Case Study: Istanbul Technical University turns whiteboard control sketches into MATLAB code with MathWorks

Deep Learning Turns Professor’s Whiteboard Sketches into Working Code

Istanbul Technical University faced a challenge in teaching control theory, where professors' hand-drawn whiteboard diagrams were difficult for students to interpret. The process of converting these feedback control architecture (FCA) sketches into digital models for analysis in MATLAB and Simulink was too time-consuming to perform live during a lecture, hindering student visualization and comprehension.

To solve this, a team at the university, using MathWorks tools including MATLAB, Simulink, and Deep Learning Toolbox, developed an AI-powered pipeline. This solution uses a deep neural network to instantly convert a photograph of a whiteboard sketch into a working Simulink model and code. The MathWorks environment was crucial for integrating the computer vision and deep learning components. The resulting system successfully translates sketches into dynamic models in real-time, with character recognition achieving 96% accuracy, significantly accelerating the teaching process and providing a powerful new tool for visualization.

Istanbul Technical University

Tufan Kumbasar

Director