Evaluation of Calibration Method for Redundant Dual Arm Industrial Robots Using Internal Sensors

Sammanfattning: To maintain a high position accuracy in robots is essential in today’s industry. Because of different factors, as for example wear and tear, the robots may need to be recalibrated after a while to maintain their high precision. Today the calibration can be costly. The robot is often sent away to be calibrated and will therefore cause stoppages in production, and if the calibration is performed in-house, education of staff and purchases of expensive external measurement equipment are required. In this thesis, a calibration method for the dual arm, seven degrees of freedom (per arm) robot YuMi [YuMi 2016] from ABB is evaluated. The idea is to connect the two arms to each other, perform specified movements and use solely the internal sensors in the robot to do the calibration. The arms will also be docked in fixed positions in the robot’s work space and while fixed, they will be moved into different configurations to obtain additional measurement data. A challenging part is that the robot has an extra joint in each arm compared to common six-axis industrial robots, which makes the robot inherently redundant. The redundancy makes it possible to reach a certain point with the robot tool using several different arm configurations, this makes it more difficult to estimate the robot’s parameters. In this thesis, several different types of calibrations were performed. Because the calibration methods have some drawbacks that need to be further investigated, the correct parameters were only identified in some cases. In those cases where the real parameters were not found, only the repeatability, and not the accuracy, was improved. The methods showed these results in simulations, while the actual robot was not successfully calibrated in experiments.