Fusing odometry from wheel encoder and VIO

Hi, so I have a simple robot simulation. I use the gazebo diffdrive plugin so it will generate odometry based on the wheel. I also put the camera and imu plugin, and use the VINS-FUSION (https://github.com/HKUST-Aerial-Robotics/VINS-Fusion), it will also generate an odometry message. Now my tf tree looks like this: :https://drive.google.com/file/d/14Vrukgt4M1uKCGpVhargOPqjwwvgbZE/view?usp=sharing

I want to fuse these two odometry messages using the robot_localization package (the ekf node). But since my wheel encoder odometry frame does not connect to the vins-fusion odometry frames, I got the following errors: image description https://drive.google.com/file/d/1C2aDTdXwFt6Yt4YiWO7qN8JVk77hQQS6/view?usp=sharing

Any suggestion about how to use robot-localization, or any other pkgs, to fuse these two odometry messages together? Thanks!!

Asked by Dadaism6 on 2021-04-29 21:29:08 UTC

Comments

Answers

Hi have you found a solution to this? I am currently facing the same issue as you

Asked by ROSlearner on 2021-05-19 12:50:52 UTC

Comments

Hello there, I never been in this case, but If they are completely disconnected the tf transform can't link them. Since you already have an odometry frame, you should probably create a link between your drone odom and your world as well.

Both branches should be connected in some way, and being connected to a common 'world' seems logical (note; for your robot it could be odom--> mapx --> world )

hope this help (else, more details)

Asked by my-name-could-be-D on 2021-05-21 14:05:44 UTC