A design and control strategy for a compliant delta manipulator

Abstract

Compliant robots have a bevy of benefits, particularly when interacting with humans or fragile objects as the maximum output force is limited by the compliance. The relatively low inertia of a parallel configuration, such as a Delta manipulator, complements this; however, compliance affects precision and speed, two hallmark characteristics of such a configuration. This may be compensated for via adaptive control. Iterative Learning Control (ILC) may be adopted to compensate for the flexibility by learning how the system reacts across the workspace over time; but, this can lead to a latency when the end-effector drags behind its ideal path due to loading and inertia. This paper introduces a two-stage ILC strategy that corrects for this latency with a separate ILC controller. Thus, it is possible to converge to the desired output without the need for complex models, which may be brittle to tuning, and without risking the instabilities that may arise from a single controller with rigid-body assumptions. Limitations in control due to frequency dependency of compliant materials are also investigated.