• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 99 subscribers
  • Views 258 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA4VM: Visual_Localization execution along with Realsense running

Aditya Varadaraj
Prodigy 100 points
Part Number: TDA4VM

Hi

I am trying to execute roslaunch ti_vl visloc.launch while  roslaunch realsense2_camera rs_camera.launch is running and putting the input image topic in visloc.launch as /camera/color/image_raw which is the output from rs_camera.launch. I am recieving the following error:

[ INFO] [1700009089.817898998]: Created Publisher for topic: vis_localize/out_image
[ INFO] [1700009089.824087875]: Created Publisher for topic: vis_localize/pose
[ INFO] [1700009089.827646446]: Created Publisher for topic: vis_localize/map
[ INFO] [1700009089.838358822]: Subscribed to the topic: /camera/color/image_raw
[ INFO] [1700009089.839243049]: processCompleteEvtHdlr Launched.
[00:45:58.68968.968219]:ERROR:[VISLOC_run:0283] VISLOC_popFreeInputDesc() failed. Incoming Frame is dropped!
[00:45:58.69042.042946]:ERROR:[VISLOC_run:0283] VISLOC_popFreeInputDesc() failed. Incoming Frame is dropped!

Please suggest how to proceed further to resolve this issue.

  • 0
    TI__Guru 50403 points

    Hi Aditya,

    Looking through the README for ti_vl, it looks like there is a limitation for using this ROS node as-is with a live input:

    ## Known Issues
1. The provided sparse 3D map is for the input bag file we provide. Both the input bag file and the sparse 3D map are created using CARLA simulator. Since the sparse 3D map should cover the area where the input image is being captured for localization. This demo does not work for live camera input with the provided sparse 3D map.

    For reference, linking the README: https://git.ti.com/cgit/processor-sdk-vision/jacinto_ros_perception/tree/ros1/nodes/ti_vl/README.md. So to make this demo work with live camera, then sparse 3D map should be created again.

    Regards,

    Takuma

  • 0
    TI__Expert 3040 points

    Hello Aditya,

    As Takuma said, this demo doesn't work with live camera. For this localization demo to work, the 3D map should be built first with camera and lidar. For this demo, the map has been built using simulator. Currently we do not provide a tool to build the 3D map for this localization demo. AprilTag seems to be an easier way to implement visual localization and SLAM.