论文部分内容阅读
通过已知质心精确反解计算仿人机器人各关节的角度是一个经常遇到的问题。在双足行走,平衡控制等领域都很常见。但对于自由度高的仿人机器人系统,质心逆运算比较困难,尤其在双足支撑情况下,问题变为一个多自由度的并联机构,此时需要额外的约束和限制条件,使得计算非常复杂。本文基于Levenberg-Marquardt算法来解决复杂关节的逆解问题,研究在给定踝关节的情况下,用假定质心固定身体上的简化模型来使得真实质心逼近目标点,然后通过重复逼近缩小误差。我们通过NAO仿人机器人模型上的模拟验证了该算法实现了较高的准确性和计算效率。
It is a common problem to calculate the angles of the joints of humanoid robot by knowing the inverse of the known centroid. In the bipedal walking, balance control and other fields are very common. However, for high-freedom humanoid robotic systems, inverse centroid calculation is more difficult. Especially in the case of biped support, the problem becomes a multi-degree-of-freedom parallel mechanism. At this time, additional constraints and constraints are required, making the calculation very complicated . In this paper, the Levenberg-Marquardt algorithm is used to solve the inverse problem of complex joints. In the case of given ankle joint, the simplified centroid model with fixed centroid is used to approximate the true centroid to the target point, then the error is reduced by repeated approximation. We simulate the NAO humanoid robot model to verify that the algorithm achieves high accuracy and computational efficiency.