Parallelization of the Value-Iteration algorithm for Partially Observable Markov Decision Processes |
Dennis Noer
|
| Abstract | Partially observable Markov decision processes (POMDP) is a strong framework for mission planning in a partially observable and stochastic environment. To determine the solution of a POMDP, algorithms with a running time in the PSPACE-Hard area are utilized. In this project, we will explore the potential for reducing the computation time using the massive parallel processing power of modern Graphic Processing Units (GPU). Our experiments show that the GPU platform can accelerate a PBVI based POMDP solver, while more advanced synchronization features are needed to exploit the full potential. |
| Type | Bachelor thesis [Academic thesis] |
| Year | 2013 |
| Publisher | Technical University of Denmark, Department of Applied Mathematics and Computer Science / DTU Co |
| Address | Matematiktorvet, Building 303B, DK-2800 Kgs. Lyngby, Denmark, compute@compute.dtu.dk |
| Series | B.Sc.-2013-31 |
| Note | |
| Electronic version(s) | [pdf] |
| Publication link | http://www.compute.dtu.dk/English.aspx |
| BibTeX data | [bibtex] |
| IMM Group(s) | Computer Science & Engineering |