Abstract
This paper presents an NC milling simulation kernel that uses voxel models as an underlying representation of workpiece shapes. To achieve simplicity and efficiency in the boolean operations, which is the main process of milling simulation, our kernel uses Adaptively-sampled Distance Fields (ADFs). By the nature of ADFs, whose fundamental data being a simple scalar value, i.e. distance, use of ADFs introduced simplicity both in data structure and processing algorithm in our prototype implementation. This enabled better simulation performances compared with conventional B-Rep based simulators. The adaptiveness of ADFs also achieved the desired balance between model precision and data size.
