Re: [ompl-users] PRM and RRT sample bounds

[Mark M. <mm...@ri...>]

You can create a Goal-derived class where the isSatisfied method always returns false. Use this in combination with the ompl::base::IterationTerminationCondition class to call solve. Next, set the goal you want and call solve again with ompl::base::plannerAlwaysTerminatingCondition(), which makes the planner terminate immediately. The internal data is not cleared between repeated calls to solve, so I think this might work.

Mark



> On Oct 5, 2017, at 11:55 AM, James Lee <JuH...@st...> wrote:
> 
> Thanks for that Mark. Its not exactly what I needed, however.
> 
> What I want to do is find a solution using a roadmap or branch that contains 1000 sample states.
> But when I use the iteration termination condition to and set the value to, say 1000, the algorithm seems to only generate ~480 states.
> I also want to perform benchmark test using set number of sample states, so ideally I don't want to pass the object through Planner::solve().
> 
> Is there another way? or do I need to hard code this?
> 
> James
> 
> From: Mark Moll <mar...@gm... <mailto:mar...@gm...>> on behalf of Mark Moll <mm...@ri... <mailto:mm...@ri...>>
> Sent: Thursday, 5 October 2017 7:39:41 PM
> To: James Lee
> Cc: omp...@li... <mailto:omp...@li...>; Chanyeol Yoo; Robert Fitch
> Subject: Re: [ompl-users] PRM and RRT sample bounds
> 
> Hi James,
> 
> Perhaps this is what you want:
> ompl::base::IterationTerminationCondition
> http://ompl.kavrakilab.org/classompl_1_1base_1_1IterationTerminationCondition.html <http://ompl.kavrakilab.org/classompl_1_1base_1_1IterationTerminationCondition.html>
> ompl::base::IterationTerminationCondition Class Reference <http://ompl.kavrakilab.org/classompl_1_1base_1_1IterationTerminationCondition.html>
> ompl.kavrakilab.org <http://ompl.kavrakilab.org/>
> Public Member Functions IterationTerminationCondition (unsigned int numIterations) Construct a termination condition that can be evaluated numIterations times before ...
> 
> You can pass an object of this type to Planner::solve().
> 
> Mark
> 
> 
> 
>> On Oct 5, 2017, at 5:39 AM, James Lee <JuH...@st... <mailto:JuH...@st...>> wrote:
>> 
>> Hi everyone,
>> 
>> I am trying to implement PRM and RRT (and their respective variants) on OMPL.
>> 
>> Is there a way to add a sample bound to these algorithms? For example, the roadmap or branch is grown until the 1000th sample has been generated, and uses said roadmap or branches to output a trajectory/path.
>> 
>> Regards,
>> 
>> James
> 
>