ModalAI Forum
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Groups
    • Register
    • Login

    Attempt to improve odometry using Opticalflow(&rangefinder)

    VOXL 2
    1
    2
    189
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • Jetson NanoJ
      Jetson Nano
      last edited by

      Hey @Eric-Katzfey, @James-Strawson @Tom-Kaufmann,

      I've been using Starling for quite a while. Congratulations on a great product.
      One of the things I noticed was that when using voxl-mapper for a significant time (like 15mins), the accumulated drift becomes a serious issue. I want to try and reduce the accumulated drift.

      An option to solve this issue is to use an OpticalFlow/Rangefinder with an external FC. I have completed this setup successfully. I can use my MAV in position mode with the fused odometry without issues. VOA works as well.

      I know from the docs and GitLab that voxl-mapper takes 2 inputs whub_body_wrt_fixed and tof. Is there a way to pass the fused odometry instead of whub_body_wrt_fixed? [ I want the mapper to use this data instead of whub_body_wrt_fixed]. The message type of the fused odometry is nav_msgs/Odometry.

      1. Will I need to modify the voxl-mavlink-server to achieve this?
      2. How do I get the mapper to use the fused odometry data?
      3. How do I synchronize these new position estimates and the pointcloud?

      I'm currently working on this project. So any suggestions or ideas will be instantly implemented, tested and reported back. i hope to hear from you.

      Thank you

      Jetson NanoJ 1 Reply Last reply Reply Quote 0
      • Jetson NanoJ
        Jetson Nano @Jetson Nano
        last edited by

        @Chad-Sweet, @James-Strawson, @tom any updates?

        1 Reply Last reply Reply Quote 0
        • First post
          Last post
        Powered by NodeBB | Contributors