This class is the root node of the rPOMCP graph. More...
#include <AIToolbox/POMDP/Algorithms/Utils/rPOMCPGraph.hpp>
Public Member Functions | |
| HeadBeliefNode (size_t A, RandomEngine &rand) | |
| Basic constructor. More... | |
| HeadBeliefNode (size_t A, size_t beliefSize, const AIToolbox::POMDP::Belief &b, RandomEngine &rand) | |
| Constructor from belief. More... | |
| HeadBeliefNode (size_t A, BeliefNode< UseEntropy > &&bn, RandomEngine &rand) | |
| Constructor from BeliefNode. More... | |
| bool | isSampleBeliefEmpty () const |
| Whether we have no particles in the sampling belief. More... | |
| size_t | sampleBelief () const |
| Samples the internal sampling belief. More... | |
| size_t | getMostCommonParticle () const |
| Useful if the agents wants a guess of what the current state is. More... | |
 Public Member Functions inherited from AIToolbox::POMDP::BeliefNode< UseEntropy > | |
| BeliefNode () | |
| void | updateBeliefAndKnowledge (size_t s) |
| This function updates the knowledge measure after adding a new belief particle. More... | |
| double | getKnowledgeMeasure () const |
| This function returns the current estimate for reward for this node. More... | |
| void | updateBeliefAndKnowledge (const size_t s) |
| void | updateBeliefAndKnowledge (const size_t s) |
Additional Inherited Members | |
| Public Attributes inherited from AIToolbox::POMDP::BeliefNode< UseEntropy > | |
| unsigned | N |
| Counter for number of times we went through this belief node. More... | |
| ActionNodes< UseEntropy > | children |
| double | V |
| Estimated value for this belief, taking into account future rewards/actions. More... | |
| double | actionsV |
| Estimated value for the actions (could be mean, max, or other) More... | |
| size_t | bestAction |
| Tracker of best available action in MAX-mode, to select node value. More... | |
| Protected Attributes inherited from AIToolbox::POMDP::BeliefNode< UseEntropy > | |
| TrackBelief< UseEntropy > | trackBelief_ |
| This is a particle belief which is easy to update. More... | |
| double | knowledgeMeasure_ |
| Estimated entropy/max-belief for this node. More... | |
Detailed Description
template<bool UseEntropy>
class AIToolbox::POMDP::HeadBeliefNode< UseEntropy >
This class is the root node of the rPOMCP graph.
This converts the unordered belief map of an ordinary belief node into a vector. This should speed up the sampling process considerably, since the head node is the one that gets sampled the most.
Note that for this reason this node does not use the trackBelief_ field. It uses the sampleBelief_ instead.
Constructor & Destructor Documentation
◆ HeadBeliefNode() [1/3]
| AIToolbox::POMDP::HeadBeliefNode< UseEntropy >::HeadBeliefNode | ( | size_t | A, |
| RandomEngine & | rand | ||
| ) |
Basic constructor.
◆ HeadBeliefNode() [2/3]
| AIToolbox::POMDP::HeadBeliefNode< UseEntropy >::HeadBeliefNode | ( | size_t | A, |
| size_t | beliefSize, | ||
| const AIToolbox::POMDP::Belief & | b, | ||
| RandomEngine & | rand | ||
| ) |
Constructor from belief.
This constructor samples the input belief in order to create the sample belief. We create beliefSize samples.
We don't store the belief directly since the sampleBelief must also be constructible from the particle beliefs of normal belief nodes (when we use rPOMCP for multiple timesteps). So we use a particle belief in both cases.
◆ HeadBeliefNode() [3/3]
| AIToolbox::POMDP::HeadBeliefNode< UseEntropy >::HeadBeliefNode | ( | size_t | A, |
| BeliefNode< UseEntropy > && | bn, | ||
| RandomEngine & | rand | ||
| ) |
Constructor from BeliefNode.
This constructor converts the input BeliefNode into a new head node. It converts the track belief of the node into our sample belief.
Member Function Documentation
◆ getMostCommonParticle()
| size_t AIToolbox::POMDP::HeadBeliefNode< UseEntropy >::getMostCommonParticle |
Useful if the agents wants a guess of what the current state is.
◆ isSampleBeliefEmpty()
| bool AIToolbox::POMDP::HeadBeliefNode< UseEntropy >::isSampleBeliefEmpty |
Whether we have no particles in the sampling belief.
◆ sampleBelief()
| size_t AIToolbox::POMDP::HeadBeliefNode< UseEntropy >::sampleBelief |
Samples the internal sampling belief.
The documentation for this class was generated from the following file:
- include/AIToolbox/POMDP/Algorithms/Utils/rPOMCPGraph.hpp