Real-Time Robot Positioning based on Measurement Feedback Control

摘要

Industrial robots are well introduced into automated production processes. Their mechanical design is dominated by some major key factors like required flexibility, different payload demands, working range, working speed, combination with different working tools and robot costs. The final achieved position accuracy of the robot tool centre point (TCP) is based on the combination of these key factors. In general, the smallest movement steps and the repeatability of robots are much lower than the absolute achievable accuracy. The positioning results and especially the programmed paths of the TCP show relatively large differences between the programmed nominal paths related to the final achieved movements in reality. These differences can be detected using the Absolute Tracker with its very high dynamic performance, especially if the 6DoF capability is included. Each set of tracker measurements (X,Y,Z,i,j,k and the timestamp t) are describing the true position and orientation of the measured object in space. The data rate of 1000 measurement per second allowed always a very accurate observation of relatively fast movements. For all inspection tasks and for static positioning processes a short time delay until the measured data are shown on the screen and stored in the computer doesn't have any impact on the results. To correct and improve robot positions during the movement, measurement data have to be available in real-time that it is possible to use these measurements as an input for a feedback control loop. Since last year the Absolute Tracker provides such a real-time interface with a maximum data rate of 1000 Hz via an EtherCAT fieldbus. Meanwhile several implementations are realized, mainly within research projects to gain first experiences and proof which improvements are achievable. It was immediately recognized that now the robot controller becomes the bottleneck because its capability to react on control commands is the limit now. In addition each robot supplier has its own interface and specific commands available. The best and ultimate solution to reposition the robot TCP in real-time would be develop a robot independent device like a specific end-effector which allows a fast feedback control based on the measurement results with a relatively small working range of about 100 mm. We started now to develop a first model of such a real-time controlled end-effector and want to introduce this idea and hopefully show first results.