Authors: Da Bao Lao, Wei Hu Zhou, Zi Li Zhang, Xue You Yang, Ji Gui Zhu
Abstract: To overcome the disadvantages of existing large-scale coordinate measurement systems about low level of automation and poor parallelism, this paper introduces a novel system which characters multi-sensor measuring network, while a single sensor measures angle via a couple of non-parallel planar laser beams rotating around a fixed shaft. The novel method for measuring angle improves automation in large-scale coordinate measurement. Furthermore, measuring network assures that measurement accuracy keeps to the same level even when the measuring range is enlarged. Measurement model and calibration are the key technologies of the system. They are the main sources of measuring errors. Therefore, to minimize the error, a practical global calibration method for the multi-sensor network system is proposed, with constructing a precise mathematical model. In this model, the parameters of the system are divided to inner and outer parts. While the inner parameters are obtained by aided instrument, the outer which means transform parameters between different sensors are figured out in spot calibration. Finally, the objective function based on transform coefficients between different sensor coordinate systems is defined and the outer parameters are optimized by bundle adjustment method. Thus, the whole system can finish global calibration by one time，and no medi-coordinate-frame is needed．Therewith，transformation time of coordinate frame decreases and calibration process is facilitated. Finally，measurement experiments are conducted with root mean square(RMS) error of not more than 0.1 mm.