MotivationManipulation of microparts requires different strategies than in the macroworld. The human operator or the electronic controller has to be provided with reliable information from the microworld. Video microcopy widely used requires visibility thus severely restricting the flexibility of handling maneu- vers. A different strategy is to use local process information, such as contact forces, to control precise and flexible operations. Different micro- physical effects, such as adhesion, additionally aggravate manipula- tion at the microscale. Methods to avoid or control sticking have to be found in order to perform precise positioning and assembly of microparts. |
ApproachIn this project, we explore the force controlled active pushing of microparts. Manipulation by pushing is programmable, does not require carrying or lifting of objects and works by accessing the workpiece W from one side only. As a pushing tool, we chose an AFM (atomic force microscope) cantilever with an integrated piezoresistive force sensor to measure the contact force. W's local position is inferred from contact events derived from the force signal (Fig. right) and the manipulator's joint sensors. |
ResultsWe have built a 3-degree-of-freedom robot with a resolution of 0.5 mm. Statistical analysis of the pushing signal was used to extract contact events from the pushing force, such as touch, loss of contact, and hitting of obstacles. Built into a finite state machine controller, this allowed us to push 500 micron sized parts on a straight path, orient them, and align them with a (moveable) reference with a final accuracy of 3 mum / 0.1°. A graphical representation of such a push-alignment with the reference, as well as a top view of the initial and final configurations are shown below. |
Alignment sequence |
Initial position |
Aligned Position |
Download a MPEG-video showing the alignment process (6.7 MB!).
back to Wolfgang's Home page |
|