The Department of Energy desires to increase radioactive waste treatment automation in order to minimize radiation worker exposure. Many treatment processes require maintaining tool orientation and surface distance including inspection and dismantling tasks. This paper expands on previous shape primitive virtual fixture generation techniques to increase their expressiveness based on surface models. A robotic hardware-agnostic virtual fixture generation pipeline is constructed and tested with varied superellipsoid and supertoroid models. The pipeline converts surface models into a bi-directional graphs representing layers of offset guidance virtual fixtures. Surface models function as forbidden region virtual fixtures and are combined with the graph structure into task virtual fixtures. Results show a correlation between distance to the surface and graph layer vertices. Surface model concavity also affects the growth of offset layer vertices. Incorrect model surface normals do result in irregular graph layers and the pipeline has been tested on CAD models.

This content is only available via PDF.
You do not currently have access to this content.