Something wrong when try to creat new tree in RRT

31 views
Skip to first unread message

jumping

unread,
Aug 3, 2017, 11:44:03 AM8/3/17
to PR2_Users
hello, 

I've been working on a project in which i need to using two trees in RRT. For easy to understand, I just modify a few code as shown attached.  The OMPL compiled successfully. But when comes to Moveit, I am getting this error：

main: malloc.c:2372: sysmalloc: Assertion `(old_top == (((mbinptr) (((char *) &((av)->bins[((1) - 1) * 2])) - __builtin_offsetof (struct malloc_chunk, fd)))) && old_size == 0) || ((unsigned long) (old_size) >= (unsigned long)((((__builtin_offsetof (struct malloc_chunk, fd_nextsize))+((2 *(sizeof(size_t))) - 1)) & ~((2 *(sizeof(size_t))) - 1))) && ((old_top)->size & 0x1) && ((unsigned long) old_end & pagemask) == 0)' failed.
  
It seems that this error occurred only when I add a new tree, So I wonder how to creat a new tree successfully in RRT.

If there is someone out there who will help me, i thank him alot in advance :)
RRT.cpp
RRT.h
Screenshot from 2017-08-0322_51_26.png

Devon Ash

unread,
Aug 3, 2017, 11:59:50 AM8/3/17
to jumping, PR2_Users
I believe I've seen this before. Try re-installing your move_group debian as it may have some corrupted bits somewhere.

--
You received this message because you are subscribed to the Google Groups "PR2_Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pr2_users+unsubscribe@googlegroups.com.
To post to this group, send email to pr2_...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pr2_users/7ce7fe26-e062-4443-a9fc-d41f17d8ed8d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

jumping

unread,
Aug 4, 2017, 3:27:52 AM8/4/17
to PR2_Users, 2435...@qq.com
Hello，Devon Ash

Thank you for your kind help.
  
It seems that this error occurred only when I add a new tree. I mean if I have not modified the RRT.h and RRT.cpp, Moveit works. When I try to add a new tree as shown attached, this error will occur and I try this in Ros Indigo and Kinect.


On Thursday, August 3, 2017 at 11:59:50 PM UTC+8, Devon Ash wrote:
I believe I've seen this before. Try re-installing your move_group debian as it may have some corrupted bits somewhere.
On Thu, Aug 3, 2017 at 11:44 AM, jumping <2435...@qq.com> wrote:
hello, 

I've been working on a project in which i need to using two trees in RRT. For easy to understand, I just modify a few code as shown attached.  The OMPL compiled successfully. But when comes to Moveit, I am getting this error：

main: malloc.c:2372: sysmalloc: Assertion `(old_top == (((mbinptr) (((char *) &((av)->bins[((1) - 1) * 2])) - __builtin_offsetof (struct malloc_chunk, fd)))) && old_size == 0) || ((unsigned long) (old_size) >= (unsigned long)((((__builtin_offsetof (struct malloc_chunk, fd_nextsize))+((2 *(sizeof(size_t))) - 1)) & ~((2 *(sizeof(size_t))) - 1))) && ((old_top)->size & 0x1) && ((unsigned long) old_end & pagemask) == 0)' failed.
  
It seems that this error occurred only when I add a new tree, So I wonder how to creat a new tree successfully in RRT.

If there is someone out there who will help me, i thank him alot in advance :)

--
You received this message because you are subscribed to the Google Groups "PR2_Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pr2_users+...@googlegroups.com.

jumping

unread,
Aug 4, 2017, 3:30:59 AM8/4/17
to PR2_Users, 2435...@qq.com
/*This is RRT.cpp */

#include "ompl/geometric/planners/rrt/RRT.h"
#include "ompl/base/goals/GoalSampleableRegion.h"
#include "ompl/tools/config/SelfConfig.h"
#include <limits>

ompl::geometric::RRT::RRT(const base::SpaceInformationPtr &si) : base::Planner(si, "RRT")
{
    specs_.approximateSolutions = true;
    specs_.directed = true;

    goalBias_ = 0.05;
    maxDistance_ = 0.0;
    lastGoalMotion_ = NULL;

    Planner::declareParam<double>("range", this, &RRT::setRange, &RRT::getRange, "0.:1.:10000.");
    Planner::declareParam<double>("goal_bias", this, &RRT::setGoalBias, &RRT::getGoalBias, "0.:.05:1.");
}

ompl::geometric::RRT::~RRT()
{
    freeMemory();
}

void ompl::geometric::RRT::clear()
{
    Planner::clear();
    sampler_.reset();
    freeMemory();
    if (nn_)
        nn_->clear();
    if (tnn_)
        tnn_->clear();
    lastGoalMotion_ = NULL;
}

void ompl::geometric::RRT::setup()
{
    Planner::setup();
    tools::SelfConfig sc(si_, getName());
    sc.configurePlannerRange(maxDistance_);

    if (!nn_)
        nn_.reset(tools::SelfConfig::getDefaultNearestNeighbors<Motion*>(si_->getStateSpace()));
    nn_->setDistanceFunction(boost::bind(&RRT::distanceFunction, this, _1, _2));
  if (!tnn_)
        tnn_.reset(tools::SelfConfig::getDefaultNearestNeighbors<Motion*>(si_->getStateSpace()));
    tnn_->setDistanceFunction(boost::bind(&RRT::distanceFunction, this, _1, _2));
}

void ompl::geometric::RRT::freeMemory()
{
   std::vector<Motion*> motions;
    if (nn_)
    {
        nn_->list(motions);
        for (unsigned int i = 0 ; i < motions.size() ; ++i)
        {
            if (motions[i]->state)
                si_->freeState(motions[i]->state);
            delete motions[i];
        }
    }

 if (tnn_)
    {
        tnn_->list(motions);
        for (unsigned int i = 0 ; i < motions.size() ; ++i)
        {
            if (motions[i]->state)
                si_->freeState(motions[i]->state);
            delete motions[i];
        }
    }
}

ompl::base::PlannerStatus ompl::geometric::RRT::solve(const base::PlannerTerminationCondition &ptc)
{
    checkValidity();
    base::Goal                 *goal   = pdef_->getGoal().get();
    base::GoalSampleableRegion *goal_s = dynamic_cast<base::GoalSampleableRegion*>(goal);

    while (const base::State *st = pis_.nextStart())
    {
        Motion *motion = new Motion(si_);
        si_->copyState(motion->state, st);
        nn_->add(motion);
    }

    if (nn_->size() == 0)
    {
        OMPL_ERROR("%s: There are no valid initial states!", getName().c_str());
        return base::PlannerStatus::INVALID_START;
    }

    if (!sampler_)
        sampler_ = si_->allocStateSampler();

    OMPL_INFORM("%s: Starting planning with %u states already in datastructure ", getName().c_str(), nn_->size());

    Motion *solution  = NULL;
    Motion *approxsol = NULL;
    double  approxdif = std::numeric_limits<double>::infinity();
    Motion *rmotion   = new Motion(si_);
    base::State *rstate = rmotion->state;
    base::State *xstate = si_->allocState();
 

    while (ptc == false)
    {

        /* sample random state (with goal biasing) */
     if (goal_s && rng_.uniform01() < goalBias_ && goal_s->canSample())
           goal_s->sampleGoal(rstate);
     else
          sampler_->sampleUniform(rstate);

        /* find closest state in the tree */
        Motion *nmotion = nn_->nearest(rmotion);
        base::State *dstate = rstate;

        /* find state to add */
        double d = si_->distance(nmotion->state, rstate);
        if (d > maxDistance_)
        {
            si_->getStateSpace()->interpolate(nmotion->state, rstate, maxDistance_ / d, xstate);
            dstate = xstate;
        }

        if (si_->checkMotion(nmotion->state, dstate))
        {
            /* create a motion */
            Motion *motion = new Motion(si_);
            si_->copyState(motion->state, dstate);
            motion->parent = nmotion;

            nn_->add(motion);
            double dist = 0.0;
            bool sat = goal->isSatisfied(motion->state, &dist);
            if (sat)
            {
                approxdif = dist;
                solution = motion;
                break;
            }
            if (dist < approxdif)
            {
                approxdif = dist;
                approxsol = motion;
            }
        }
    }

    bool solved = false;
    bool approximate = false;
    if (solution == NULL)
    {
        solution = approxsol;
        approximate = true;
    }

    if (solution != NULL)
    {
        lastGoalMotion_ = solution;

        /* construct the solution path */
        std::vector<Motion*> mpath;
        while (solution != NULL)
        {
            mpath.push_back(solution);
            solution = solution->parent;
        }

        /* set the solution path */
        PathGeometric *path = new PathGeometric(si_);
        for (int i = mpath.size() - 1 ; i >= 0 ; --i)
            path->append(mpath[i]->state);
        pdef_->addSolutionPath(base::PathPtr(path), approximate, approxdif, getName());
        solved = true;
    }

    si_->freeState(xstate);
    if (rmotion->state)
        si_->freeState(rmotion->state);
    delete rmotion;

    OMPL_INFORM("%s: Created %u states", getName().c_str(), nn_->size());

    return base::PlannerStatus(solved, approximate);
}

void ompl::geometric::RRT::getPlannerData(base::PlannerData &data) const
{
    Planner::getPlannerData(data);

    std::vector<Motion*> motions;
    if (nn_)
        nn_->list(motions);

    if (lastGoalMotion_)
        data.addGoalVertex(base::PlannerDataVertex(lastGoalMotion_->state));

    for (unsigned int i = 0 ; i < motions.size() ; ++i)
    {
        if (motions[i]->parent == NULL)
            data.addStartVertex(base::PlannerDataVertex(motions[i]->state));
        else
            data.addEdge(base::PlannerDataVertex(motions[i]->parent->state),
                         base::PlannerDataVertex(motions[i]->state));
    }
}



/* This is RRT.h */

#ifndef OMPL_GEOMETRIC_PLANNERS_RRT_RRT_
#define OMPL_GEOMETRIC_PLANNERS_RRT_RRT_

#include "ompl/geometric/planners/PlannerIncludes.h"
#include "ompl/datastructures/NearestNeighbors.h"

namespace ompl
{

    namespace geometric
    {

        /**
           @anchor gRRT
           @par Short description
           RRT is a tree-based motion planner that uses the following
           idea: RRT samples a random state @b qr in the state space,
           then finds the state @b qc among the previously seen states
           that is closest to @b qr and expands from @b qc towards @b
           qr, until a state @b qm is reached. @b qm is then added to
           the exploration tree.
           @par External documentation
           J. Kuffner and S.M. LaValle, RRT-connect: An efficient approach to single-query path planning, in <em>Proc. 2000 IEEE Intl. Conf. on Robotics and Automation</em>, pp. 995–1001, Apr. 2000. DOI: [10.1109/ROBOT.2000.844730](http://dx.doi.org/10.1109/ROBOT.2000.844730)<br>
        */

        /** \brief Rapidly-exploring Random Trees */
        class RRT : public base::Planner
        {
        public:

            /** \brief Constructor */
            RRT(const base::SpaceInformationPtr &si);

            virtual ~RRT();

            virtual void getPlannerData(base::PlannerData &data) const;

            virtual base::PlannerStatus solve(const base::PlannerTerminationCondition &ptc);

            virtual void clear();

            /** \brief Set the goal bias

                In the process of randomly selecting states in
                the state space to attempt to go towards, the
                algorithm may in fact choose the actual goal state, if
                it knows it, with some probability. This probability
                is a real number between 0.0 and 1.0; its value should
                usually be around 0.05 and should not be too large. It
                is probably a good idea to use the default value. */
            void setGoalBias(double goalBias)
            {
                goalBias_ = goalBias;
            }

            /** \brief Get the goal bias the planner is using */
            double getGoalBias() const
            {
                return goalBias_;
            }

            /** \brief Set the range the planner is supposed to use.

                This parameter greatly influences the runtime of the
                algorithm. It represents the maximum length of a
                motion to be added in the tree of motions. */
            void setRange(double distance)
            {
                maxDistance_ = distance;
            }

            /** \brief Get the range the planner is using */
            double getRange() const
            {
                return maxDistance_;
            }

            /** \brief Set a different nearest neighbors datastructure */
            template<template<typename T> class NN>
            void setNearestNeighbors()
            {
                nn_.reset(new NN<Motion*>());
                tnn_.reset(new NN<Motion*>());
            }

            virtual void setup();

        protected:


            /** \brief Representation of a motion

                This only contains pointers to parent motions as we
                only need to go backwards in the tree. */
            class Motion
            {
            public:

                Motion() : state(NULL), parent(NULL)
                {
                }

                /** \brief Constructor that allocates memory for the state */
                Motion(const base::SpaceInformationPtr &si) : state(si->allocState()), parent(NULL)
                {
                }

                ~Motion()
                {
                }

                /** \brief The state contained by the motion */
                base::State       *state;

                /** \brief The parent motion in the exploration tree */
                Motion            *parent;

            };

            /** \brief Free the memory allocated by this planner */
            void freeMemory();

            /** \brief Compute distance between motions (actually distance between contained states) */
            double distanceFunction(const Motion *a, const Motion *b) const
            {
                return si_->distance(a->state, b->state);
            }

            /** \brief State sampler */
            base::StateSamplerPtr                          sampler_;

            /** \brief A nearest-neighbors datastructure containing the tree of motions */
            typedef boost::shared_ptr< NearestNeighbors<Motion*> > TreeData;
            TreeData                      nn_;
            TreeData                      tnn_;

            /** \brief The fraction of time the goal is picked as the state to expand towards (if such a state is available) */
            double                                         goalBias_;

            /** \brief The maximum length of a motion to be added to a tree */
            double                                         maxDistance_;

            /** \brief The random number generator */
            RNG                                            rng_;

            /** \brief The most recent goal motion.  Used for PlannerData computation */
            Motion                                         *lastGoalMotion_;
        };

    }
}

#endif
Reply all
Reply to author
Forward
0 new messages