Design of Programmable Matter

Ara N. Knaian

FULL TEXT

Full Text, PDF

Full Text, E-Book (PRC)

GALLERY

Distributed PostScript rendering on the 1000-node centimeter-scale paintable display. Each node has a processor, infrared communications, and an LED.

A sliding-cube modular robot node being assembled.

A test of horizontal lattice-unit translation using magnetic surface-drive actuation. The images were taken 33 milliseconds apart.

ABSTRACT

Programmable matter is a proposed digital material having computation, sensing, actuation, and display as continuous properties active over its whole extent. Programmable matter would have many exciting applications, like paintable displays, shape-changing robots and tools, rapid prototyping, and sculpture-based haptic interfaces. Programmable matter would be composed of millimeter-scale autonomous microsystem particles, without internal moving parts, bound by electromagnetic forces or an adhesive binder.

Particles can dissipate 10 mW heat, and store 6 J energy in an internal zinc-air battery. Photovoltaic cells provide 300 &mu W outdoors and 3.0 &mu W indoors. Painted systems can store battery reactants in the paint binder; 6 J / mm3 can be stored, and diffusion is fast enough to transport reactants to the particles. Capacitive power transfer is an efficient method to transfer power to sparse, randomly placed particles. Power from capacitive transfer is proportional to VDD 2: 100.W at 3.3V and 12 mW at 35V. Inter-particle communication is possible via optical, near-field, and far-field electromagnetic systems. Optical systems allow communication with low area (sub-mm) particles, and 24 pJ/bit. Near-field electromagnetic gives precisely controlled neighborhoods, localization capability, and 37 pJ/bit. Far-field radio communication between widely spaced particles may be possible at 60 GHz; antennas that fit inside 1 mm3 exist; complete transceivers do not. A 32-bit CPU uses less than 0.26 mm2 die area, 256K x 8 SRAM uses 1.1 mm2, and 256K x 8 FLASH uses 0.32 mm2. Direct-drive electric and magnetic field systems allow actuation without moving parts inside the particles. Magnetic surface-drive motors designed for operation without bearings are not power-efficient, and parasitic interactions between permanent magnets may limit their usefulness at millimeter particle dimensions. Electrostatic surface-drive motors are power-efficient, but practical only at particle dimensions below a few millimeters.

We constructed a prototype paintable display; a distributed PostScript rendering system with 1000 randomly-placed 3.4 cm nodes, each with a CPU, IR communications, and LED. The system is used to render the letter "A." We present a design, not yet constructed, for a literal paintable display, with 1.0 mm rendering particles, each with a microprocessor and memory, and 110 &mu m display particles, with tri-color LED.s and simpler circuitry. Storage of zinc-air battery reactants in the paint binder would provide an 8 hour battery life, and capacitive power distribution would allow continuous operation.

We constructed a prototype sliding-cube modular robot, with 3.4 cm nodes. The system uses magnetic surface-drive actuation. We demonstrate horizontal lattice-unit translation. We describe a design, not yet constructed, for a sliding-cube modular robot with 2 mm nodes. The cubes use standard-process CMOS IC's, inserted into a cubic space frame and wire-bonded together. Arrays of passivated electrodes, 1 &mu m from the surface of the cubes, are used for electrostatic surface-drive actuation, zero-power latching, power transfer, localization, and communication. The design allows actuation from any contacting position. Energy is stored in a standard SMT capacitor inside each node, which is recharged by power transfer through chains of contacting nodes.


Ara Knaian <ara@mit.edu>