Research Group: Vision Feedback for Micro- and Nanohandling
At micro- and nanolevel, manipulation- and assembly procedures require continuous feedback from ongoing processes. Integrated sensors are very helpful but can only deliver incomplete process data. Additionally there is indeterminacy caused by effects such as thermal expansion, backlash or force effect. The researchgroup 'Vision Feedback for Micro- and Nanohandling' works on the extraction of sensor information from image data with the objective of process automation. Several imaging modalities are employed, including CCD-cameras, optical microscopes, scanning electron microscopes and medical imaging devices.
- Object tracking and depth estimation
- Object recognition and classification
- 3D Image Processing
- Magnetic resonance imaging and propulsion
The research area object tracking is dealing with the development of real-time algorithms for the tracking of moving micro and nano objects, like micro grippers or manipulated carbon nanotubes, in noisy Scanning Electron Microscope (SEM) video streams. In particular the use of a SEM makes high demands on the image processing. High update rates of the pose data enforce a short image acquisition time of the SEM images. Hence the image noise increases, because frame averaging or averaging of the detector signal is time consuming. Therefore algorithms with high robustness against noise with real-time capability are needed. Mobile microrobots with piezo slip-stick actuation and more than one degree of freedom mostly do not have internal pose sensors. One possibility for fast pose estimation with a high accuracy is the application of video cameras. If accuracy in the micrometer or even in the nanometer range is required, a light microscope respectively a scanning electron microscope (SEM) is needed. Robust object tacking algorithms are required for fast and accurate pose estimation.
Figure 1: Microgripper and carbon nanotube. The gripper position can be calculated well
using the active contour. Depth information is only available after 3D imaging.
Video 1: AFM cantilever in bending experiment. An active contour fits
to the cantilever shape and allows reconstruction of the bending curve.
The goal of pattern recognition in general is the automatic classification of signals. Current research projects at the Division Microrobotics and Control Engineering incorporate plenty of interesting applications of pattern recognition. For instance defect work pieces or tools may be sort-out before microassembly. Basically pattern recognition may be subdivided into the extraction of relevant features (especially from image data) und the process of classification itself. Our research activities cover both topics.
During feature generation from microscopic images problems arise that require consideration of the particular method of image acquisition. This includes the limited depth-of-focus in light microscopy and the special properties of noise in scanning electron microscopy. Various feature selection strategies are applied
Classification is a mathematical problem with a large number of proposed solutions. The approach of Support Vector Machines is of special interest for our purposes. Latest object of research is the utilization of additional process data in the optimization procedure.
Figure 2: Oocytes in capillary tube. Oocyte shape as well as the peculiarity of animal- and vegetal pole are detected automatically and allow conclusion of oocyte viability.
Contact person: Tim Wortmann
In micro- and nanotechnology, handling and manipulation processes are important. Due to the size of the structures, Scanning Electron Microscopes (SEM) are employed. For precise handling, there is a need to know the position of the objects and the tools in all three dimensions of space. For solving this problem, the 2D information of SEM images is not sufficient. There is a need for 3D information. By means of tilting the sample table or tilting the electron beam by a special lens system, pairs of stereoscopic images can be generated. With such images the third dimension (relative depth) can be determined. Until now, in most of the cases, only area-based algorithms for surface reconstruction have been used in SEM. Area-based algorithms are not suited for the observation of technical handling processes in an SEM. The reasons for this are the pixel accuracy of the results (disparity maps), the lack of noise reduction in low-texture regions of the calculated disparity maps and lack of invariance to contrast differences between the input images. Therefore, in AMiR a 3D-imaging system using beam tilting and based on biologically motivated stereoscopic methods is being developed. This system calculates a high density disparity map in sub-pixel accuracy and is optimized for the observations of technical processes in an SEM.
Contact person: Robert Tunnell
Magnetic resonance imaging has become a widespread technique in clinical practice over the last decades. There are more than 20000 installations worldwide. The main advantage of MR-imaging is the high contrast between different soft tissues. Also it does not incorporate ionizing radiation. Generally the magnetic fields used for MR-imaging may also be utilized to move magnetic particles. This introduces new fields of application such as targeted drug delivery and embolization, for example in order to fight tumors.
For these purposes the magnetic propulsion is only usable if there is also a continuous position control. The research of our group is focused on simultaneous imaging and propulsion, incorporating the same hardware (gradient coils). It has to be taken into account that MR-imaging states high requirements regarding the homogeneity of the applied magnetic fields. If magnetic material is inserted into the volume under observation, this assumption is violated which results in a number of defects in the spatial registration of measured signals or even a complete loss of signal. As a result, image artifacts arise with a volume of several orders of magnitude above the magnetic particle's volume.
Current Active Projects
- NanoMa -Nano-Actuators and Nano-Sensors for Medical Applications
- FIBLYS - Building an Analyzing Focused Ion Beam for Nanotechnology
- HYDROMEL - Hybrid ultra precision manufacturing process based on positional- and self-assembly for complex micro-products
- NANOHAND - Micro-Nano System for Automatic Handling of Nanoobjects
- 3D-REM - 3D-Echtzeit-Bildgebungs- und Messsystem für Rasterelektronenmikroskop
- NANORAC – Nanorobotics for Assembly and Characterization
- ZuNaMi - Zukünftige Verfahren der Nano-/Mikroproduktion
- NanoStoRe - Mikroroboterzelle zur automatisierten Handhabung und Montage von CNTs für die Integration von Mikro- und Nanoobjekten innerhalb eines Rasterelektronenmikroskops
- ROBOSEM - Development of a Smart Nanorobot for Sensor-based Handling in a Scanning Electron Microscope
- REMROB - Entwicklung eines flexiblen Mikroroboters zur Handhabung im Rasterelektronenmikroskop
- Christian Dahmen, M.Sc. (Head of the group and contact person)
- Christian Geldmann
Teaching offers are only available in German. For questions please contact us via email.
- C. Dahmen: "Focus-based depth estimation in the SEM", Proc. of Int. Symposium on Optomechatronic Technologies (ISOT), San Diego, CA, U.S.A., 17-19 November, 2008
- S.Fatikow, D.Jasper, C.Edeler, Ch.Dahmen: "Flexible visual feedback for automated nanohandling inside an SEM", Proc. of Int. Symposium on Optomechatronic Technologies (ISOT), San Diego, CA, U.S.A., 17-19 November, 2008
- C. Dahmen, T. Wortmann, and S. Fatikow: "OlVis: A Modular Image Processing Software Architecture and Applications for Micro- and Nanohandling", Proc. of the Eighth IASTED Int. Conference on Visualization, Imaging and Image Processing (VIIP 2008), Palma de Mallorca, Spain, September 1-3, 2008, pp.245-250
V. Eichhorn, S. Fatikow, Th. Wich, Ch. Dahmen, T. Sievers, K.N. Andersen, K. Carlson and P. Bøggild: "Depth-Detection Methods for Microgripper based CNT Manipulation in a Scanning Electron Microscope", Journal of Micro-Nano Mechatronics, Vol.4, Issue 1, pp.27-36, Springer, 2008.
T. Sievers: Echtzeit-Objektverfolgung im Rasterelektronenmikroskop. In D. Wagner et al. (Hrsg.): GI-Edition Lecture Notes in Informatics, Ausgezeichnete Informatikdissertationen 2007, pp. 279-288, Köllen, 2008
D.Jasper, C.Dahmen and S.Fatikow: "CameraMan - Robot Cell with Flexible Vision Feedback for Automated Nanohandling inside SEMs", IEEE Conference on Automation Science and Engineering (CASE 2007), Scottsdale, USA, September 22-25, 2007, pp.51-56
M. Jähnisch, S. Fatikow: "3D Vision Feedback for Nanohandling Monitoring in a Scanning Electron Microscope", International Journal of Optomechatronics, vol. 1, no. 1, pp. 4-26, 2007, Koh Young Best Paper Award
T. Sievers, S. Fatikow: "Real-Time Object Tracking for the Robot-Based Nanohandling in a Scanning Electron Microscope", Journal of Micromechatronics - Special Issue on Micro/Nanohandling, pp. 267-284 (18), 2006