A default rejection sampling scheme that samples uniformly from the entire planning domain. Samples are rejected until one is found that has a heuristic solution estimate that is less than the current solution. In general, direct sampling of the informed subset is much better, but this is a general default. More...
#include <ompl/base/samplers/informed/RejectionInfSampler.h>

Public Member Functions | |
RejectionInfSampler (const ProblemDefinitionPtr &probDefn, unsigned int maxNumberCalls) | |
Construct a rejection sampler that only generates states with a heuristic solution estimate that is less than the cost of the current solution. | |
bool | sampleUniform (State *statePtr, const Cost &maxCost) override |
Sample uniformly in the subset of the state space whose heuristic solution estimates are less than the provided cost, i.e. in the interval [0, maxCost). Returns false if such a state was not found in the specified number of iterations. | |
bool | sampleUniform (State *statePtr, const Cost &minCost, const Cost &maxCost) override |
Sample uniformly in the subset of the state space whose heuristic solution estimates are between the provided costs, [minCost, maxCost). Returns false if such a state was not found in the specified number of iterations. | |
bool | hasInformedMeasure () const override |
Whether the sampler can provide a measure of the informed subset. | |
double | getInformedMeasure (const Cost &) const override |
The measure of the subset of the state space defined by the current solution cost that is being searched. As rejection sampling has no closed-form knowledge of the informed subset, the measure of the informed space is always the measure of the entire space. | |
double | getInformedMeasure (const Cost &, const Cost &) const override |
The measure of the subset of the state space defined by the current solution cost that is being searched. As rejection sampling has no closed-form knowledge of the informed subset, the measure of the informed space is always the measure of the entire space. | |
![]() | |
InformedSampler (const InformedSampler &)=delete | |
InformedSampler & | operator= (const InformedSampler &)=delete |
InformedSampler (const ProblemDefinitionPtr &probDefn, unsigned int maxNumberCalls) | |
Construct a sampler that only generates states with a heuristic solution estimate that is less than the cost of the current solution. Requires a function pointer to a method to query the cost of the current solution. If iteration is required, only maxNumberCalls are attempted, to assure that the function returns. | |
virtual bool | sampleUniform (State *statePtr, const Cost &maxCost)=0 |
Sample uniformly in the subset of the state space whose heuristic solution estimates are less than the provided cost, i.e. in the interval [0, maxCost). Returns false if such a state was not found in the specified number of iterations. | |
virtual bool | sampleUniform (State *statePtr, const Cost &minCost, const Cost &maxCost)=0 |
Sample uniformly in the subset of the state space whose heuristic solution estimates are between the provided costs, [minCost, maxCost). Returns false if such a state was not found in the specified number of iterations. | |
virtual bool | hasInformedMeasure () const =0 |
Whether the sampler can provide a measure of the informed subset. | |
virtual double | getInformedMeasure (const Cost ¤tCost) const =0 |
The measure of the subset of the state space defined by the current solution cost that is being searched. Does not consider problem boundaries but returns the measure of the entire space if no solution has been found or if a closed form expression for the measure does not exist. | |
virtual double | getInformedMeasure (const Cost &minCost, const Cost &maxCost) const |
The measure of the subset of the state space defined by the current solution cost that is being searched. Does not consider problem boundaries but returns the measure of the entire space if no solution has been found or if a closed form expression for the measure does not exist. By default calls the 1-argument overloaded version with the min and max costs and subtracts the differences; however, there may be more efficient ways to do this for some cost functions. | |
virtual Cost | heuristicSolnCost (const State *statePtr) const |
A helper function to calculate the heuristic estimate of the solution cost for a given state using the optimization objective stored in the problem definition. | |
ProblemDefinitionPtr | getProblemDefn () const |
unsigned int | getMaxNumberOfIters () const |
Additional Inherited Members | |
![]() | |
ProblemDefinitionPtr | probDefn_ |
A copy of the problem definition. | |
OptimizationObjectivePtr | opt_ |
A copy of the optimization objective. | |
StateSpacePtr | space_ |
A copy of the state space. | |
unsigned int | numIters_ |
The number of iterations I'm allowed to attempt. | |
Detailed Description
A default rejection sampling scheme that samples uniformly from the entire planning domain. Samples are rejected until one is found that has a heuristic solution estimate that is less than the current solution. In general, direct sampling of the informed subset is much better, but this is a general default.
Definition at line 53 of file RejectionInfSampler.h.
Constructor & Destructor Documentation
◆ RejectionInfSampler()
ompl::base::RejectionInfSampler::RejectionInfSampler | ( | const ProblemDefinitionPtr & | probDefn, |
unsigned int | maxNumberCalls | ||
) |
Construct a rejection sampler that only generates states with a heuristic solution estimate that is less than the cost of the current solution.
Definition at line 45 of file RejectionInfSampler.cpp.
Member Function Documentation
◆ getInformedMeasure() [1/2]
|
overridevirtual |
The measure of the subset of the state space defined by the current solution cost that is being searched. As rejection sampling has no closed-form knowledge of the informed subset, the measure of the informed space is always the measure of the entire space.
Implements ompl::base::InformedSampler.
Definition at line 107 of file RejectionInfSampler.cpp.
◆ getInformedMeasure() [2/2]
|
overridevirtual |
The measure of the subset of the state space defined by the current solution cost that is being searched. As rejection sampling has no closed-form knowledge of the informed subset, the measure of the informed space is always the measure of the entire space.
Reimplemented from ompl::base::InformedSampler.
Definition at line 112 of file RejectionInfSampler.cpp.
◆ hasInformedMeasure()
|
overridevirtual |
Whether the sampler can provide a measure of the informed subset.
Implements ompl::base::InformedSampler.
Definition at line 102 of file RejectionInfSampler.cpp.
◆ sampleUniform() [1/2]
|
overridevirtual |
Sample uniformly in the subset of the state space whose heuristic solution estimates are less than the provided cost, i.e. in the interval [0, maxCost). Returns false if such a state was not found in the specified number of iterations.
Implements ompl::base::InformedSampler.
Definition at line 60 of file RejectionInfSampler.cpp.
◆ sampleUniform() [2/2]
|
overridevirtual |
Sample uniformly in the subset of the state space whose heuristic solution estimates are between the provided costs, [minCost, maxCost). Returns false if such a state was not found in the specified number of iterations.
Implements ompl::base::InformedSampler.
Definition at line 70 of file RejectionInfSampler.cpp.
The documentation for this class was generated from the following files:
- ompl/base/samplers/informed/RejectionInfSampler.h
- ompl/base/samplers/informed/src/RejectionInfSampler.cpp