Young engineers shine with innovative tech
- Wits University
EIE Open Day 2024: Can a robot do medical triage? Can a face detector spot you when you’re wearing a mask? Can geysers be smart?
The fourth-year Electrical and Information Engineering students have been hard at work once again. During the 2024 Open Day hosted by the School, their projects were on display for industry professionals, students, and families to admire.
From a noisy, high-voltage Tesla coil to biomechanical resistance testers and even a robot wine sniffer, the event had it all. The day concluded with a tree-planting ceremony and awards for the top three posters, chosen by peers and a secret judging panel.
Winners
First place went to Sam Cohen and Asher Radowsky, who developed a system using Large Language Models (LLMs) for medical triage. Their project aimed to examine the ethical decisions embedded in LLMs by testing them with hypothetical emergencies, such as a multi-vehicle crash. They challenged, among others, OpenAI's ChatGPT, Google’s Gemini, and the Chinese ChatGLM with various scenarios and meticulously documented how each model responded, comparing these decisions with medical standards to evaluate alignment.
Second prize was awarded to Yishai Basserabie and Jacob Sacks for their real-time face detection system. They replicated and enhanced a recently proposed algorithm, aiming to improve both speed and accuracy. On the day, their system performed live, recognising faces even when partially obscured by masks. Their work highlighted areas where the algorithm could still improve, providing insights into the complexities of real-world applications.
Third place went to Okuhle Luningo and Bhekanani Mkhize for their project focusing on optimising energy use through smart geysers. Using Model Predictive Control and LSTM models, they developed a system to shift water heater loads away from peak hours. With electric water heaters contributing significantly to South Africa’s evening electricity demand—equivalent to Stage 3 loadshedding—their aim was to flatten the usage curve and ease grid strain. Analysing hot water consumption data from Johannesburg, they simulated water usage patterns and tested their models, demonstrating impressive energy savings. A single-occupant household’s energy consumption dropped from 15.33 kWh to 9.00 kWh, while three-occupant households saw reductions from 35.49 kWh to 18.00 kWh. Though the models struggled with unpredictable usage patterns, their work highlighted the potential of such systems for smarter, cost-effective energy management.
The posters reflected a blend of technical ingenuity and real-world relevance, showcasing the students' readiness to tackle pressing challenges. The event ended with a tree-planting and well-deserved celebrations, marking a day of creativity, learning, and engineering excellence.