Kinematic error model plays an important role in the kinematic calibration of robot manipulators. In order to guarantee the best calibration result, the error model should meet the requirements of completeness, minimality and continuity. For the conventional serial robots, these issues have been intensively studied and a consensus has been reached during the past decades. However, for parallel manipulators, the problem of minimality, namely the determination of identifiable parameters mainly relies on numerical studies. There is a lack of effective methods to analyze the parameters’ identifiability in the error modeling process. Besides, there is still no agreement on the formula of the maximum number of identifiable parameters. In our previous work, a complete, minimal and continuous error modeling approach for parallel manipulators has been proposed based on product of exponential (POE) formula and the equation that the maximal number of identifiable parameters $4r+2p+6$ was proved. However, the correctness and effectiveness of this method in kinematic calibration have not been verified on any parallel manipulator or robot. Therefore, in this paper a kinematic calibration is conducted on a prototype of a 3-PRRU parallel manipulator based on the proposed error model. The method of validating the completeness and minimality are respectively given by numerical simulations. Furthermore, to illustrate the advantage of the proposed error model, the kinematic parameters of the prototype are also calibrated based on the conventional error model which is derived by means of differentiating the limbs’ inverse kinematics. By comparing the calibration results, it can be concluded that the error model satisfying the three criteria can better improve the positioning accuracy. In addition, since the calibration method is described uniformly, it can be applied to most parallel manipulators by referring to the calibration procedure this paper presented.

运动误差模型在机器人操纵器的运动学标定中起着重要作用。为了保证最佳的校准结果，误差模型应满足完整性，最小性和连续性的要求。对于传统的串行机器人，在过去的几十年中对这些问题进行了深入研究，并已达成共识。然而，对于并联机械手，最小化问题，即可识别参数的确定主要依赖于数值研究。在错误建模过程中，缺乏有效的方法来分析参数的可识别性。此外，关于可识别参数的最大数量的公式仍未达成共识。在我们之前的工作中，$4\mathrm{[R}+\mathrm{2个}p+6$被证明。但是，这种方法在运动学标定中的正确性和有效性尚未在任何并联机械手或机器人上得到验证。因此，在本文中，对3- P的原型进行了运动学校准基于提出的误差模型的RRU并联机械手。数值模拟分别给出了验证完整性和最小性的方法。此外，为了说明所提出的误差模型的优点，还基于传统的误差模型对原型的运动学参数进行了校准，该传统的误差模型是通过区分肢体的逆运动学而得出的。通过比较校准结果，可以得出结论，满足三个标准的误差模型可以更好地提高定位精度。另外，由于对校准方法进行了统一描述，因此通过参考本文介绍的校准程序，它可以应用于大多数并联机械手。