空间机器人对大尺寸非合作航天器的超近距离视觉测量方法研究

The requirements for the near-range measurement on orbit are accuracy, reliability and in real-time in the approaching and berthing missions of space robots to non-cooperative target. Based on the research of the large size, non-cooperative targets, this project focuses on the study of the key technologies of on-orbit pattern recognition，relative position and attitude measurement to targets in near-range. The main contributions of this project are as follows: 1)The object selection and pattern recognition methods are proposed to the non-cooperative targets. 2) The pattern recognition and the relative motion determination method are proposed to the rotational objects based on the corner, line features. 3) A new visual method using two collaborated cameras for space robots is proposed to determine the position and attitude of the non-cooperative target at the final mission stages. When one or two feature points of the target are respectively recognized by them and when totally three feature points are recognized, the relative pose and attitude information of the target will be obtained by using the extended P3P algorithm. 4) The solver and the stability of the results for the extended PnP problem are proposed. The methods of this project will be estimated and evaluated by the simulation system and physical system. The method may put forward the key approach to realize the measurement of the large size non-cooperative target at the "Last One Meter" distance for space robot.

空间机器人对非合作目标的接近和停靠任务，对天基近距离测量提出了高精度、实时性和高可靠性的要求。本课题以大尺寸、非合作飞行器为研究对象，开展近距离对非合作目标的目标识别、位置姿态测量的关键技术研究。主要研究内容包括：1) 非合作目标测量对象的选择与识别，即在待服务的目标上寻找并识别合适的测量对象并进行识别；2）提出接近过程中基于角点和直线等特征的卫星形状特征识别方法，用以在目标的姿态为旋转情况下的相对运动参数获取；3）提出一种新的双目协作视觉测量方法，建立视觉测量模型，完成在最低3个特征点的条件下对目标的相对运动测量；4）对扩展PnP问题进行求解算法及解的稳定性研究。测量方法将在仿真系统和气浮台实验系统上进行验证。课题所提出的新视觉测量方法为突破空间机器人与大尺寸非合作目标的"最后一米"范围内的测量约束问题提供了关键性的技术途径。

超近距离相对位姿视觉测量是在轨服务任务中的一项关键技术，是对目标航天器进行交会接近、抓捕或对接等操作的前提条件。申请人在基金委的资助下，开展了空间机器人对大尺寸非合作航天器的超近距离视觉测量方法研究。具体包括以下几个方面：.1）根据飞行器的特点，研究了非合作目标测量与抓捕对象的选择与自主识别的理论和实际问题；.2）针对非合作目标上比较常见的矩形特征，提出一种基于双目立体视觉和协作视觉的交会接近相对测量方法；.3）针对非合作目标上比较常见的圆形特征，提出一种基于双目立体视觉的交会接近相对测量方法；.4）提出了一种扩展PNP测量的新方法，来解决近距离对非合作目标的位置和姿态进行测量时存在的问题；.5) 提出了一种基于结构光视觉的超近距离大目标相对位姿测量方法；.6）提出了一种较为经济实用、合理可行的半物理仿真实验方法来验证基于双目视觉的近距离交会接近关键算法；.7）提出了一种基于粒子滤波-支持向量机的ToF相机测量误差校正方法。