Assembly by Intelligent Scaffolding ; CU-CS-1080-11

We propose a novel class of algorithms for autonomously assembling structures from inert building blocks guided by intelligent scaffolding components. Intelligent scaffold units are equipped with sensing, actuation, computation and communication abilities and facilitate the attachment of inert building blocks to the structure. After attaching an inert building block, the scaffold structure reconfigures to attach the next block until the structure is completed. The proposed algorithms are scale-free and independent of the implementation of the locomotion of building blocks and intelligent scaffolding blocks. For example, movement of building and scaffold blocks can be achieved using manipulating robots or self-assembly in a well-stirred liquid. In a robotic assembly context, the intelligent scaffolds take the role of markers on the structure and allow for reducing the perception and coordination requirements on the robotic team. In this paper, we describe algorithms for converting any desired structure that can be represented as 3D lattice into a finite state machine that is executed by intelligent scaffolding blocks; we prove that all finite structures can be assembled using intelligent scaffolds; and we provide examples of simulations that assemble a square, a fractal structure, and a model of a space station, each using only three intelligent scaffold components.