Skip to content

How to adapt the hand-eye calibration matrix from the real world to the Gazebo simulation environment? #12

Description

@a-small-white-guy

After performing eye-in-hand calibration between the D435 camera and the RML robotic arm in the real world, I converted the resulting matrix into the relative pose between the camera and link6 (the end-effector of the robotic arm).However, I am not sure whether this matrix represents the transformation from the optical frame of the camera to the end-effector frame, or some other frame.
In addition, after transforming according to the optical frame convention, I found that in Gazebo the optical frame of the D435 and the robot end-effector only have a translation along one axis, whereas in the real world there is an offset along one axis between the physical center of the camera and the robot end-effector.
The optical frame of the D435 only has a spatial rotation relative to its camera_link, with no translation vector.My goal is to adapt the real-world calibration matrix for simulation so that after proper alignment, I can use the real-world matrix for related training.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions