基于改进LM算法的冗余机器人运动学参数标定

Kinematic Parameter Calibration of Redundant Robots Based on Improved LM Algorithm

机器人的定位精度是衡量机器人性能的重要指标,相较于重复定位精度可以从几何参数上提高几何参数,绝对定位精度还需对非几何参数进行校准。对于冗余自由度机器人,非几何参数求解更为复杂。在分析了LM算法的基础上,提出了一种基于改进LM算法的冗余机器人标定算法。首先,通过混合五参数MD-H方法建立机器人正运动学模型,采用速度级雅可比矩阵求解机器人运动学参数。实验过程中,采用深度相机对机器人末端位置进行测量。采用改进LM算法,根据测量结果对运动学参数进行校准。在改进LM算法中,通过变初始步长提高算法的适应性,并动态更新阻尼因子提高了收敛速度。改进后Quick-LM算法相较传统LM算法收敛速度提高46.7%,使机器人最大误差降低了9.18 mm,证明了改进后算法的有效性和实用性。

As robots are widely used in industrial production and human-robot collaboration,the positioning accuracy of robots has become an important indicator of robot performance.Compared with the repeatable positioning accuracy which can be improved geometrically from the geometric parameters,the absolute positioning accuracy also requires calibration of the non-geometric parameters.For redundant degree of freedom robots,the non-geometric parameter solution is more complicated.Based on the analysis of the LM algorithm,a redundant robot calibration algorithm based on the improved LM algorithm is proposed.Firstly,the robot positive kinematic model is established by the hybrid five-parameter MD-H method,and the velocity-level Jacobi matrix is used to solve the robot kinematic parameters.During the experiment,a depth camera is used to measure the robot end position.The kinematic parameters are calibrated according to the measurement results using the improved LM algorithm.In the improved LM algorithm,the adaptability of the algorithm is improved by varying the initial step size,and the convergence speed is improved by dynamically updating the damping factor.The improved Quick-LM algorithm increases the convergence speed by 46.7%compared with the traditional LM algorithm and reduces the maximum error of the robot by 9.18 mm,which proves the effectiveness and practicality of the improved algorithm.