Abstract It is a big challenge to ensure the low energy consumption and stability of the hybrid mechanism under the dynamic load condition, and it is more obvious in the series-parallel mechanism. In this paper, a 6 degrees-of-freedom (DOF) serial-parallel hybrid humanoid arm by constituting the distribution of “2-1-3” is constructed, and a dynamic load distribution optimization method is studied systematically. The kinematics and dynamics model of the humanoid arm are established by Denavit–Hartenberg (DH) method, Lagrange formulation and the principle of virtual work which can clearly show the coupling characteristics between the components, respectively. Meanwhile, the optimal space point has been attained by a genetic algorithm via the dynamic performance evaluation index and force mapping performance evaluation index are derived based on the dynamic model, and the comprehensive performance evaluation index is defined by the weighted summation method which convert multiple comprehensive performance evaluation model into a single-objective model. Furthermore, the position and posture trajectory of the best performance and the optimal comprehensive objective function are solved by a genetic algorithm, which enables the robot to complete the movement with the lowest energy consumption, the fastest and smoothest way, and the feasibility of this method is verified by numerical example. The work laid a theoretical foundation for the engineering application of the humanoid arm.

中文翻译：

---

串并联混合仿人手臂的动态载荷分布优化

摘要 保证混合动力机构在动态负载条件下的低能耗和稳定性是一个很大的挑战，在串并联机构中更为明显。本文通过构造“2-1-3”的分布构造了一个6自由度（DOF）串并联混合人形手臂，并系统研究了动态载荷分布优化方法。采用Denavit-Hartenberg(DH)方法、拉格朗日公式和虚功原理建立了仿人手臂的运动学和动力学模型，分别可以清晰地显示部件之间的耦合特性。同时，基于动力学模型推导出动态性能评价指标和力图性能评价指标，通过遗传算法得到最优空间点，并通过加权求和法将多项综合性能评价转换为综合性能评价指标。模型转化为单目标模型。进而通过遗传算法求解出性能最佳的位置姿态轨迹和最优综合目标函数，使机器人能够以最低的能耗、最快、最流畅的方式完成运动，该方法的可行性通过数值例子验证。该工作为仿人手臂的工程应用奠定了理论基础。综合绩效评价指标采用加权求和法定义，将多个综合绩效评价模型转化为单目标模型。进而通过遗传算法求解出性能最佳的位置姿态轨迹和最优综合目标函数，使机器人能够以最低的能耗、最快、最流畅的方式完成运动，该方法的可行性通过数值例子验证。该工作为仿人手臂的工程应用奠定了理论基础。综合绩效评价指标采用加权求和法定义，将多个综合绩效评价模型转化为单目标模型。进而通过遗传算法求解出性能最佳的位置姿态轨迹和最优综合目标函数，使机器人能够以最低的能耗、最快、最流畅的方式完成运动，该方法的可行性通过数值例子验证。该工作为仿人手臂的工程应用奠定了理论基础。通过遗传算法求解出性能最佳的位置姿态轨迹和最优综合目标函数，使机器人能够以最低的能耗、最快、最流畅的方式完成运动，验证了该方法的可行性通过数值例子。该工作为仿人手臂的工程应用奠定了理论基础。通过遗传算法求解出性能最佳的位置姿态轨迹和最优综合目标函数，使机器人能够以最低的能耗、最快、最流畅的方式完成运动，验证了该方法的可行性通过数值例子。该工作为仿人手臂的工程应用奠定了理论基础。