Distributed Manipulation: Using Complexity to Reduce Complexity


Professor Jonathan E. Luntz

Department of Mechanical Engineering

University of Michigan


A general trend in robotic manipulation involves reducing the complexity of a system while preserving the ability of the system to position and orient objects. These reductions in complexity often take the form of reduced degrees of freedom in actuation, and the reduction or elimination of sensing and feedback, although these "minimalistic" systems often require complex planning and iterative algorithms. Distributed manipulation systems take a slightly different approach by adding degrees of freedom in actuation to reduce or eliminate sensing and feedback as well as sensing and iteration.

Distributed manipulation systems induce motions on objects through the application of many external forces. Such systems generally involve redundant actuation, and hence provide tremendous manipulation power. Many forms of distributed manipulation exist, three of which will be discussed. First, actuator arrays perform distributed manipulation using a planar array of many small stationary elements (called cells) which cooperate to manipulate larger objects. A particular macroscopic actuator array will be presented along with the design and analysis of both open and closed-loop manipulation strategies for positioning and orienting objects in the plane on such macroscopic, spatially discrete arrays with hybrid dynamics using distributed control. A second form of distributed manipulation uses air flow fields along which small flat objects "float". A new approach to using air flow will be presented which uses potential flow theory to position and orient through a superposition of simple flow sources and sinks. Finally, ideas for using snake-like or tentacle-like robots for distributed manipulation will be presented.


Friday, February 2, 2001

3:30 - 5:00 p.m.

1500 EECS