At the microscale, soft materials have mechanical properties, namely stiffness and damping, close to that of the actuation and sensing mechanisms of robotic micromanipulation systems. Therefore, during gripping tasks, samples have enough variation to induce instabilities that damage the robotic system or the sample. Commonly used control techniques at the microscale lack the robustness required for the micromanipulation of soft materials. The contribution of the work has been to propose a new approach for force control at the microscale by the design of a low dimensional controller. Our approach consists in defining the position of the microrobotic fingers as an uncertain parameter (it is related to the uncertainty of the diameter of the objects to be manipulated) and both the stiffness and the damping of the manipulated object as two other uncertain parameters.
Experimental setup for robust force control at the microscale (the enlarged view (a)
and (b) shows respectively a non compressed and a compressed gripped Expancel microsphere (AkzoNobel)
The proposed methodology consists first on the experimental identification of the diameter, the stiffness and the damping of a broad set of different soft micro-objects (which can be biological cells, biomaterials, etc.). From a LPV model, a set of elementary models corresponding to each characterized micro-object are then defined. After that, the gripping force control is designed through a robust observer based controller and a multimodel eigenstructure assignment. The shortcoming of this approach is that it cannot deal with no measurable uncertain parameters. In gripping tasks, the size, the stiffness and the damping of the samples are part of uncertain parameters which are not measurable. To overcome this issue, the use of a self-scheduled form of the observer is proposed. The most relevant multimodel constraints required for the synthesis of the robust controller have been defined through a worst case analysis. This analysis allowed the determination, from a set of 65 samples, that the mechanical properties of two samples are relevant for the determination of the multimodel constraints. The order of the controller is equal to the number of observers, and is potentially low which is of importance from application point of view. Results have shows the first experimental demonstration of robust gripping force control at the microscale for the manipulation of a broad set of soft materials with varying mechanical properties. The control approach allows the development of new robotic procedures for biological manipulation were samples can vary greatly. This is often the case in biology.
Experimental robust force control results have been obtained with the manipulation of a set of 65 Expancel microspheres. The spheres are composed of a polymer shell encapsulating a gas. They are deformable, soft and resilient, with properties in the range of biological samples which makes them very attractive for force control experiments at the microscale. The compressibility test of Expancel microspheres perfromed by AkzoNobel can be shown through the following link [Video 1].
References
M. Boudaoud and S. Régnier, “An overview on gripping force measurement at the micro and nano-scales using two-fingers microrobotic systems”. International Journal of Advanced Robotic Systems 2014, 11:45 | doi: 10.5772/57571.
M. Boudaoud, M. G. De Faria,Y. Haddab, S. Haliyo, Y. Le Gorrec, P. Lutz and S. Régnier. Robust microscale grasping through a multimodel design: synthesis and real time implementation. Control Engineering Practice, vol 39, pp. 12-22, 2015.
M. Boudaoud, M. G. De Faria,Y. Haddab, S. Haliyo, Y. Le Gorrec, P. Lutz and S. Régnier “Robust Microscale Gripping Using a Self-Scheduled Dynamic Controller”. IFAC World Congress, Cape Town, South Africa, 2014.
