Research Areas‎ > ‎

Multi-Scale Manipulation & Assembly

The MSRAL is interested in applying flexible manufacturing techniques to micro, meso, and nano-scale manipulation tasks utilizing simple or minimalistic actuation and sensing schemes. Therefore, we have designed, prototyped, and customized a flexible automation system for meso-,micro-, and nano-scale manipulation tasks. This test-bed has an inverted optical microscope with a CCD camera attached to one of the ports. As a result, whatever is observed in the field of view (FOV) of the microscope can be routed to a control PC for image processing. There are multiple micromanipulators in the system configuration which are capable of being computer controlled with minimum incremental motions on the order of 60 nm along each degree of freedom. There is a computer controlled motorized XY stage on the microscope platform for automating planar movements of a sample residing on the stage.  For example, during meso-scale manipulations, a 4X objective is placed on the microscope yielding a FOV of approximately 3.3 mm x 2.5 mm. The images from the camera are 640 x 480 pixels in size, thus each pixel in the image corresponds to approximately 5 µm. Typically, each manipulator is outfitted with one or two 5 µm or 25 µm tip diameter tungsten probes, which can also be attached in-line with a 10g capacity load cell to sense forces at the mN level. Custom image processing and control software used to interface the hardware components of the platform with a quasi-static dynamics simulator and path planning algorithms to simulate and execute various tasks.

                MSRAL Meso-, Micro-, and Nano-scale Manipulation Test-bed

Field of View with 4X objective showing planar part with four manipulators.

Meso-scale Manipulation & Assembly

For manipulation and assembly tasks at the meso-scale, surface forces, such as stiction, friction, and electrostatic forces, dominate. Most parts are planar so access to the parts from the top is possible. Grasping the part with a suction gripper can be performed, however it can only be used to approximately position the parts due to the difficulties in part release resulting from aforementioned dominant surface forces. Therefore, manipulation and assembly with a gripper is not a preferred technique and manipulations with point probes are required. To study this problem, the canonical peg-in-the-hole problem at the meso-scale had been investigated. The peg-in-the-hole problem involves assembling a planar, rectangular part into a planar, rectangular slot with uncertainties.

The parts are about 40um thick and manufactured out of beryllium copper using a photochemical machining process. The hole or fixture is attached to a glass microscope slide which is coated with a thin layer of mineral oil. This system has uncertainties from three different sources. There are uncertainties in estimating (or sensing) the state of system (peg); errors in the control/actuation of the probe position relative to the part; and uncertainties in the geometric (from manufacturing) and physical parameters (from modeling) of the system. Two types of support surface friction models have been considered for the system: a three-point support surface and a viscous damping layer along the surface. Open-loop plans from a sampling-based motion planner have been implemented experimentally to successfully manipulate the peg into the hole. These manipulations involved a single 5um tip probe on the active (computer-controlled) manipulator.

Canonical peg-in-the-hole task: Move part from configuration A to configuration B.

Typical part and fixture dimensions for experimental trials.

Manipulation task: Experimental results obtained from an automatically generated open loop plan.

Subsequent work on closing-the-loop has been performed and also resulted in successful manipulations of the peg into the hole. In this work, both manipulators are used. This time the passive (manual) probe is outfitted with the single tip probe (STP) while the active manipulator has a dual tip probe (DTP). Three types of interactions with system are allowed: (a) one-point contact with the DTP, (b) two-point contact with the DTP, and (c) one-point contact with the DTP along with one-point contact with the STP. These types of interactions are combined in different ways to generate robust motion primitives to predictably manipulate the position and orientation of peg and move it from the starting position to the goal location.

         (a)                               (b)                                    (c)

Planar manipulation with a single degree-of-freedom, dual-tip probe
and a passive single-tip probe. There are three sets of operations that can
be performed.


Pushing with one-point contact with the DTP

Robust one-point sticking contact with CCW rotation

Pushing with two-point contact: The DTP is shown with the exaggerated misalignment between its two tips for better visualization.

Robust two point sticking contact

Robust rotational motion primitive using the DTP and STP.

Robust rotation


Planing with robust motion primitives.  The assembly task considered is shown above.  For such a task, the environment are currently being ignored. For such a task, the planning algorithm relies on composing the simple robust motions defined above. The following three higher level robust motion primitives are first constructed using these simple ones.
  1. Robust translation in the X-direction:  Use the DTP to push the part in the X direction while maintaining two-point sticking contact
  2. Robust translation in the Y direction:  Compose a robust motion with one point sticking contact and intended rotation followed by a robust to-two-point-contact motion
  3. Robust rotation:  Use the primitive as is, as shown above.


Meso-scale Part Manipulation & Assembly with Robust Motion Primitives

Video showing the execution of the automatically generated assembly plan using robust motion primitives.

Related Publications

  1. D. Cappelleri, J. Fink, B. Mukundakrishnan, V. Kumar and J. Trinkle, “Designing Open Loop Plans for Planar Micro Manipulation”, Proceedings of the IEEE International Conference on Robotic and Automation (ICRA), Orlando, FL, 2006.    
  2. D. Cappelleri, J. Fink, V. Kumar, “Modeling Uncertainty for Planar Meso-scale Manipulation and Assembly”, Proceedings of the ASME International Design Engineering Technical Conference (IDETC), Philadelphia, 2006.
  3. P. Cheng, D. Cappelleri, B. Gavrea and V. Kumar, “Planning and Control of Meso-scale Manipulation Tasks with Uncertainties”, Proceedings of Robotics: Science and Systems (RSS), Atlanta, GA, June 27-30, 2007
  4. D. Cappelleri, P.Cheng, B. Gavrea, and V. Kumar, “Meso-manipulation: System, Modeling, Planning and Control”, Video, IEEE International Conference on Robotic and Automation (ICRA), Pasadena, CA, 2008.
  5. D. Cappelleri, M. Fatovic, U. Shah, “Caging Micromanipulation for Automated Microassembly”, Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), Shanghai, China, May, 2011.
  6. D. Cappelleri, M. Fatovic, Z. Fu, "Caging Grasps for Micromanipulation & Microassembly", IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), San Francisco, CA, USA, September 25-30, 2011.
  7. D. Cappelleri, P. Cheng, J. Fink, G. Gavrea, V. Kumar, "Automated Assembly for Meso-Scale Parts", IEEE Transactions on Automation Science and Engineering, vol.8, no.3, pp.598-613, July 2011.
Copyright Notice
Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the publishers.