Real-time vibration suppression in robotic end effectors

Title:  Real-time vibration suppression in robotic end effectors
Author: Jacob Heyden & Georg Emmanouilidis
Date & Time: June 4th, 13:15–14:00
Location: Seminar Room M 3170-73 in the M-building, LTH
Supervisor: Yiannis Karayiannidis, Jon Davidsson (Beckhoff Automation AB) & Daniel Jovanovski (Beckhoff Automation AB)
Examiner:  Björn Olofsson
Opponents: Dante Neckmar


Abstract:
Industrial robotic manipulators are widely used in manufacturing for high-speed pick-and-place operations. Therefore, there is value in optimizing the trajectory execution speed and positional accuracy to reduce cycle times. However, pushing these acceleration limits often induces structural residual vibrations at the end-effector, which degrade process quality and necessitate long settling times. Traditional methods to mitigate this, such as mathematical plant inversion, may lead to stability issues due to unmodeled physical constraints and shifting payload inertia. This thesis evaluates the performance of a control architecture that integrates real-time system identification with feed-forward input shaping to actively suppress end-effector vibrations. The evaluated control architecture utilizes real-time Fast Fourier Transform (FFT) frequency identification, coupled with ZV, ZVD, and ZVDD shapers. The results show that it is possible to successfully suppress vibrations without relying on full-state mathematical models, although robustness to model errors proves more valuable than theoretical execution speed. The ZVDD shaper performed best, consistently achieving the fastest absolute settling time by absorbing the parameter estimation errors caused by discrete digital execution. Furthermore, while continuous real-time shaping is dynamically viable, mid-trajectory parameter updates induce additional compensating movements that extend the vibrational settling time. Consequently, the findings demonstrate that defaulting to pre-calibrated structural vibrations yields higher performance. The system maintains optimal cycle times by relying on the broad sensitivity curve of the robust shaper to inherently absorb normal frequency outliers and dynamic payload shifts. By reserving real-time adaptation strictly for drastic, unforeseen mechanical changes, the proposed architecture provides a reliable and industrial solution.