Hierarchical Segmentation of Automotive Surfaces and Fast Marching Methods.

Prasad Atkar

Abstract

Complete automation of trajectory planning tools for material deposition/removal applications has become increasingly necessary to reduce the ``concept-to-consumer" timeline in product development. Robotic spray painting is such an application where it is necessary to generate spray gun trajectories in such a way that the constraint on the uniformity of paint deposition is satisfied along with the minimization of cycle time and paint waste. To meet these output criteria, it is necessary to segment a complex automobile body into geometrically simple and topologically monotonic cells. By employing tools from differential geometry, we characterize the relationships of input parameters and output characteristics to solve the segmentation problem. In this talk, I will describe procedures to segment a surface embedded in R^3 based on the interaction of the material deposition pattern with the surface geometry, surface topology and the geometry of the resultant passes.

Our trajectory generation tools rely on offsetting curves sideways within the surface. As a detour from the segmentation procedures, I will briefly describe the Fast Marching Methods (developed by Sethian) that track the motion of complex fronts on a curved surface, where the fronts can develop sharp corners and change topology as they evolve on the surface.

The URL to Sethian's book.

Presentation Slides

Ppt (2.3Mb)

 

Speaker Appointments

For appointments, please contact Prasad Atkar (prasada@andrew.cmu.edu).