Research project MiniVision

parlab

research> activities

Research projects

MiniVision for micro-assembly

James Mure-Dubois

Project program

Partners: CSEM

Keywords

Computer vision, micro-assembly, 3D vision, multifocus, depth from focus

Objective of the project

Present global vision systems are not adequate for robot control in an assembly environment where individual components have sizes in the range of 1 to 100 micrometers, since current systems do not provide sufficient resolution in the whole workspace when they are fixed, and they are too bulky to be brought close enough to the components. A local 3D vision system (fig. 1) is expected to provide two decisive advantages: high accuracy and high exibility. The project aims to research 3D vision sensors easily embedded in a micro-assembly robot

Figure 1: Micro-assembly with global vs. local vision system.

Figure 2: BW image of nails scene.

Figure 3: Range image of nails scene.

Figure 4: 3D rendering of nails scene

Methodology

Accuracy, size and weight constraints for the embedded vision sensor were defined. Based on those constraints, a screening of 3D vision methods was performed. The most promising method, multifocus acquisition, was selected for further investigation. The image acquisition and processing element of the embedded system were prototyped and tested on typical assembly scenes. The accuracy was compared to measurements with a reference microscope system.

Results

Experiments ( fig. 2,3 & 4) carried out with multifocus imaging. The applicability of multifocus imaging to micro-assembly was confirmed:

imager package's mass (20g) fits within embedment constraints.
Accuracy (20 µm) is sufficient for many assembly tasks.
Real-time processing is possible for regions of interest.

Further steps in the development of an embedded prototype could include the design of a suitable actuator for the imager package, and reduction of the processing time through on-chip processing.

Publication

[1] J. Mure-Dubois, H. Hügli, "Embedded 3D vision system for automated micro-assembly", Proc. Conf. Two- and Three-Dimensional Methods for Inspection and Metrology IV, Proc. SPIE Vol. 6382-20, 2006 - Abstract - Postprint on RERO