中文摘要：

针对空间机械臂系统的一系列特殊需求，提出了一种空间机械臂轨迹规划方法。假设机械臂在关节空间下存在一条可以用两段高次样条曲线分段描述、满足所有空间机械臂运动特性要求的理想轨迹。这两段样条轨迹之间的连接点参数能够影响两段样条函数在空间中的扭曲形状，从而使机械臂在遵从样条轨迹运动的同时避开所有障碍。首先建立理想轨迹的分段描述方程，将方程中的未知量用与中间点有关的参数进行描述，将不能够确定的中间点相关参数提取出来作为待定参数，从而将空间机械臂轨迹规划问题转变为一个多目标优化求解问题；其次应用遗传算法进行求解，在分析空间机械臂控制需求和障碍的笛卡儿空间描述的基础上，通过加权系数法建立关于笛卡儿空间机械臂末端轨迹长度、关节空间机械臂运动角度、运动过程中关节最大扭矩、机械臂总运动时间和碰撞情况的遗传算法适应度评定函数。最后应用遗传算法在关节空间下规划出一条无碰撞、动力学特性满足裕度要求、轨迹长度和运动时间较短的理想轨迹。另外，上述方法具有拓展性，通过多次分解轨迹求取中间点，能够使机械臂避开非常复杂的障碍。利用该方法在通过C语言建立的空间机械臂仿真平台上进行验证，结果表明该方法稳定、有效，生成的轨迹满足空间机械臂的性能要求。

英文摘要：

Meeting a series of special needs of space manipulator systems, a trajectory planning method for space manip-ulator is put forward. It is assumed that there is an ideal trajectory in joint space, which can be described by two sections of high-order spline curve and satisfy all kinetic characteristics of space manipulator. Parameters of the connection point between the two spline trajectories can influence the distorted shape of the two trajectories in space. Consequently, the ma-nipulator moves in accordance with the spline trajectories and in the same time avoids all the obstacles. Firstly, sectionalized functions of ideal trajectory are established. Unknown parameters in the functions are described using the parameters about the connection point of which the unknown parameters are extracted to be calculated. Thus the trajectory planning problem is transformed into a multi-objective optimization problem. Secondly, genetic algorithm （GA） is adopted to solve the problem. Based on the analysis on space manipulator control requirements and descriptions of obstacles in Cartesian space, a fitness function based on GA is defined by weighting coefficient method, with respect to the trajectory length of the end-effector in Cartesian space, the angles in joint space and the maximum torque during movement, the total movement time and the cases of interferences and collisions. At last, an ideal collision-free trajectory with short length and ideal movement time in the joint space is planned by using GA, and its dynamic features satisfies margin requirements. Additionally, the method proposed has wide expansibility. After decomposing trajectories and calculating connection points for many times, the manipulator can avoid obstacles in very complex circumstances. Simulation is accomplished on space manipulator simulation platform established by C language. And results show that the method is stable and effective, and the trajectory generated satisfies the special performance requirements of space manipulator.