verify the scientific validity of this strategy. After tests and validation in a laboratory environment, the results have shown that the time of fiber configuration is reduced by 1/5 and 100% of positioners can reach a precision of 10μm with great stability.
|