Abstract

The making, maintenance and use of digital fabrication machines is a core competency for industrial economies. However, innovation in this domain is hampered by the pervasive use of GCode: an antiquated interface that lies between machine users and machine controllers. GCode only allows for one-way data transfer, and prevents us from smoothly crossing the boundary between high- and low-level planning and control tasks even though those tasks have deep physical connections. This makes it difficult to develop new machines, train new operators, and educate new machine designers. This thesis develops a modular, feedback-native control architecture and applies it to machines that can think about what they are doing, providing new insights for their operators and designers.

The architecture that I develop rebuilds machine controllers as distributed systems that are modular across hardware and software. This includes work in heterogeneous networking, clock synchronization, lightweight data serialization and transport, and the development of a distributed programming model that combines dataflow for real-time operation with scripting for machine tasking.

This enables me to develop model-based machine controllers. First, I build a motion planner that leverages its understanding of machine physics to improve machine speeds compared to heuristics-based planners. I then extend this planner to optimize motion and process control simultaneously for a 3D printer. I develop methods for high- and low-level printer control, replacing the state-of-the-art workflow with a feedback-based alternative that learns both material and machine parameters. This enables us to use a wider range of renewable and high-performance materials, greatly reduces user burden, and provides useful physical insights. Finally, I develop a CNC milling machine, showing how models can be used to build feedback tools for CNC machine builders and CNC programmers. In each of these cases, models are developed on the hardware where they will be used, reducing our reliance on standalone testing equipment through computational metrology.

With the work in this thesis, machine controllers are no longer black boxes; they are distributed algorithms made of recognizable software design patterns and operated via physical models rather than tacit knowledge and heuristics. With successful industrial adoption, this paradigm will enable the next generation of engineers and scientists to rapidly invent, modify and deploy novel machine systems as they work through the coming decades’ most pressing issues.