ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
2017-11-07 08:21:29 -0500 | received badge | ● Famous Question (source) |
2015-04-24 04:16:58 -0500 | received badge | ● Famous Question (source) |
2015-04-24 04:16:58 -0500 | received badge | ● Notable Question (source) |
2015-04-24 04:16:58 -0500 | received badge | ● Popular Question (source) |
2015-02-16 02:25:11 -0500 | received badge | ● Famous Question (source) |
2015-02-16 02:25:11 -0500 | received badge | ● Notable Question (source) |
2015-02-02 17:33:37 -0500 | asked a question | PS3 controller on LUbuntu 14.04 without UInput I see on the ps3joy package page ( http://wiki.ros.org/ps3joy ) that the package is only supported for versions up to 12.10. This seems to correspond with the general google results that show these controllers working up to at least 12.04, sometimes up to 12.10. We are currently running LUbuntu 14.04 (kernel 3.10.54+) with ROS Indigo on an ARM platform (ODROID XU3). Does anyone have any experience in using a PS3 controller without the UInput module (that seems to have been changed to a built-in post 12.10, hence sixad can't use it)? The sixpair program appears to work fine (we require wireless control, so a USB tether is simply not acceptable), but sixad and ps3joy fall over looking for UInput. Any suggestions/help is appreciated. Thanks, aSM |
2014-12-18 16:57:47 -0500 | commented answer | Undefined reference to 'setup' and 'loop' using CMake for custom project This corrected the issue. Entry required was: generate_arduino_firmware(program SKETCH file.ino SRCS ${ROS_LIB_DIR}/time.cpp BOARD mega2560 PORT /dev/ttyACM0) Why is there a SRCS and a SKETCH entry if they behave the same, as I have shown by simply renaming one to the other? |
2014-12-18 08:18:47 -0500 | received badge | ● Popular Question (source) |
2014-12-18 02:42:32 -0500 | commented answer | Undefined reference to 'setup' and 'loop' using CMake for custom project You are correct, gvdhoorn; this is a workaround and not a fix. I will submit the bug report to the issue tracker as per your suggestion. |
2014-12-17 20:24:11 -0500 | answered a question | Undefined reference to 'setup' and 'loop' using CMake for custom project So it appears that rosserial_arduino/CMake can't work with .ino files, unless there is something specifically wrong with my .ino file. Copying and renaming the .ino file to a .cpp file fixed all of the issues, and the code uploaded correctly. Bug? EDIT 1: A symbolic link (ln -s) from the ../sketchbook/<project>/<project>.ino file to the ../firmware/<project>.cpp file also fixes the issue, meaning I can still have the students update the code in one location for both the CLI and GUI. EDIT 2: Obviously an update was required to the ./firmware/CMakeLists.txt to target the new .cpp file, after the .cpp symbolic link was created. |
2014-12-17 20:09:19 -0500 | asked a question | Undefined reference to 'setup' and 'loop' using CMake for custom project I am trying to implement a CLI uploader for an Arduino MEGA2560 using rosserial_arduino and CMake, based upon http://wiki.ros.org/rosserial_arduino... under LUbuntu 14.04 with ROS-Indigo on an ODROID (ARM) development board. The tutorial works fine, as can be seen here: However, when I create a new catkin_package for my desired file (which is actually a symbolic link to a ../sketchbook/ Arduino project due to the need for both IDE and non-IDE uploading (some students are competent with CLI, others GUI)), I get the following output (I have removed the Policy CMP0022 warnings from this output): (more) |
2014-07-21 03:38:23 -0500 | received badge | ● Notable Question (source) |
2014-07-21 03:38:23 -0500 | received badge | ● Popular Question (source) |
2014-04-28 23:09:28 -0500 | asked a question | RViz crashes upon first /scan I'm having a serious problem getting RViz to operate on a FitPC2 (dreaded GMA500 embedded system) with a SICK TiM310 scanner. The FitPC2 is currently running Ubuntu 12.04, ROS-Groovy and RViz 1.9.34 as our primary research platforms are designed around a PandaBoard, but this specific research problem requires some of the x86 packages that are not yet ARMHF compiled for Hydro. We will upgrade along that path at some point. Currently I am able to run RViz fine, but as soon as the first /scan data is published by the SICK_TiM3xx package, RViz crashes with the following terminal information: http://pastebin.com/euHWXPBD I have confirmed that the output of the LiDAR is correct by other means, but it would be extremely valuable to us to be able to visualise the scan data in real-time on the platform. Any help that someone could provide would be greatly appreciated. |
2014-01-28 17:28:40 -0500 | marked best answer | octomap_server globally referenced pointcloud and transform Hi, I am trying to pass globally referenced point clouds and trajectories to octomap_server, but it's not producing the desired results? The point cloud and ray tracing are being inserted from the single origin of the first scan. |
2013-12-10 04:56:02 -0500 | marked best answer | octomap_server performance issues? Hi, I'm currently running octomap_server from the latest svn (checked out on 10/01/13) under ROS-Fuerte. EDIT1: I'm working with complex 3d laser scan data which is moving at a reasonable speed (1m/s, perhaps more). The sensor data is of a building and grounds complex covering 300m by 200m by 50m in elevation throughout the loop. EDIT2: More details about the point cloud and ROS .bag file: If I publish the data in real time (playing back a .bag file through rosbag which contains the PointCloud2 scan and relevant transform), octomap_server quickly falls behind and within 60 seconds it hits 100% CPU load and starts to drop scans. This has been observed on both a Core 2 Duo and a Core i7 (1st generation), both with 4gb of RAM under Ubuntu 11.10. Is this an issue with the configuration paramaters in the .launch file? Currently latch is 'false', 'sensor_model/max_range' is '3.0' and resolution is '0.1' (I would prefer 0.05, but it can't even handle 0.1). Would there be a significant performance increase by implementing an octomap octree in code outside of octomap_server; i.e. write my own node? Is octomap_server capable of operating with real-time sensor data? Thanks in advance. |
2013-07-25 00:29:52 -0500 | received badge | ● Famous Question (source) |
2013-03-02 21:53:35 -0500 | received badge | ● Famous Question (source) |
2013-01-29 16:54:36 -0500 | received badge | ● Notable Question (source) |
2013-01-22 20:25:21 -0500 | received badge | ● Popular Question (source) |
2013-01-22 15:13:09 -0500 | commented answer | Overlay octomap with devel branch (source) for ROS Indeed, octomap_server depends on ros-fuerte-octomap-ros, which is linked to ros-fuerte-octomap. If you could take a look at the EDIT1: section in the original posting, that would be appreciated. Still having issues in linking the openmp development branch to octomap_server. |
2013-01-21 18:08:00 -0500 | commented answer | Overlay octomap with devel branch (source) for ROS Ah, yes, I was confused in thinking that it was a ROS-package. Unfortunately, neither of the suggested fixes had any luck for me, with octomap_server still seeming to use the ros-fuerte-octomap libraries, verified by 'ldd octomap_server_node'. |
2013-01-20 11:31:23 -0500 | asked a question | Overlay octomap with devel branch (source) for ROS I have been attempting to test the new OpenMP implementation for OctoMap (provided by Armin in this question: http://answers.ros.org/question/52127/octomap_server-performance-issues/ ) due to the requirements of the point cloud that I'm trying to use. I downloaded the latest branch from https://github.com/OctoMap/octomap/tree/devel-openmp and then: This completes with no problem, as does: My .bashrc file contains: Even still, octomap_server (which I have been editing in a non-standard location, hence the redirect) still finds the includes and libraries from the ros-fuerte-octomap install, which comes with ros-fuerte-octomap-ros which is a dependency of octomap_server. I assume that I'm missing something with installing the octomap source from a development branch, but I can't work out what it is. *EDIT1: Finally got back to trying to get octomap_server to build with the OpenMP development branch. My CMakeLists.txt file is: Doing a (more) |
2013-01-20 11:21:09 -0500 | commented answer | octomap_server performance issues? Performance on the existing source build of Octomap was greatly increased by increasing the chunk size of the point cloud being fed to octomap_server. I've opened a new question with regards to the devel-openmp branch installation problems. |
2013-01-20 10:31:08 -0500 | received badge | ● Famous Question (source) |
2013-01-17 10:43:49 -0500 | received badge | ● Notable Question (source) |
2013-01-14 18:10:59 -0500 | received badge | ● Scholar (source) |