public class BeliefMDPGenerator extends java.lang.Object implements DomainGenerator
PODomain
and turning it into
a BeliefMDP, which can then be input to any MDP solver to solve the POMDP. To get the belief MDP reward function,
create and instance of the static inner class BeliefMDPGenerator.BeliefRF
.
BeliefState
and EnumerableBeliefState
interfaces. The getTransitions method (used by planning algorithms that require the full transition dynamics)
also operates by iterating over all possible observations. If your domain has many observations, consider using
a sample-based MDP planning algorithm.
Modifier and Type | Class and Description |
---|---|
class |
BeliefMDPGenerator.BeliefAction
A Belief MDP action.
|
static class |
BeliefMDPGenerator.BeliefRF
A class for turning a POMDP reward function into a Belief MDP reward function.
|
static class |
BeliefMDPGenerator.GroundedBeliefAction
A
GroundedAction implementation for a Belief MDP that curries the
GroundedAction for the underlying POMDP. |
static class |
BeliefMDPGenerator.ObjectParameterizedGroundedBeliefAction
A
GroundedAction implementation for a Belief MDP that curries
an AbstractObjectParameterizedGroundedAction GroundedAction for the underlying POMDP. |
Modifier and Type | Field and Description |
---|---|
protected PODomain |
podomain
The input POMDP domain
|
Constructor and Description |
---|
BeliefMDPGenerator(PODomain podomain)
Initializes
|
Modifier and Type | Method and Description |
---|---|
Domain |
generateDomain()
Returns a newly instanced Domain object
|
protected PODomain podomain
public BeliefMDPGenerator(PODomain podomain)
podomain
- the input POMDP domain that will be turned into a Belief MDP.public Domain generateDomain()
DomainGenerator
generateDomain
in interface DomainGenerator