空间近场完全非合作目标鲁棒相对位姿确定方法研究

The relative position and attitude determination for completely near-field non-cooperative spatial target is one of the key supporting technologies in several important space missions, such as space robot autonomous manipulation, spacecraft assembly on-orbit, space attack-defense, and multi-spacecraft formation flying. To meet the increasing requirements of task, this technology urgently need a breakthrough in the problem of the space stray light interference and the missing kinematic parameters such as the position of mass centroid and the inertia moment. Based on the in-depth analysis of polarized stereoscopic measurement gross error and kinematic parameter identification coupling of the target, through establishing the optimal joint identification and estimation mechanism of model, parameters and state, this project focus on the research of modeling and parameter calibration for polarized stereoscopic measurement gross error under space flare and shadow interference, and the optimal joint identification and estimation theory and analysis of the mixture of system parameters, model and state for the relative position and attitude determination for absolute near-field non-cooperative spatial target, and a fast robustness filter algorithm of relative position and attitude determination for the absolute near-field non-cooperative spatial target based on IMM-EM. It will also establish an actual measurement database to verify the effectiveness of the proposed method. Related outcomes will play a positive role in enriching the relative spacecraft navigation theory methods and improving the capability of spacecraft space near field measurement.

空间近场完全非合作目标相对位姿确定是空间机器人自主操控、航天器在轨组装、空间攻防、多航天器编队等重要航天任务的关键支撑技术之一。为了满足越来越高的任务要求，该技术迫切需要突破空间杂散光干扰以及目标质心位置、转动惯量等运动学参数缺失等问题。本课题拟在深入分析偏振立体视觉量测粗差与目标运动学参数辨识耦合机理的基础上，通过建立模式、参数与状态的最优联合辨识与估计机制，重点研究空间耀光及阴影干扰下的偏振立体视觉系统量测粗差建模与参数标定，面向空间近场完全非合作目标相对位姿确定的混合系统参数、模式与状态最优联合辨识与估计方法理论与分析，基于IMM-EM的空间近场完全非合作目标鲁棒相对位姿确定快速滤波算法，并建立实测数据库对提出方法有效性进行验证。相关成果对丰富航天器相对导航理论方法，提高航天器空间近场测量能力具有积极促进作用和研究价值。

空间近场完全非合作目标相对位姿确定是空间机器人自主操控、航天器在轨组装、空间攻防、多航天器编队等重要航天任务的关键支撑技术之一。为了满足越来越高的任务要求，该技术迫切需要突破空间杂散光干扰以及目标质心位置、转动惯量等运动学参数缺失等问题。本课题在深入分析多波段偏振立体视觉量机理的基础上，采用多波段偏振对杂散光进行检测与分离，进而利用色彩相关性正则约束补充丢失信息，从而实现对杂散光的抑制；引入鲁棒多模型滤波器，设计面向量测野值的IMM滤波算法；针对空间完全非合作目标缺失质心位置及转动惯量等先验信息，引入光流量测信息，获得目标上若干特征点的图像位置和速度信息，结合非合作目标的相对运动信息，解算出非合作目标的相对位置、姿态、线性速度、角速度、转动惯量和特征点的结构。相关成果对丰富航天器相对导航理论方法，提高航天器空间近场测量 能力具有积极促进作用和研究价值。