面向立方星的相对位姿测量视觉传感器设计

Vision sensor design of relative pose measurement for CubeSats

针对立方星在轨服务任务中的近距离相对导航问题,提出一种相对位姿测量视觉传感器设计。为应对空间复杂的光照条件,设计一种具有多层结构的立体靶标,选取波长为850nm的LED灯作为立体靶标的光源。在相机镜头处安装850nm的红外窄带滤镜以提高立体靶标成像质量。为提高视觉传感器的测量精度,提出一种迭代直接求解(iterative direct solution,IDS)的相对位姿估计算法。该算法直接利用立体靶标上的非共面点提供的深度信息获得初始相对位姿,然后引入Haralick迭代算法来优化初始测量结果。搭建六自由度实验平台对该视觉传感器进行性能测试。实验结果表明,该视觉传感器能克服背景光的干扰,且提出的IDS相对位姿估计算法的精度优于经典的P3P算法和EPnP算法。

Aiming at solving the close-range relative navigation problem for on-orbit serving(OOS)missions of CubeSats,a vision sensor design for measuring the relative pose was proposed.In order to circumvent the complex light conditions in space,a multi-layer stereo marker was designed.The LEDs with a wavelength of 850 nm were selected as the illuminators of the stereo marker.An infrared narrow band-pass filter that only let 850 nm length waves pass was added in front of the camera lens to improve the imaging quality of the stereo marker.An iterative direct solution(IDS)was derived for the relative pose estimation.This algorithm used the depth information provided by the non-coplanar points in the stereo marker to obtain an initial relative pose.The Haralick iterative algorithm was then introduced to optimize the initial relative pose.An experiment platform with six degree-of-freedom was built to test the performance of the vision sensor.The results show that the vision sensor can overcome the interference of ambient light.The accuracy of the proposed IDS algorithm for pose estimation is better than those of the classical P3P algorithm and EPnP algorithm.