Digital twin models for advanced manufacturing are constrained by the fundamental geometric representations that are currently used. Instead of a singular representation, mesh, voxel, and parametric surface representations all require multiple conversions along the digital design, manufacturing, and inspection processes. These conversions inherently introduce error and are time-consuming, complicate comparison of in-situ sensor data to the as-designed model, and result in a complex, fragmented process chain.
We introduce a novel holistic digital twin representation based on a voxelized, GPU-accelerated, adaptively sampled distance function (ASDF). The framework enables rapid comparison of in-situ sensor data to as-designed models, provides a baseline representation of as-designed geometry for control systems, serves as a foundation for path planning tools, allows rapid comparison of sensor data to as-designed models, and as such, could enable real-time, online path planning during the manufacturing process.