利用空间机器人对高价值航天器开展在轨维护是合理使用轨道资源、提升航天器的经济效益、保障空间资产安全的重要手段之一，具有显著的经济和社会意义。空间机器人抓捕非合作目标后形成复合体系统，由于该系统的非线性、耦合性强，而且初始动量未知，操作过程约束条件多，给系统的稳定控制、参数辨识、轨迹规划等带来一系列难题。对于自旋失效卫星的捕获，尚未形成系统的研究，相关的理论和方法尚不完善。针对双臂空间机器人在轨抓捕任务，本文分别对开链式捕获和闭链式协同操作问题展开研究。针对非合作目标捕获后空间机器人基座姿态难以控制的问题，本文提出了双臂空间机器人自适应反作用零空间控制算法，该算法将反作用零空间控制由单臂空间机器人拓展到了双臂空间机器人，由合作目标的抓取拓展到了非合作目标的抓取。本文还首次将自适应滤波算法应用于双臂空间机器人，解决了大惯量非合作目标抓取后系统参数突变难以辨识的难题。通过设计变遗忘因子，明显提升了参数估计速率。解决了双臂空间机器人在轨捕获非合作目标的稳定控制难题。与传统单个机械臂相比，双臂机器人更能适应装配、搬运等复杂的操作任务。本文针对双臂协同操作物体的典型任务，提出了双臂协调操作的自适应混合阻抗控制方法。该控制方法可以使双臂机器人的建模控制不依赖于物体的形状、尺寸以及刚度等信息，从而扩大了双臂机器人操作物体的范围。利用机器人操作系统(ROS)，本文还构建了双臂协同操作控制实验平台，在该平台上对自适应混合阻抗控制算法进行了有效验证。针对含有不确定性的复杂任务，本文介绍了面向操作任务的双臂机器人控制策略。通过将复杂任务分解为关键子任务，实现了全局任务规划和精细运动控制，全局任务规划负责处理整体任务安排，处理确定性问题，精细运动控制根据实时反馈，处理不确定性问题。本文以拧水瓶的任务为例，详细介绍了任务级控制的特点，首次完整实现了抓取水瓶、拧开瓶盖、倒水并拧紧瓶盖的任务。从在轨服务的发展趋势来看，双臂空间机器人在轨捕获目标并开展服务是未来的研究热点。本文的研究成果将为空间机器人技术的发展提供参考。

With the development of astronautic technology, space robots have been playing an important role in space exploration. Therefore, many countries have paid significant attention to the development of space robotic technologies. In that context, this thesisinvestigates two problems of dual-arm space robot: the problem of stablization after capturing a non-cooperative target and the problem of cooperative manipulation.First, the problem of post-capure of a non-cooperative target by using a dual-arm space robot, is analyzed. The previously developed concept of Reaction Null Sapce control is expanded from a single-arm space robot to a dual-arm space robot. To reduce the uncertainties brought by the non-cooperative target, an adaptive filtering algorithm is developed. The proposed methods are applicable to a dual-arm space robot supplying on-orbit services.Next,to grasp and manipulate various objects, a new adaptive hybrid impedance control is proposed based on the master-slave structure. The new adaptive variablestiffness controller is only installed on the slave arm side, the master arm is responsible for the absolute motion trajectory. To verify the effectiveness of the algorithm, a dual-armrobotic system is set up with two UR5 arms and Robot Operation System (ROS).This thesis also give an overview of motion planning for dual-armrobot manipulation tasks, including motion generation and control strategies. Taking the bottle caping task as an example, the task-level control is divided into gross motion planning and fine motionplanning. The gross motion planning considers its overall global movement strategy, while the fine motion planning aims to deal with uncertainties during the task. This is the first time to accomplish the entire task of twisting off the bottle cap, pouring water, and tighten the cap back.From the development trends in on-orbit servicing, dual-arm robot will be one of the most promising approaches in future. This thesis presents a prospective application in this area.