: A termination condition for stopping an optimizing planner based on cost convergence More...

Inheritance diagram for ompl::base::CostConvergenceTerminationCondition:

Public Member Functions
	CostConvergenceTerminationCondition (ProblemDefinitionPtr &pdef, size_t solutionsWindow=10, double epsilon=0.1)
	Constructor. More...

void	processNewSolution (const Cost solutionCost)

Public Member Functions inherited from ompl::base::PlannerTerminationCondition
	PlannerTerminationCondition (const PlannerTerminationConditionFn &fn)
	Construct a termination condition. By default, eval() will call the externally specified function fn to decide whether the planner should terminate.

	PlannerTerminationCondition (const PlannerTerminationConditionFn &fn, double period)
	Construct a termination condition that is evaluated every period seconds. The evaluation of the condition consists of calling fn() in a separate thread. Calls to eval() will always return the last value computed by the call to fn().

bool	operator() () const
	Return true if the planner should stop its computation.

	operator bool () const
	Cast as true if the planner should stop its computation.

void	terminate () const
	Notify that the condition for termination should become true, regardless of what eval() returns. This function may be called while the condition is being evaluated by other threads.

bool	eval () const
	The implementation of some termination condition. By default, this just calls fn_()

Detailed Description

: A termination condition for stopping an optimizing planner based on cost convergence

Constructor & Destructor Documentation

ompl::base::CostConvergenceTerminationCondition::CostConvergenceTerminationCondition	(	ProblemDefinitionPtr &	pdef,
		size_t	solutionsWindow = `10`,
		double	epsilon = `0.1`
	)

Constructor.

Parameters

pdef	Problem definition, needed to get access to the optimization objective and to set a callback to get intermediate solutions.
solutionsWindow	Minimum number of solutions to use in deciding whether a planner has converged.
epsilon	Threshold to consider for convergence. This should be a positive number close to 0. If the cumulative moving average does not change by a relative fraction of epsilon after a new better solution is found, convergence has been reached and a planner should terminate..

The documentation for this class was generated from the following files: