【作者】 曾岑；

【导师】 魏小鹏；

【作者基本信息】 大连理工大学 ， 机械设计及理论， 2013， 博士

【摘要】 机械臂在空间在轨服务过程中起着非常重要的作用,系统地应用空间机械臂能提高宇航员的工作效率并能有效地节省费用,而自由漂浮空间机械臂由于其耗能低,寿命长,近些年来逐渐成为空间在轨服务研究的重点。本文以两项国家863科研项目为依托,主要研究空间机械臂的运动学建模问题、空间机械臂的运动规划、基于在轨服务的多目标任务规划、避障规划以及空间机械臂综合任务规划,并应用Orbiter系统进行仿真研究。本文的主要内容为：(1)调研了国内外空间机械臂的发展情况,介绍了空间在轨服务的任务体系及典型任务,最详细阐述了近年来国内外的空间机械臂理论研究的重点难点问题。(2)基于四点假设,建立了空间机械臂的坐标系统。对广义雅克比矩阵及虚拟机械臂的建模方法进行分析,针对目前建模方法的计算量大、系统模型不直观的问题,提出了基于互相映射的坐标系方法。(3)进行了自由漂浮空间机械臂运动规划研究,分别包括：关节空间点到点运动规划、笛卡尔空间点到点的路径规划和笛卡尔空间连续性路径规划。(4)根据当前空间机械臂的任务追踪过程以及在轨服务任务的复杂情况,建立了空间机械臂多目标在轨任务模型。结合邻域搜索算法利用非线性分段优化算法解决空间机械臂的多目标任务的问题。(5)建立了不同种类障碍物的包围盒及碰撞检测算法,使用位姿空间分段搜索的方法解决了机械臂与服务星和目标星的碰撞问题,提出了基于遗传算法的全局性避障及任务规划方法。(6)应用Orbiter系统对空间机械臂的整体任务规划进行了仿真和验证,并模拟了在轨服务观测的摄像头视角,更直观地仿真了空间机械臂系统的在轨服务任务。本文的主要创新工作如下：(1)提出了建立互映射空间坐标系的方法,改进了广义雅可比矩阵的建模方法,使得模型更直观,有效地降低了广义雅克比矩阵方法计算复杂度。(2)在基于笛卡尔空间的点到点运动规划方法中,利用基于六次多项式样条函数的双参数法规划关节的角速度变化曲线,结合牛顿法和割线法对关节参数迭代,提高了迭代效率,得到的空间机械臂运动轨迹满足任务要求。(3)在空间机械臂的连续性路径规划方面,为避免连续性路径过程中动力学奇异问题,基于分段进化算法,将连续性路径规划转化为点到点路径规划,解决了连续性路径规划不连续可微和动力学奇异点的问题,增加了机械臂规划轨迹的灵活性。(4)提出了遗传进化算法与包围盒碰撞检测算法、分段算法相结合的方法解决空间机械臂的避障规划与任务规划,可解决避障规划中所产生的对基座扰动大且耗能过多的问题。更多还原

【Abstract】 Manipulators play an increasingly important role in the on-orbit service(OOS) due to its efficiency and lower cost. Studies on OOS are gradually focusing on free-floating space manipulator attribute to the lower fuel-consuming and longer life.Five key technologies including the kinematic modeling, motion planning, multi-objects task planning, obstacle-avoiding and comprehensive planning of space robot are studied in this paper supported by National863research projects, moreover, the results are simulated by Obiter to verify the methods proposed in this paper.The contents are as follows:(1) The development of space manipulator at home and abroad is investigated and space mission architecture and several typical tasks are introduced in the paper. The focal points of all kinds of space manipulator theories are described in detail.(2) Based on four assumptions, the space manipulator system is established, the generalized Jacobi matrix method and virtual manipulator theory are analyzed. In addition, A mutual-mapping method is proposed in the paper in order for less computation complexity and intuitive model.(3) Path planning algorithms of free-flowing space manipulator is studied in the paper, including point-to-point motion planning in joint space, point-to-point motion planning in Cartesian space and continuous path planning of space manipulator based on Cartesian space.(4) Based on the complexity of task tracing and on-orbit servicing, the on-orbit multi-objects task model is established, the non-linear subsection optimal algorithm is combined with the neighborhood searching algorithm on the multi-object task model to solve the multitask problems.(5) Different obstacles bounding boxes are established and the collision detection algorithm is proposed in the paper. Then, the obstacle-avoid problem is handled by position-attitude space subsection algorithm. Finally, a global obstacle-avoidance algorithm which based on Genetic Algorithm is proposed.(6) Eventually, All of the simulation results are demonstrated using Orbiter software, and a camera viewpoint mounted on the end-effector is simulated.The innovations are as follows: (1) A mutual-mapping approach is prpposed to advance the traditional generalized Jacobi matrix method. The results shows the more intuitive model and the less computational complexity.(2) In Cartesian path planning part, a dual-parameters method is proposed for the manipulator path planning., the iterative calculation of Newtown and Secant algorithm is used for high efficiency.(3)In the continuous path planning part, a subsection evolution algorithm is proposed to solve the general continuous path tracking problems, and also the non-differentiable and non-continuous path problems, the approach can also obtain the more flexible path.(4) The Genetic Algorithm is combined with the collision detection algorithm and the subsection evolution algorithm together because of the large base disturbance and more consume.更多还原