SJET
Logic Matter: MIT 2010
Logic Matter: Digital Logic as heuristics for physical self-guided-assembly
Given the increasing complexity of the physical structures surrounding our everyday environment; buildings, machines, computers and almost every other physical object that humans interact with, the processes of assembling these complex structures are inevitably caught in a battle of time, complexity and human/machine processing power. If we are to keep up with this exponential growth in construction complexity we need to develop automated assembly logic embedded within our material parts to aid in construction. In this thesis I introduce Logic Matter as a system of passive mechanical digital logic modules for self-guided-assembly of large-scale structures. As opposed to current systems in self-reconfigurable robotics, Logic Matter introduces scalability, robustness, redundancy and local heuristics to achieve passive assembly. I propose a mechanical module that implements digital NAND logic as an effective tool for encoding local and global assembly sequences. I then show a physical prototype that successfully demonstrates the described mechanics, encoded information and passive self-guided-assembly. Finally, I show exciting potentials of Logic Matter as a new system of computing with applications in space/volume filling, surface construction, and 3D circuit assembly.
Full PDF document available upon email request.
Advisor: Terry Knight
Professor of Design and Computation, Department of Architecture, MIT
Advisor: Patrick Winston
Ford Professor of Artificial Intelligence and Computer Science, EECS, MIT
Reader: Erik Demaine
Associate Professor, EECS, MIT
Physical Mechanism
Grey and white plastic Logic Matter mechanisms after roto-molding. Output faces with values of 0 & 1 shown.
Prototype
60 unit working prototype demonstrating three dimensional single path & user programmability. Grey units act as NAND gates while white units are inputs.
User Programmability
Right-angle tetrahedron demonstrating the functionality of a NAND gate through input, output and gate decisions. Output A and Output B are the two output faces, either [0] or [1]. Input A and Input B are the two input faces. The input faces can both receive either [0] or [1] at any time. The input faces will dictate the decision in the Gate unit as to which face (Output A [1] or Output B [0]) will be utilized, directly based on the NAND truth table.
Self-Guided-Replication
A single sequence is read and written from information stored in the materials. Duplicate structures are built.
Single Path Growth & Geometry Descriptions
Random search on/in sphere geometry. The single path description allows any geometry (line, surface, volume) to be described by a single sequence of turns (left, right, up, down). Turn sequences correlate to digital logic gate inputs/outputs.
Growth with Robustness & Redundancy
Single path assembly of a random growth showing redundant inputs. Single path shown in orange/green and redundant nodes shown in black.
Failure and Dissassembly
A complex structure with a single point of failure can be reassembled accurately by reading the adjacent unit's storage information. This example shows only local knowledge in necessary for construction because the materials allow the read-write storage of information.
Programmable Sequences of Growth
Binary gradient sequences & resultant spatial output. Orange units are inputs equal to 1, grey units are gates.
Unit Roto-Molding Process
Roto-molding sphere constructed from MDF wood. Releasing process from a two part rubber mold.
Assembly Sequence
Assembly animation from 60 individual units to final global configuration.
