Teenagers Build AI Glasses for Under $100

Three 15-year-old students from Santa Clara created text-to-speech smart glasses powered by artificial intelligence. The wearable technology helps visually impaired students access reading materials from any format.

The prototype required five months of development and cost less than $100 to build. Akhil Nagori explained the motivation behind the project: "Our main goal was to create an easy, cost-efficient way to transcribe text from any format for visually impaired students."

The glasses operate by taking pictures of text, extracting the content, and converting it to audio played through tiny speakers built into the frames. The device uses a Raspberry Pi computer board, camera, battery, and speakers. It also includes a small on-and-off switch.

Performance metrics show the prototype achieves over 90% accuracy when translating text to speech. Lucas Shengwen Yen highlighted the importance of speed: "One of the most important aspects of our project is the load time. And that averaged around 13 seconds."

The inspiration came from a personal experience. Nagori traveled to India to visit family, including his great-uncle who is visually impaired and works as a cashier. He observed: "He has all these boxes filled with these braille receipts. He has to go through them line by line. When I saw that, I said, 'There's got to be an easier way that's not so tedious.'"

The students faced three main challenges during development: hardware design, software programming, and data collection through testing.

Hardware Design

Evann Sun led the hardware design using Fusion 360 CAD software and a 3D printer to create custom frames. The team researched average glasses dimensions for middle to high school students to ensure proper fit. Sun explained: "Since we're trying to have all of the components on the glasses, we had to custom-design the areas for all of them."

Battery life was a critical consideration. Sun stated: "We want students to use this for, give or take, the entire school day. We are really concerned about the battery life, especially when we are using such a small battery."

Software Training

Nagori custom-trained a convolutional recurrent neural network (CRNN) using a dataset of 800 images. The training data came from school textbooks and other educational materials, featuring colorful images and various font styles.

The team collected images themselves under three lighting conditions: classroom lighting, low lighting, and outdoor lighting to ensure the model worked in different environments.

Testing and Refinement

The students conducted extensive testing to improve accuracy. Sun described the process: "After all our hardware and software were done, we tested our software part. We would input images that we saw online or that we took ourselves into our software model. Then, it extracted the text and gave us an MP3 file, which we could use to improve our accuracy."

The Thermo Fisher Scientific Junior Innovators Challenge represents one of the top STEM research competitions for students. The selection process is highly competitive: middle school students must first compete at local science or engineering fairs, where judges nominate the top 10% of projects. Approximately 2,000 winners apply for the national competition, but only 300 are selected. From those, judges choose 30 finalists who present their research in Washington, D.C.

The team faced significant obstacles during their journey. Nagori admitted: "We had a lot of all-nighters." The most dramatic challenge occurred hours before their presentation when the glasses suffered a critical failure.

Yen described the crisis: "On the flight there, some of our soldering came off the Raspberry Pi. Without the soldering, nothing worked, and the glasses wouldn't start up. We were all in panic mode."

The team scrambled to fix the device. Yen continued: "The night before we were presenting, my dad ran to the closest mechanic store and got a soldering iron. The three of us put on masks, hunched over, and fixed it."

Despite not receiving an initial nomination, judges from the state level attended their competition and recognized the project's value. Sun reflected on the experience: "I think it really taught us that even if we don't get what we want the first time, as long as we work hard and stay committed, we can come back and be better."

In addition to the $10,000 prize for the glasses project, Nagori earned the Thermo Fisher Scientific Leadership Award, and Sun received $10,000 for the Lemelson Foundation Award for Invention.

The research remains in the prototype phase, but the students have ambitious expansion plans. They received a $5,000 grant to scale their glasses and reach more of their community.

Nagori outlined their immediate goals: "We're currently working on implementing a lot of our glasses throughout California."

The team is already preparing for larger-scale production. Nagori noted they have "a big 3D printer in my garage right now with 30 Raspbe[rry Pis], 30 cameras, 30 batteries."

The project demonstrates how innovative technology can be developed affordably, contrasting with Silicon Valley companies that are raising billions to develop wearable AI products. While major tech firms invest massive resources, these teenagers proved that impactful solutions can be created for under $100.

Their work addresses a real-world problem for visually impaired students, providing a cost-effective alternative to expensive assistive technology. The combination of AI-powered text recognition, real-time audio conversion, and portable design creates a practical tool for educational accessibility.