Lead researchers

Mariadas Roshan

Sheila Sutjipto

Project based at

Lead Partner Organisation

Dynamic Motion and Vision-Based Study of Steel Coil Raking – Pilot Project

Start: July 2024

Expected end date: October 2025 (12-month duration + 2 months of site testing and data collection)

Background

Steel coils are a critical product in the manufacturing sector, but the way they are stacked can create significant operational challenges. At present, misalignments and uneven layering can lead to material waste, production stoppages, and the need for manual intervention. These issues not only reduce efficiency but also pose safety risks for workers. 

Project Aims

The InfraBuild Steel Coil Project aims to develop new technologies that improve the reliability and safety of coil stacking. By combining computer vision, artificial intelligence, and collaborative robotics, the project seeks to: 

• • Reduce material waste and production downtime, 
  • Minimise the need for manual corrections, 
  • Enhance workplace safety, and 
  • Build the foundation for future automation in heavy manufacturing. 

Project Stages

The whole project is designed as a staged approach: 

1. 1. Vision System Development – design and test a camera-based system to detect coil start/end points and track layering patterns. 
   2. Quality Assessment – build models that automatically distinguish between properly and improperly stacked coils. 
   3. Collaborative Robotics – explore and propose robotic solutions to correct stacking issues autonomously. 

Stage 1 progress

The current phase of the project is Stage 1: Vision System Development. The team is working on a proof-of-concept system that uses industrial cameras and computer vision algorithms to monitor coil positioning and layering. This stage will provide the foundation for later phases, where automated quality checks and robotic assistance will be developed. 

Outcomes

Upon completing this research, we expect to develop:

• Proof-of-concept vision system that demonstrates how computer vision can detect coil start/end points and track layering patterns in real time. 
• Foundation for automation by generating the data and insights needed to develop future classification models (Stage 2) and robotic correction systems (Stage 3). 
• Research contributions including new methods in industrial computer vision and opportunities for academic publications and student training. 

Supervisory Team

• Dr Sheila Sutjipto – Postdoctoral Researcher & Project Co-lead, University of Technology Sydney 

• Dr Mariadas Capsran Roshan – Postdoctoral Researcher & Project Co-lead, Swinburne University of Technology 

• Professor Mats Isaksson – Program Co-Lead, Swinburne University of Technology 

• Dr Michelle Dunn – Program Co-Lead, Swinburne University of Technology 

• Professor John McCormick – Chief Investigator, Swinburne University of Technology 

• Professor Chris McCarthy – Chief Investigator, Swinburne University of Technology 

• Professor Teresa Vidal-Calleja – Program Lead, University of Technology Sydney 

• Associate Professor Gavin Paul – Chief Investigator, University of Technology Sydney 

• Dr Victor Hernandez Moreno – Chief Investigator, University of Technology Sydney 

• Professor Jonothan Roberts – Chief Investigator, Queensland University of Technology 

Research Outputs

Publications

We are currently preparing a conference paper that presents the algorithms developed in this project. The paper will also benchmark our approach against existing methods reported in the literature, with the goal of highlighting improvements in coil feature detection and monitoring.