Tim Quatmann
|
ae9360af03
|
Use the new MakeStateSetObservationClosed transformer for the belief-exploration based pomdp model checker
|
5 years ago |
|
Tim Quatmann
|
ee350ca384
|
Use same precision as BeliefValueType when dealing with triangulation resolutions.
|
5 years ago |
|
Tim Quatmann
|
55c4408c6a
|
Storing the observation resolutions as a float so that we can increase the resolution more accurately with a non-integer factor
|
5 years ago |
|
Tim Quatmann
|
2f2a007896
|
Implemented 'guessing' of initial pomdp schedulers for multiple guesses
|
5 years ago |
|
TimQu
|
7504f6f315
|
Improved statistics output for refinements, added detection of fixpoints
|
5 years ago |
|
Tim Quatmann
|
1763f0c582
|
Making sure that we only store the best bounds found so far. Also added some output for the resulting values in each iteration.
|
5 years ago |
|
Tim Quatmann
|
fa624d2a20
|
Introduced new settings for controlling the refinement strategy and whether to produce only upper and/or lower bounds
|
5 years ago |
|
Tim Quatmann
|
37490a8eca
|
Started to integrate new refinement options.
|
5 years ago |
|
Tim Quatmann
|
75d792e987
|
Implemented refinement heuristic.
|
5 years ago |
|
Tim Quatmann
|
03889958da
|
Added a switch to control the size of the under-approximation via command line.
|
5 years ago |
|
Tim Quatmann
|
c2837bb749
|
ApproximatePOMDPModelchecker: Improved output a bit.
|
5 years ago |
|
Tim Quatmann
|
c3847d05af
|
Scaling the rating of an observation with the current resolution.
|
5 years ago |
|
Tim Quatmann
|
c2ddea1480
|
First (re-) implementation of refinement. (probably needs some testing/debugging)
|
5 years ago |
|
Tim Quatmann
|
71e0654498
|
Changed method signatures to new data structures.
|
5 years ago |
|
Tim Quatmann
|
8b0e582ef4
|
Use the new BeliefMdpExplorer also for the underapproximation.
|
5 years ago |
|
Tim Quatmann
|
ab26b69435
|
Added BeliefMdpExplorer which does most of the work when exploring (triangulated Variants of) the BeliefMdp.
|
5 years ago |
|
Tim Quatmann
|
0b552e6813
|
Renamed BeliefGrid to BeliefManager
|
5 years ago |
|
Tim Quatmann
|
d184d67b53
|
Refactored under-approximation code a bit.
|
5 years ago |
|
Tim Quatmann
|
6fee61feb1
|
POMDP: Started to split belief logic from exploration logic.
|
5 years ago |
|
Tim Quatmann
|
e76efd14d5
|
POMDP: Filling the statistics struct with information. Also incorporated aborting (SIGTERM, i.e. CTRL+C)
|
5 years ago |
|
Tim Quatmann
|
6f3fab8e80
|
Added a statistics struct to the approximatePOMDP model checker
|
5 years ago |
Alexander Bork
|
311362d995
|
Removal of some more obsolete code
|
5 years ago |
|
Alexander Bork
|
0507da4ffa
|
Adjusted Refinement Procedure for rewards
|
5 years ago |
|
Alexander Bork
|
8e30e27eb9
|
Removal of obsolete code
|
5 years ago |
|
Tim Quatmann
|
b3493b5888
|
Grid: Added cli setting to cache subsimplices.
|
5 years ago |
|
Tim Quatmann
|
a11ec691a9
|
Introduced options in the ApproximatePOMDPModelChecker.
|
5 years ago |
|
Alexander Bork
|
c2582058c9
|
Added first version of refinement with reuse of previous results
|
5 years ago |
|
Alexander Bork
|
c74a6ea201
|
Removal of unused extractActions function
|
5 years ago |
|
Alexander Bork
|
e1bd87b91a
|
Added documentation
|
5 years ago |
|
Alexander Bork
|
53d5e9edf5
|
Code Cleanup
|
5 years ago |
|
Alexander Bork
|
91232a2b11
|
Changed data structure for belief distributions from vector to map to exploit sparsity
|
5 years ago |
|
Alexander Bork
|
e02ea57a58
|
Rudimentary version of the refinement loop
|
5 years ago |
|
Alexander Bork
|
a5fc9fc5b7
|
Added naive underapproximation by unrolling the belief support up to a given size
|
5 years ago |
|
Alexander Bork
|
4bbb02dcaa
|
Wrong method for underapproximation for future reference
|
5 years ago |
|
Alexander Bork
|
8d3254b8d0
|
Use boost::bimap for the belief space <-> state space mapping
|
5 years ago |
|
Alexander Bork
|
0facf4a572
|
Preparation work for the implementation of the refinement procedure
|
5 years ago |
|
Alexander Bork
|
94b93f013c
|
Added option to stop approximation space exploration early if difference between over- and under-approximation is under a given threshold
|
5 years ago |
|
Alexander Bork
|
f416cc8291
|
Added flag to toggle caching of subsimplex and lambda values
|
5 years ago |
|
Alexander Bork
|
aca676a0a5
|
Added model generation and checking for initial approximation bounds
|
5 years ago |
|
Alexander Bork
|
8992b70da3
|
Made Value Iteration its own function to reduce duplicate code
|
5 years ago |
|
Alexander Bork
|
fe81e0d7cf
|
Smaller touch-ups (Removal of unused code, pass-by-reference)
|
5 years ago |
|
Alexander Bork
|
877c15ed43
|
Removed obsolete function to create transition matrices from a data structure not used anymore
|
5 years ago |
|
Alexander Bork
|
b7b213571d
|
Refactoring of underapproximation procedures to reduce code duplication
|
5 years ago |
|
Alexander Bork
|
8f81958268
|
Refactoring of reachability reward and probability methods to reduce code duplication
|
5 years ago |
|
Alexander Bork
|
4664b4244b
|
Refactoring of on-the-fly computation to reduce code duplication
|
5 years ago |
|
Alexander Bork
|
bbd3ec7287
|
Fix of wrong MDP underapproximation
|
5 years ago |
|
Alexander Bork
|
b9c0af6628
|
Added on-the-fly belief grid generation for rewards
|
6 years ago |
|
Alexander Bork
|
21e417bdac
|
Added on-the-fly belief grid generation to avoid computations for unreachable beliefs
|
6 years ago |
|
Alexander Bork
|
bc52aa86ca
|
Added procedure to repeat probability computation with higher resolution
|
6 years ago |
|
Alexander Bork
|
4b8664c521
|
Added reward under-approximation
|
6 years ago |
