Absolute kinematic calibration of six-axis serial industrial robots using relative measurements

标准的、未校准的工业机器人固有的定位精度不足是其在制造过程自动化中使用的限制因素，例如外壳直接公司(EDI)的制造过程。用于制造的机器人通常被设计为具有出色的重复性(由控制器中存储的一组关节角度定义的连续达到相同位置和方向或姿势的能力)，但缺乏精确度(移动到必须由控制器计算关节角度的姿势)。为了克服精确度的缺点，机器人通常被用于“记录和播放”模式，即由技术人员手动示教姿势，并在长时间的生产过程中重复。在处理较短的定制产品生产批次时，例如在EDI提供3000多种不同产品的情况下，手动教授所有必需的姿势是不可行的，而这些姿势是基于零件尺寸或CAD模型计算的。这种方法要求对机器人进行标定，以满足精度要求。现有的运动学标定方法需要机器人末端执行器相对于机器人底座的绝对位姿的准确和精确的知识。要做到这一点，需要一个复杂而昂贵的测量系统，而且通常需要将机器人移到异地的校准测量单元中。这项研究项目将利用关节控制方面的最新进展，通过使用位置敏感设备和线性编码器的新组合对刀尖位置进行相对测量，来识别六个关节角度中的误差。一旦确定了联合零值，将使用编码器测量来确定剩余的参数误差。由于观察到关节角度误差通常大于其余参数的误差，因此开发了一种测量系统，该系统将允许在使用相同的实验数据集的同时，将关节角度校准与其余参数分开。该项目的目标是开发一套完整的、通用的系列式工业机器人运动学标定系统。

The inherent lack of positioning accuracy in standard, un-calibrated industrial robots is a limiting factor for their use in the automation of manufacturing processes, such as those at Enclosures Direct Inc. (EDI). Robots utilized for manufacturing are generally designed to have excellent repeatability (ability to continually reach the same position and orientation, or pose, defined by a set of joint angles stored in the controller) but are lacking in accuracy (moving to a pose for which the joint angles must be computed by the controller). To overcome the accuracy shortcomings the robots are typically used in a "record and play" mode where poses are manually taught by a technician and repeated for long production runs. When dealing with shorter production runs of customized products, as is the case at EDI where over 3000 different products are offered, it is not viable to manually teach all of the required poses which are instead calculated based on part dimensions or a CAD model. This approach requires the robots to be calibrated in order to meet accuracy requirements. Existing kinematic calibration methods require accurate and precise knowledge of the absolute pose of the robot end effector with respect to the robot base. Achieving this requires an elaborate and expensive measurement system and often requires moving the robot offsite to a calibrated measurement cell. This research project will use recent advances in joint mastering to identify the errors in the six joint angles using relative measurements of tool point position using a novel combination of position sensitive devices and a linear encoder. Once the joint zero values have been identified, the encoder measurements will be used to identify the remaining parameter errors. Observations that the joint angle error is generally larger than the error in the remaining parameters has led to the development of a measurement system that will allow for the separation of the joint angle calibration from the remaining parameters while employing the same experimental data set. The goal of the proposed project is to develop a complete and general system for the kinematic calibration of serial industrial robots.