Design-Intent Compilation for Heterogeneous Fabrication

Researchers have introduced a compiler architecture for heterogeneous fabrication that keeps the source design separate from the machine-facing output. The core idea is to define geometry plus named, typed spatial attribute fields over a shared object domain, then lower those attributes through translation models into backend-specific instructions.

That matters because fabrication workflows often force designers to encode intent as material fractions, voxel labels, mesh partitions, or slicer settings tied to a particular printer. In practice, that makes the same object hard to reuse across different toolchains, especially when the design includes appearance, mechanical behavior, process state, or measured data.

The system is built around three layers: source modeling, attribute translation, and backend compilation. Translation models derive required realization attributes from the source or from higher-level intent, while backend compilers emit outputs such as voxel material stacks, process G-code, or configured slicer project files.

The approach has been validated with fabricated objects and measurements across sampled volumetric data, CT-derived visual and mechanical models, Shore-hardness fields, and full-color fields, using both material jetting and material extrusion. The practical upside for developers is a more reusable pipeline for heterogeneous fabrication research, plus an open-source Python package that exposes the representation, translation framework, compiler interface, and workflows.