Research Project

Advancement and Promotion of AI-Automated Welding Technology for Continuity Plates in Steel Built-up Box Columns

Bringing computer vision, robotics, and intelligent quality inspection together to automate one of structural steel fabrication's most safety-critical welds.

Explore the Project Get in Touch

Overview

Why This Research Matters

Continuity plates connect beam flanges through steel built-up box columns in moment-resisting frames, and their weld quality directly governs the seismic and structural performance of the connection. Fabricating these welds today still relies heavily on manual, highly skilled welders working inside constrained, hard-to-access box-column geometries — a process that is slow, inconsistent, and increasingly difficult to staff.

This project develops and validates an AI-automated welding workflow purpose-built for continuity-plate-to-box-column joints: combining machine-vision joint tracking, adaptive welding parameter control, and AI-based defect detection to deliver welds that are faster to produce, more consistent, and verifiable against code requirements — with the long-term goal of promoting adoption of this technology across the structural steel fabrication industry.

  • 01

    Quality & Consistency

    Reduce weld defects and variability introduced by manual welding in confined box-column joints.

  • 02

    Automation & Efficiency

    Shorten fabrication cycle time and ease the shortage of experienced structural welders.

  • 03

    Verifiable Inspection

    Pair automated welding with AI-assisted inspection for traceable, code-aligned quality records.

  • 04

    Industry Adoption

    Package findings into guidelines and demonstrations that support real-world technology transfer.

Technology

An AI-Driven Automated Welding Pipeline

The system integrates perception, robotic motion, and quality assurance into a closed loop around the continuity-plate joint.

1. Vision-Based Joint Recognition

3D scanning and machine vision locate the continuity plate – box column joint and map seam geometry in real time.

2. Adaptive Welding Path Planning

AI models convert seam data into optimized torch trajectories and parameters for the confined box geometry.

3. Robotic Welding Execution

A robotic welding cell executes the planned path with closed-loop monitoring and real-time correction.

4. AI Quality Inspection

Deep-learning defect detection screens welds for porosity, undercut, and profile deviations against code criteria.

Research Team

People Behind the Project

A cross-disciplinary team spanning structural engineering, robotics, and machine learning.

Principal Investigator

Professor, Department of Civil Engineering

[University / Institution Name]

Co-Investigator

Professor, Robotics & Automation

[University / Institution Name]

Research Assistant

Graduate Researcher, AI & Computer Vision

[University / Institution Name]

Research Assistant

Graduate Researcher, Structural Testing

[University / Institution Name]

News

Latest Updates

  • 2026

    Project Website Launched

    This site now tracks project progress, publications, and team information.

  • TBD

    Robotic Welding Cell Setup

    Details on experimental setup and first continuity-plate welding trials will be posted here.

Contact

Get in Touch

Interested in collaborating, learning more about the research, or discussing industry adoption? Reach out to the project team.

Institution
[Department, University / Institution Name]
Address
[Street Address, City, Country]