Efficient Reconfiguration of Lattice-Based Modular Robots

摘要

Modular robots consist of many small units thatrnattach together and can perform local motions. By combiningrnthese motions, we can achieve a reconfiguration of the globalrnshape. The term modular comes from the idea of groupingrntogether a fixed number of units into a module, which behavesrnas a larger individual component.rnRecently, a fair amount of research has focused on Crystallinernrobots, whose units (and modules) fit on a cubic lattice. Whenrnthe proper module size is formed, these robots can reconfigurernin linear time within a rather physically restrictive model, or inrnO(log n) time in a more unrestricted theoretical model.rnIn this paper, we show that the results for Crystalline robotsrnalso apply to two other modular robots: M-TRAN and Molecube.rnThe common requirement, for each robot type, is that a fixedrnnumber of units combine to create modules of specified shapes.rnIn this way, we are able to simulate the actions of Crystallinernmodules. Previous reconfiguration bounds thus transfer automatically,rnas long as the robots are composed of the module shapesrnthat we specify.