Skip to content

Clarification on motion_anchor_pos_b semantics between training and real-robot deployment #3

Description

@guanzimeng4-ops

Hi, thank you for releasing the training and deployment code.

I am trying to understand how motion_anchor_pos_b is reproduced on the real robot without motion capture or global torso-position estimation. After comparing the paper, the training repository, and RoboNaldo_Deploy, I found what appears to be a significant difference in observation semantics.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions