SUBJECT: Ph.D. Proposal Presentation
BY: David Lynn
TIME: Monday, July 31, 2017, 3:00 p.m.
PLACE: Love Building, 210
TITLE: Direct Servo Control of Positional Derivatives for 5-axis CNC Machine Tools Using Densely-Sampled Toolpaths
COMMITTEE: Dr. Thomas Kurfess, Chair (ME)
Dr. Christopher Saldana (ME)
Dr. Tommy Tucker (Industry)
Dr. Cassandra Telenko (ME)
Dr. Jarek Rossignac (CS)
Dr. Moneer Helu (NIST)


The representation of volumetric geometry in a discrete sense can be accomplished through the use of a voxel model, which consists of many small cubes that are used to make up the volume. The use of voxel models in computer-aided manufacturing (CAM) operations enables planning of toolpaths for highly complex surfaces that would be infeasible with traditional analytical modeling approaches. However, generation of toolpaths for CNC machining from voxel models creates many small linear movements that are difficult for a machine tool control system to process at high speed. The machine controller cannot create motion profiles for the movements quickly enough to maintain the tool velocity at the programmed level when the size of the movements is sufficiently small. This dissertation seeks to develop motion control schemes that enable rapid processing of high density toolpaths without compromising tool velocity. Instead of supplying traditional G-Code to the controller, the CAM system will perform all motion profile generation and supply timestamped positional derivative commands directly to the servo axes of the machine tool. A prototype implementation will be performed using an open-source CNC control system, MachineKit, and will provide a path to generalize the approach to any machine tool control platform. The effects of higher-order positional derivative (jerk and snap) continuity along the motion profile will be investigated as a means to smooth the motion of the machine tool. Additionally, implementation guidelines will be developed for controlling the machine from a cloud computing platform that has already been implemented in prior work.