A High-fidelity Digital Twin Platform with An Engineering Mechanism Inference through Vision Systems

While mass production systems have successfully implemented automation processes, job-shop manufacturers face a critical challenge: experienced workers are retiring, taking decades of specialized knowledge with them, while finding skilled replacements becomes increasingly difficult. This project addresses this problem by creating a “digital twin” – essentially a highly accurate virtual copy of a real manufacturing workspace that can see and understand what’s happening on the factory floor. Think of it as creating a knowledgeable digital assistant that learns from expert workers and can guide less experienced operators through complex tasks. The system uses artificial intelligence that learns the best ways to set up equipment from thousands of manufacturing scenarios and predicts potential problems before they occur. It can visually recognize tools and fixtures, answer questions about processes, and even coordinate robotic arms to automate setup tasks. For the manufacturing partner, this technology means preserving valuable expertise that would otherwise be lost, training new workers more quickly and effectively, reducing setup errors, and significantly cutting down the time needed to prepare machines for production – all while maintaining the precision and quality that customers expect.

Faculty Supervisor:

Patrick Lee

Student:

Partner:

UnleashifAI Inc.

Discipline:

Computer science

Sector:

Information and cultural industries; Professional, scientific and technical services

University:

University of Toronto

Program: