ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
2017-06-07 05:02:36 -0500 | received badge | ● Famous Question (source) |
2016-02-15 07:44:51 -0500 | received badge | ● Taxonomist |
2015-06-10 19:23:05 -0500 | received badge | ● Famous Question (source) |
2015-06-10 19:23:05 -0500 | received badge | ● Notable Question (source) |
2015-03-23 01:11:26 -0500 | received badge | ● Good Answer (source) |
2014-11-23 02:00:27 -0500 | received badge | ● Nice Answer (source) |
2014-11-10 11:12:30 -0500 | received badge | ● Famous Question (source) |
2014-10-20 03:28:26 -0500 | received badge | ● Necromancer (source) |
2014-10-20 03:28:26 -0500 | received badge | ● Teacher (source) |
2014-10-16 07:54:47 -0500 | received badge | ● Notable Question (source) |
2014-10-15 11:32:30 -0500 | received badge | ● Famous Question (source) |
2014-10-14 04:14:30 -0500 | answered a question | Transform Quaternion It seems that creating a Matrix3X3 from the quaternion is not necessary since a transform already contains the rotation matrix. If you are working with TF, inside every Quaternion there is a Matrix3x3 called m_basis. When you call Transform::getRotation(), what you get is From http://docs.ros.org/indigo/api/tf/htm... : So it could be that you may get away with: |
2014-08-17 12:47:35 -0500 | received badge | ● Popular Question (source) |
2014-08-03 11:02:42 -0500 | received badge | ● Notable Question (source) |
2014-07-31 18:00:37 -0500 | received badge | ● Popular Question (source) |
2014-07-31 11:05:29 -0500 | asked a question | How to calculate covariance matrix for monocular SLAM? Hi there, I am using a single camera for visual odometry. The camera is fixed to the robot and I use the common approach of matching descriptors between consecutive frames and then estimating the rotation and translation using SVD. The camera is already calibrated and I already have the points in 3D position. From the rotation matrix and the translation matrix I can obtain the the rotation angles and the new position x,y,z of the robot. My question then is: How can I estimate the covariance matrix for theses values in order to publish an nav_msgs::Odometry matrix? I suspect that I have to backpropagate the points from one frame to the other, but I am not sure. Any references to papers/books are also welcome =) Thanks in advance, Pedro |
2014-07-18 02:45:14 -0500 | received badge | ● Enthusiast |
2014-07-14 10:19:59 -0500 | answered a question | Why is image_proc only show the upper left half of my image? Hi, I think I have found the answer. The problem is how to treat the "pixel binning" fields of the Camera Info Messages. The driver I am using actually publishes that the camera has already processed the original image and has used a 2x2 pixel binning to reduce the size of the image alongo with the new (reduced) dimension of the image. The package image_proc however, uses that information once again to generate an empty matrix for the rectified image. i.e.: My camera driver uses Format7 Video Mode 0 to produce a 1024x1024 image (after binning its original 2048x2048 using 2x2 bins). The problem is that it also publishes the binning information, so binning_x: 2 Which causes some problems for image_proc for it tries to generate a (1024/2)x(1024/2) empty matrix which will be partially filled up with the rectified frame... See ya |
2014-07-14 06:49:02 -0500 | asked a question | Why is image_proc only show the upper left half of my image? Hello, header: While the header of the output of /camera/image_raw is: So the problem is that, when I run $ROS_NAMESPACE=camera image_proc image_proc I get images to be published at /camera/image_rect that have the following header: I can see with image_view that the image in the rectified version is only the 512x512 upper left of the original 1024x1024 image_raw Does anyone know why image_proc is only grabbing only the first upper 512x512 pixels of my image? It seems to me to be some kind of ROI problem, but I cannot see where... Thanks in advance, Pedro |
2014-07-14 02:40:10 -0500 | received badge | ● Popular Question (source) |
2014-07-11 02:03:11 -0500 | asked a question | Is there a node/package for periodic camera calibration? Hi there, I have a single rgb camera with known intrinsic parameters mounted to a robot together with a chessboard pattern fixed the robot's base. The chess board is fixed and I already publish its transform relative to the base, but the camera position may change (tilt) overtime which will certainly influence on the visual measurements I am making. So my question is: Is there a node/package that I can use that automatically publishes the TF for a camera given a fixed pattern? I remind you that I already have the intrinsics parameters and that theses are published using CameraInfo. Thanks, Pedro |
2014-07-11 01:55:56 -0500 | received badge | ● Supporter (source) |
2014-07-09 12:13:39 -0500 | received badge | ● Notable Question (source) |
2014-07-08 06:22:29 -0500 | received badge | ● Popular Question (source) |
2014-07-07 16:13:36 -0500 | asked a question | Why use multiple node handles? Hi, I am currently developing a ROS packages to to PointGrey Grasshopper3 and I've seen that some packages like the camera1394 use more than one node handler. Does anyone know why? Thanks, Pedro |
2014-07-07 16:08:17 -0500 | commented question | Failed to process package 'qt_gui_cpp' on Debian sid Nop... Moved to Ubuntu 12.04. Although I like Debian better, it is too much hassle |
2014-04-09 09:33:14 -0500 | answered a question | Can we install ros Hydro on Ubuntu 13.10 Worry no more! Where there is a hack... there is a way! Now, I have just installed Ubuntu 13.10 to later find out that it is not supported by Hydro... Sad but true... Anyway, the thing is that the meta-package ros-hydro-desktop-full installs also gazebo, which is a simulator that requires libogre-dev to be installed. libogre-dev on the other hand needs libboost1.49 but everyone else has already already moved on to libboost1.53... So what do we do? Simple... We compile libogre-dev and make it look like it is native package installed! This means that you have to: 1) Install libboost1.53-all-dev ( I install it all, so dependency won't be a problem) 2) Download libogre-dev source code from http://bitbucket.org/cabalistic/ogredeps 3) Cmake the project and generate the Makefile 4) Use 'sudo checkinstall' ( http://www.linuxjournal.com/content/u... ) to generate and install your libogre-dev package. BUT REMEMBER TO CHANGE THE FOLLOWING: Name of the Package: libogre-dev Version: 1.7.4+dfsg1-7 This way you are tricking ubuntu's package management system into thinking you have the original package. OBS: At this point, you have installed a libogre-dev that has been compiled with libboost1.53. I don't guarantee that gazebo will run perfectly or that it will work at all... 5) Proceed with ROS for Raring as described here: http://wiki.ros.org/hydro/Installatio... See ya |
2013-11-29 13:45:27 -0500 | received badge | ● Student (source) |
2013-11-09 22:52:54 -0500 | received badge | ● Popular Question (source) |
2013-11-09 22:52:54 -0500 | received badge | ● Notable Question (source) |
2013-11-09 22:52:54 -0500 | received badge | ● Famous Question (source) |
2013-10-08 05:50:25 -0500 | asked a question | Failed to process package 'qt_gui_cpp' on Debian sid Hi, I am trying to install ROS Hydro in Debian sid. I have already installed PCL, Gazebo, Sdformat and now I was trying to install all packages but somehow I get a problem while trying to process package 'qt_gui_cpp' at aroun 51%. The coomand I used was : I included the --make-args -d just so to get more info, but could not figure out what is actually going on. Does anyone see anything? Thank you =) The logs is as follows: (more) |