Control of haptic devices for tele-microassembly
|Project members:||I. Polat, MSc (Ilhan), prof.dr. C.W. Scherer (Carsten)|
|Keywords:||Robotics and mechatronics, Optimization-based control, Data driven and fault tolerant control, Model-based control|
Accurate control of haptic micro-assembly devices requires fine-tuned multivariable impedance and admittance shaping for the realization of scaling for the two-sided interaction between micro- and macro-domains. A particular challenge arises from the inclusion of stiffness directionality in order to avoid the damage of work-pieces. Model-based H_infinity–synthesis is an ideal tool for optimal loop-shaping which has not been systematically employed in haptic controller synthesis. As a second major challenge in general manipulator systems, one has to keep up stability and performance even if the slave (or operator) interacts with a strongly varying environment.
In haptic micro-assembly, the superposition of external forces acting on the workpiece leads to position-dependent nonlinear characteristics. In order to avoid overly conservative designs based on crude passivity techniques for ensuring stability and performance robustness, it is of fundamental relevance to develop suitably structured uncertainty models (such multi-variable sector-conditions on force nonlinearities) for the work-piece environment in haptic micro-assembly.
As a substantial benefit, these techniques even offer the opportunity to design controllers that adapt themselves to measurable changes of the environment in order to even further enhance performance. In view of the generic tuning-complexity, classical ad-hoc gain-scheduling techniques seem inferior to recently developed one-shot algorithms for optimization-based scheduled controller synthesis, and their extension to dynamic integral quadratic constraints which are currently under development. Thus it would be beneficiary if the state-of-art modelling techniques can be modified in order to obtain a parameter-varying model so that current advances regarding LPV gain-scheduled controller design can be applied.