Future Work

Only a few areas of compiling for computational origami have been explored in this thesis. Many more remain. Among them are:

• A new type of input and output should be added to the language, the FLOATING INPUT and FLOATING OUTPUT. These channels would be free to move around during optimization, and would solve some of the hand-optimization problems encountered in section 3.4. Fixed position INPUTs and OUTPUTs will still be necessary for interfacing with external connections, and for making sure that outputs and inputs line up when two origami arrays are connected or the outputs of an array are fed back to the inputs to produce a finite state machine.
• The annealing schedule should be experimented with. The shape of the temperature vs. iteration curve can be very important in determining the quality of the resultant system.
• A number of aspects of the placement and routing algorithm could be improved. Specifically, the determination of goal columns needs some work since the modules are shifted to make room for wires in a very unintelligent fashion.
• Other iterative optimization methods should be investigated. For example, genetic algorithms [24][11][8][7] have proven themselves worthwhile in many different applications.
• The boundary conditions of origami arrays should be analyzed and used to perform time-domain shifting of data. Among the applications that could be envisioned are parallel to serial and serial to parallel converters, time-domain multiplexers, and packet switchers. A new language should be developed to allow the programmer to express both data flow and time constraints in a concise, intuitive manner.