16/10/2026
An AI assistant that helps visually impaired users understand their surroundings, flag potential hazards and answer questions — all while processing their visual data on-device to preserve their privacy.
That is what Badrinath Sandhya (Electrical Engineering), Chan Zun Mun Terence (Computer Engineering) and Hsu Myat Noe (Computer Science) set out to build with EdgeSight, earning them a place among the Second Prize winners at the Intel Cup 2026 Undergraduate Embedded System Design Competition in Shanghai.
This year’s edition brought together 230 teams from 120 universities worldwide to develop a working system over three months around the theme of AI application and innovation.
With EdgeSight, a small camera observes the user’s surroundings, enabling the device to describe scenes, identify potential hazards, and respond to questions through voice. By processing this information on device rather than relying on the cloud, EdgeSight can respond quickly while keeping images of the user’s surroundings private.
Getting large AI models to work on an embedded platform meant navigating memory and computing constraints while keeping the system responsive. The team had to optimise the pipeline and decide what processing should happen at each stage.
They also considered how much information a user needs. Flagging every change could result in constant and unnecessary warnings, so the team developed reasoning and warning-suppression logic to distinguish ordinary changes from situations requiring attention.
Sitting at the intersection of hardware, embedded systems and AI, EdgeSight drew on the team’s varied backgrounds. Together, they approached EdgeSight as an integrated system, considering how each component affected the others.
Supervisor Dr Rajesh Panicker (Electrical and Computer Engineering) encouraged the team to go beyond demonstrating an AI model and build a functional system. For Sandhya, the project pushed her to move “from models on a laptop to something that works in real time on embedded hardware”. Across the team, the experience also reinforced AI’s potential to create meaningful social impact.
NUS Electrical and Computer Engineering NUS Computing