← Industry Showcase

Construction

TACKLE Lab

AI-Driven Pre-Planning for Construction: From 2D Drawings to Optimized Panel Manufacturing

https://tacklelab.ca/ ↗

Showcase video

Showcase presentation — 6 min 49 s

Infographic

Download
TACKLE Lab infographic

About this showcase

This work introduces an integrated AI framework that automates and optimizes key pre-planning phases in construction management. It begins with BIRD, an end-to-end framework that converts 2D perspective drawings into scaled 3D building envelope models for accurate design visualization. The workflow then transitions to JACK, which employs joint training and cascaded neural networks to predict material requirement quantities across interdependent building components. Finally, PAAD applies a genetic evolutionary algorithm based on 2D bin packing to generate optimal panel manufacturing layouts and minimize material waste.

The research is led by Dr. Andrew Fisher (andrew.fisher@unb.ca) and the TACKLE Lab (https://tacklelab.ca/), dedicated to advancing artificial intelligence, machine learning, and computational optimization in real world settings. The lab uses this project to focus on bridging the gap between digital architectural design, material knowledge automation, and off-site manufacturing optimization to drive real-world operational efficiency and sustainability across the construction industry.

At a glance

  • Company TACKLE Lab
  • SectorConstruction
  • Presented byDr. Andrew Fisher, University of New Brunswick
  • Research projectMitacs IT22468 — Design Automation and Optimization Using Artificial Intelligence
  • SupervisorsPawan Lingras (co-supervisors: Muntasir Billah, Vijay Mago)
  • UniversitiesSaint Mary's University
  • Website https://tacklelab.ca/

Partner organization: Greenstone Brandon Plant Ltd.

Contact

Dr. Andrew Fisher

andrew.fisher@unb.ca