Barbosa's profile - activity

This worked fine. thank you!

Hello guys,

I am subscribing to a topic with a normal subscriber (used only for testing) and also subsribing with another approximateTime synchronizer subscriber (from message_filters) and the thing is, although the topics are being received, one of them has corrupted data.... it is a geometry_msgs::PoseStamped. While rostopic echo (and also the normal subscriber) print the normal values such as position.x= 0.5, the synchronized subscriber receives (or prints, i'm using %f) 56243784258694623847252734942351.000.

Has anyone had the same problem or has any ideas of why is this happening?

I will try this simpler way of doing it! Thanks!

I think this will help, thanks a lot. I did try this approach too but I was having some trouble with errors, maybe now this will help clear them out.

Hi,

So I was trying to have a transform published over tf as a result of a callback call, which made me think I had to have a tf::TransformBroadcaster as a global variable. However these can't be declared before the ros::NodeHandle is created.

I have tried using a boost::shared_ptr and almost got the hang of it but I still didn't quite understand how to use these boost pointers. Here's what I tried to do:

boost::shared_ptr<tf::TransformBroadcaster> pose_broadcaster;
void poseCallback(const nav_msgs::Odometry& pose){

    geometry_msgs::TransformStamped pose_trans;
    pose_trans.header.stamp = pose.header.stamp;
    pose_trans.header.frame_id = "odometry_frame"; 
    pose_trans.child_frame_id = "base_link";

    pose_trans.transform.translation.x = pose.pose.pose.position.x;
    pose_trans.transform.translation.y = pose.pose.pose.position.y;
    pose_trans.transform.translation.z = pose.pose.pose.position.z;
    pose_trans.transform.rotation = pose.pose.pose.orientation;
    ROS_INFO("Sending TF broadcast");
    pose_broadcaster.sendTransform(pose_trans);


}
int main(){
    ros::init(argc, argv, "pioneer_tf");
    ros::NodeHandle n;
    ros::Subscriber pose_sub=n.subscribe("pose", 1000, poseCallback);
    boost::shared_ptr<tf::TransformBroadcaster> ptr;
    pose_broadcaster= boost::shared_ptr<tf::TransformBroadcaster>(ptr);//problem here
   ros::spin();
}

I obviously got it wrong in the initialization of the pointer, what's the right way of doing it in this case?

--- EDIT ---

This problem is solved now, but I have a similar issue at the moment with a transformListener

It is basically the same set up, with a global variable (pointer declared as global and initialized in main) and the listener being called in a callback. The thing is, even though it compiles, the callback is never called, but if I comment the line where it is called inside the callback, it works and does everything normally... Any ideas of why the callback might not be called?

--- EDIT2 ---

This problem is totally solved. It was just a matter of making the TF buffer bigger and using a try/catch.

Hi, I am currently trying to capture images and localizing a colored landmark in the kinect frame. I first tried detecting the color in a downsampled pointcloud but that was just too slow, and so now I am doing color segmentation with OpenCV hoping that i can also mask a depth image with which I can build a smaller point cloud.

Here is my current problem.

    void depthCb(const sensor_msgs::ImageConstPtr& msg)
  {
    cv_bridge::CvImageConstPtr cv_ptr;
    try
    {
      cv_ptr = cv_bridge::toCvShare(msg, enc::TYPE_16UC1);
    }
    catch (cv_bridge::Exception& e)
    {
      ROS_ERROR("cv_bridge exception: %s", e.what());
      return;
    }

     //dst is the mask resultant from color segmentation, class member
   Mat result= dst & src;
    cv::imshow("depth_filtered", result);

    cv::imshow("depth_image", cv_ptr->image);
    cv::waitKey(3);

    image_pub_.publish(cv_ptr->toImageMsg());
  }

This is a callback for the depth raw data. When I run this I get an exception

[ERROR] [1349283297.431785349]: cv_bridge exception: Unsupported conversion from [16UC1] to [mono16]

I chose to convert to mono16 because I tought it was basically the same thing since the kinect image is 16UC1. I was trying to avoid writting the code to convert this, so it would be nice if there was any existing solution.

I am sort of new to openCV as well as ROS itself so if you have any ideas of how to overcome this particular obstacle or if you know of a better method to achieve my goal, please share. Thanks in advance.

EDIT

This worked, but it kind of slowed down the code

void depthCb(const sensor_msgs::ImageConstPtr& msg)
  {
    cv_bridge::CvImageConstPtr cv_ptr;
    try
    {
      cv_ptr = cv_bridge::toCvShare(msg, enc::TYPE_16UC1);
    }
    catch (cv_bridge::Exception& e)
    {
      ROS_ERROR("cv_bridge exception: %s", e.what());
      return;
    }

     ROS_INFO("Applying mask to depth image");
   Mat m=Mat(cv_ptr->image.size(), cv_ptr->image.type());

   cvtColor(dst,m,CV_RGB2GRAY); 
   Mat result=Mat(cv_ptr->image.size(), cv_ptr->image.type());
   cv_ptr->image.copyTo(result, m); 

    cv::imshow("depth_filtered", result);
    cv::imshow("depth_image", cv_ptr->image);
    cv::waitKey(1);

  }

Hello,

I built a node which simply reads a file and builds a vector of data from it. However, when i tried running the node from roslaunch instead of rosrun, according to the log file, it ca n longer open the said data.txt file, causing my node to shutdown. I tried moving the file around but the problem persists. Here is the code

ifstream vel("ros_workspace/odometry/traj.txt");
  if(!vel.is_open()
{
    ROS_INFO("file not found!");
    ros::shutdown();
  }

I am new to ROS and not very familiar with the PATHS, don't even know if the problem comes from there, so this might be a very simple question.

Thanks! Barbosa