An analysis of Cartesian deviations in robot work cycles
Robots have proven to be extremely useful in performing a variety of tasks, especially those that are repetitive in nature. Most applications require good accuracy and repeatability, therefore positioning errors need to be minimized to enhance any practical application. Information concerning the source of positioning errors and their influence on the work cycle is inherently useful in the study of robotics. The Teachmover is a five-jointed Robot arm that is widely used for educational purposes. The robot does not have a built-in feedback system and therefore does not have a high repeatability, which limits its usefulness to classroom learning. This project attempted to identify alternatives to correct deviations from the programmed points in a Teachmover's routine. A comprehensive analysis of the variables affecting position errors is presented. A strategy to locate a repeatable home position (starting point) was defined, together with a detailed methodology to analyze the deviations. This methodology included statistical tests which are common to a similar class of problems. The results indicated that the load on the robot and the working speed were the main causes of the deviations. It was found that neither the area of execution in the working envelope nor the number of cycles of execution affect the error. Linear relationships were found to exist between load and speed with the positioning error. It was established that there was no correlation between load and speed.