Providing an understanding of space in game and simulation environments is one of the major challenges associated with moving artificially intelligent characters through these environments. The usage of some form of navigation mesh has become the standard method to provide a representation of the walkable space in game environments to characters moving around in that environment. There is currently no standardized best method of producing a navigation mesh. In fact, producing an optimal navigation mesh has been shown to be an NP-Hard problem. Current approaches are a patchwork of divergent methods all of which have issues either in the time to create the navigation meshes (e.g., the best looking navigation meshes have traditionally been produced by hand which is time consuming), generate substandard quality navigation meshes (e.g., many of the automatic mesh production algorithms result in highly triangulated meshes that pose problems for character navigation), or yield meshes that contain gaps of areas that should be included in the mesh and are not (e.g., existing growth-based methods are unable to adapt to non-axis-aligned geometry and as such tend to provide a poor representation of the walkable space in complex environments).
We introduce the Planar Adaptive Space Filling Volumes (PASFV) algorithm, Volumetric Adaptive Space Filling Volumes (VASFV) algorithm, and the Iterative Wavefront Edge Expansion Cell Decomposition (Wavefront) algorithm. These algorithms provide growth-based spatial decompositions for navigation mesh generation in either 2D (PASFV) or 3D (VASFV). These algorithms generate quick (on demand) decompositions (Wavefront), use quad/cube base spatial structures to provide more regular regions in the navigation mesh instead of triangles, and offer full coverage decompositions to avoid gaps in the navigation mesh by adapting to non-axis-aligned geometry. We have shown experimentally that the decompositions offered by PASFV and VASFV are superior both in character navigation ability, number of regions, and coverage in comparison to the existing and commonly used techniques of Space Filling Volumes, Hertel-Melhorn decomposition, Delaunay Triangulation, and Automatic Path Node Generation. Finally, we show that our Wavefront algorithm retains the superior performance of the PASFV and VASFV algorithms while providing faster decompositions that contain fewer degenerate and near degenerate regions.
Unlike traditional navigation mesh generation techniques, the PASFV and VASFV algorithms have a real time extension (Dynamic Adaptive Space Filling Volumes, DASFV) which allows the navigation mesh to adapt to changes in the geometry of the environment at runtime.
In addition, it is possible to use a navigation mesh for applications above and beyond character path planning and navigation. These multiple uses help to increase the return on the investment in creating a navigation mesh for a game or simulation environment. In particular, we will show how to use a navigation mesh for the acceleration of collision detection.
|Advisor:||Youngblood, G. Michael|
|Commitee:||Akella, Srinivas, Barnes, Tiffany, Hetyei, Gabor, Subramanian, Kalpathi|
|School:||The University of North Carolina at Charlotte|
|Department:||Information Technology (PhD)|
|School Location:||United States -- North Carolina|
|Source:||DAI-B 73/05, Dissertation Abstracts International|
|Keywords:||Agent navigation, Artificial intelligence, Navigation meshes, Pathfinding, Spatial decomposition|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
supplemental files is subject to the ProQuest Terms and Conditions of use.