Projects

Exploring intelligence.
Expanding possibility.

Exploring how intelligent systems can help us understand information and approach complex problems.

View research publications

Selected experience

Machine vision2025–2027

Enhanced light-field machine vision camera with AI image analysis/defect identification for high-tech glass products

Light-field imaging and AI analysis to detect, locate, and classify defects in high-tech glass during production. The project aims to support earlier quality decisions and reduce material waste.

Wearable robotics2023–2027

Soft wearable assistive garments for human empowerment

Soft, inflatable assistive garments combining sensing, actuation, and data-driven movement prediction. The research explores adaptive support for the lower body and core.

Circular construction2023–2027

A territorial construction system for a circular low-carbon built environment

Lower-carbon materials and modular building components designed for disassembly and reuse. Digital tools support a circular construction lifecycle, with demonstrators in Barcelona and Brussels.

Wearable health technologyMay 2022 – Jan 2023

Wearable ECG-based Real-time Arrhythmia Monitoring Using Deep Neural Networks

A wearable, ECG-based system has been developed for real-time arrhythmia monitoring using deep neural networks. The system, embedded in a garment and powered by a microcontroller, showcases exceptional performance with an accuracy of 99.6% and an F1 score of 98.8%. Notably, it maintains a low energy consumption, requiring only 4.24 mJ per inference. Furthermore, the system features active artefact noise detection utilizing acceleration data, bolstering its efficacy and reliability in real-world applications.

Edge AISep 2020 – Jan 2022

ECG-based real-time arrhythmia monitoring using quantized deep neural networks

This project introduces an efficient, lightweight, and automated ECG classification method using quantized one-dimensional deep convolutional neural networks for real-time arrhythmia monitoring. Designed for high accuracy and privacy, the solution processes raw ECG data directly, eliminating the need for complex pre-processing. The system's efficiency allows for local data analysis on edge devices such as smartphones, making it highly portable and privacy friendly. A study was conducted, validating the solution's effectiveness, boasting a sensitivity of 98.5%, specificity of 99.8%, and requiring only 5.85 mJ energy and 7.65 ms per prediction.

Biomimetic roboticsMar 2018 – Jun 2018

Artificial Intelligence and Biomimetic Robotics

This project features the fusion of biomimetic principles with robotics and artificial intelligence, culminating in a hexapod robot equipped with a simple reflex agent AI. The robot, featuring a 3D-printed exoskeleton and powered by a Raspberry Pi, uses an I2C board and soldered components to facilitate efficient operation. An ultrasound proximity sensor integrated into the system allows for effective interaction with the robot's environment. The entire system was programmed using Python, highlighting its application in practical, interactive, and autonomous robotics.

Bring a different perspective.

Discuss a collaboration