Approximate POMDP Planning Software
Our team won the POMDP track at the
ICAPS 2011 International Probabilistic
Planning Competition (IPPC
source code is available for download
. It uses SARSOP together with
APPL is a C++ toolkit for approximate POMDP planning. Originally, it is based on the SARSOP algorithm  for solving discrete POMDPs. Over the time, it has evolved and now consists of three packages:
- APPL Offline implements the SARSOP algorithm [1,2] for solving discrete POMDPs. APPL Offline has already been used by many people all over the world. It takes as input a POMDP model in the POMDP or POMDPX file format and produces a policy file.
- APPL Online implements the DESPOT algorithm for online POMDP planning . Online POMDP planning enables us to scale up and handle POMDP models too large for offline POMDP policy computation.
- APPL Continuous implements the MCVI algorithm for solving continuous POMDPs. It takes as input a POMDP model (coded in C++) and produce a policy file. It also contains a simple simulator for evaluating the quality of the computed policy.
Although all experimental software, the maturity of these three packages differ. While APPL Offline is well tested and largely stable, the other two packages are our latest efforts to expand the capability of APPL. At the moment, the three packages are independent. Maybe one day they will be merged in a single coherent framework. For bug reports and suggestions, please email email@example.com.
- H. Kurniawati, D. Hsu, and W.S. Lee. SARSOP: Efficient point-based POMDP planning by approximating optimally reachable belief spaces. In Proc. Robotics: Science and Systems, 2008.
- S.C.W. Ong, S.W. Png, D. Hsu, and W.S. Lee. POMDPs for robotic tasks with mixed observability. In Proc. Robotics: Science and Systems, 2009.
- A. Somani, N. Ye, D. Hsu, and W.S. Lee. DESPOT: online POMDP planning with regularization. In Advances in Neural Information Processing Systems (NIPS). 2013.
- H. Bai, D. Hsu, M. Kochenderfer, and W.S. Lee. Unmanned Aircraft Collision Avoidance using Continuous-State POMDPs. Proc. Robotics: Science and Systems, 2011.
- Z.W. Lim, D. Hsu, and W.S. Lee. Monte Carlo Value Iteration with Macro-Actions. Proc. Neural Information Processing Systems (NIPS), 2011.
| || || |
| ||integrated exploration|| |
This page has been viewed 23511 times.