Current Research
Parallel Computing Based Automation of Machine Tool Path Planning
Principal investigator:
Thomas Kurfess, Professor and BMW Chair of Manufacturing
Department of Mechanical Engineering
Director, Automotive Engineering
Clemson University - International Center for Automotive Research
Graduate student(s):
Joshua Tarbutton
Dima Konobrytskyi
Sponsor:
National Science Foundation, Okuma, Tucker Innovations
Brief abstract:
Significant research has been accomplished in various machining areas but there has been no generally accepted method by which to automate tool path planning for multi-axis machine tools. The goal of this research is to perform automatic path planning for multi-axis machine tools by generating new algorithms that exploit the latest graphics hardware and parallel programming technologies. This research is focused on identifying the obstacles to accomplishing this task and providing a framework by which to solve them.
The inputs to the automated path planning algorithm are a mesh part model, blank model, fixture model, machine parameters, and cutting tool information. From these inputs, an algorithm that exploits the native Graphics Processing Unit (GPU) functionality (e.g. the depth buffer), parallel processing technology (e.g. CUDA and OpenCL), and graphics concepts (e.g. ray-casting), generates a family of tool paths based on desired cost, quality, and machining time trade-offs. As part of this research a real time simulator is being developed to evaluate the effectiveness of the algorithm before actual test machining. G-codes are generated as a result of the algorithm and the tool paths produced are tested on actual machine tools.Currently, roughing paths are generated by using a contour parallel offset (CPO) approach and knowledge of the available cutters. Semi-finish and finishing paths are generated by using a z-offset approach. The various stages of the tool path are automatically generated by knowledge of the current state of the part, blank, and cutter information. The cutter contact (CC) points are adjusted in a parallel gouge prevention algorithm implemented in OpenCL for extremely accurate and fast cutter location (CL) determination.
The outcome of this research will be a software tool that manufacturing personnel can use to automatically generate tool paths that take into account all the available cutters, the machine parameters, and their relative preference in terms of cost, quality, and machining time.
Successful completion of this research will have a tremendous impact on manufacturing. This research will provide a tool for manufacturing personnel to automatically generate tool paths for complex parts resulting in significant savings of limited resources.
 |
 |
 |
Figure 1: Tiger Paw - Part, Simulattor and Machining Result |
||
Impact:
Manufacturing engineers will benefit from this research because it has the potential to eliminate the time consuming process on manually creating and linking tool paths.
Project schedule:
5/2009-5/2010 Phase I: Cutting foundations and specific cutting strategy implementation
6/2010- 10/2010 Phase II: Integration of strategies, trade-offs, and simulator
11/2010-4/2011 Phase III: Summary of automatic path generation ca
Publications:
Tarbutton, J, T.R. Kurfess, and T.M. Tucker, “Graphics Based Path Planning for Multi-Axis Machine Tools”. CAD and Applications. Accepted.
Tarbutton, J, T.R. Kurfess, and T.M. Tucker, “Machining by Ray Casting into voxel models”. International Symposium on Flexible Automation. Accepted.
Carter, J.A.; T.M. Tucker; T.R. Kurfess, "3-Axis CNC Path Planning Using Depth Buffer and Fragment Shader". Computer Aided Design and Applications, 5(5), 2008, 612-621.
