Skip to content

Incorrect results for simulated Go1 #13

@Abban-Fahim

Description

@Abban-Fahim

Hi there, thank you for this amazing work!

I was trying out this repository's code for estimating base linear velocity (which I need later as input to an RL policy) and I was pleased to find out that it already had the urdf and feet names to work with the Go1. Apart from changing the one line in leg_odometry_plugin.cpp, I did not modify any of the other code, and used it with the default simulation environment available from the unitree_ros repository (ROS1, I'm using Melodic on Docker). However, I found that the state estimator was quite unreliable. I plotted it against the base velocity against the (what I hope is) ground truth velocity provided by Gazebo and found that the estimated velocity would tend towards the opposite magnitude (-1 for a forward movement) and fluctuate a lot. You can see it in the screenshot below.

Image

Similar behavior is observed for a constant velocity in any direction (-ve x, +ve and -ve y; I have not tested angular velocity or attitude yet). Could there be some mis-configuration on my part, or could the simulation be the reason for these results? Thank you for your help!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions