How feasible is it to perform multi-topology rebar joint labeling and tying with minimal data?

We show that diverse rebar joints can be reliably annotated from minimal data using a geometry-driven, two-stage pipeline, enabling robust sim-to-real generalization for robotic rebar tying.

less than 1 minute read

Code Dataset

Geometry-Driven Perception and Action for Topology-Agnostic Robotic Rebar Tying

Huiguang Wang^a, Zekai Jin^a, Yi Shao^a,*
^a McGill University, Montreal, Canada
* Corresponding author: yi.shao2@mcgill.ca

Background & Methodology

Background

Two-stage Pseudo-labeling

Topology-agnostic Tying Policy

The code and dataset will be made publicly available after the publication of this paper.