Projects, Fall 24 | Ziad Youssfi

Weather Prediction Using tinyML

No Wi-Fi or cell connection, no problem. This weather station can forecast the clouds without the “cloud”!

By Parsheeta Roy and Shreya Ajay Kale

Posted on January 1, 2025

This project aims to develop a cost-effective weather forecasting system using the Arduino Nano 33 BLE Sense. By utilizing Tiny Machine Learning (TinyML), our model can run efficiently on constrained hardware, enabling real-time weather predictions without internet connectivity. This system can be particularly relevant for users in resource-limited areas (including... [Read More]

Tiny Tennis

How good is your swing?

By Ann Kang and Arti Jain

Posted on January 1, 2025

Tiny Tennis Coach is a wearable device strapped to a player’s wrist that can analyze a player’s swing and provide targeted feedback to help improve a player’s form. Inspired by our own experiences playing tennis, we aimed to create a convenient and real-time learning aid for beginners to get feedback... [Read More]

Smart Door Lock

Unlock a door just like you unlock your iPhone-- by looking at it!

By Belle Connaughtand and Elliot Clark

Posted on January 1, 2025

This project presents a smart door lock system integrating face and voice recognition to enhance security and accessibility in home or office environments. The system leverages two microcontroller boards: Arduino Nicla for face detection using MobileNetv1 and Arduino Nano 33 for audio detection. Both models are trained and deployed through... [Read More]

Smart Alarm Clock with Voice Control and Motion Recognition

It won't stop until you get out of bed!

By Yuhong Yao and Lai Jiang

Posted on January 1, 2025

We present a smart alarm clock that uses embedded hardware and onboard inference to recognize voice commands and detect wakeful motion before disabling. The system runs compact models on a low-power platform and remains active until it confirms full wakefulness through spoken or physical cues. [Read More]

American Sign Language Fingerspelling Interpreter

Translate sign language to characters and words.

By Rosina Ananth, Nathan Chow, and Edward Lu

Posted on January 1, 2025

American Sign Language (ASL) is the preferred language for many amongst the deaf and hard of hearing communities. Fingerspelling is a part of ASL which assigns a sign to each letter of the English alphabet and is used for proper names or when signs for words are not known. Since... [Read More]

Gestures Detection with Oximeter

If the Apple Watch can do it, so can we!

By Bangjie Xue and Qi Luo

Posted on January 1, 2025

This project aims to replicate the Apple Watch double-tap feature using on-device machine learning with an Arduino Nano BLE Sense 33. The device leverages its built-in accelerometer, gyroscope, and an externally attached MAX30102 oximeter sensor 1 to detect gestures such as double-tap, clench, and rotate. The hardware is housed in... [Read More]

Deer Detection

Helps you avoid deer on the road.

By Amman Kazi, Akshat Sahay, and Swamynathan Siva

Posted on January 1, 2025

The goal of this project is to design a real-time deer detection system for vehicles using embedded technology. We use sensor fusion (thermal imaging, TOF sensors) and onboard processing to catch deer near the road and warn drivers faster than they can react. If the system catches a deer, it... [Read More]

Virtual Boxing Trainer

Use it to learn to throw a punch, a hook, an upper, or a block.

By Gokulapriyan Balaji, Prithviraj Vijayaraj, and Revanth Varma Kantheti

Posted on January 1, 2025

The Virtual Boxing Trainer project aims to revolutionize boxing training by integrating wearable technology with intelligent data analysis. Using Arduino Nano BLE 33 boards equipped with inertial measurement units (IMUs), this system records, classifies, and analyzes boxing moves, delivering an interactive and engaging training experience for users. Designed for fitness... [Read More]

Baseball Simulator

Can you hit a home-run?

By Belle Connaughtand and Elliot Clark

Posted on January 1, 2025

Our project implements a real-time baseball detection and scoring system using YOLOv8 for object detection and Bluetooth Low Energy (BLE) for external device communication. The system integrates a webcam to capture live video frames, processes them with a pre-trained YOLOv8m model to detect baseballs, and overlays detection results with bounding... [Read More]

AirDrums

Use your bare hands to drum away if you can’t have a drum set!

By Max Tang and Jiarong Wang

Posted on January 1, 2025

AirDrums: a virtual drum set powered by deep learning. AirDrums packages an entire drumset into just a wristband with built-in sensors. By wearing the wristband and making the motions of hitting a drum set with a drumstick, we can use deep learning to identify the motion made and which drum... [Read More]

PocketPro

Are you a beginner, intermediate, or advanced-level ping-pong player? You can find out by wearing it, and you can also get feedback.

By Brendan Wilhelm and Kobe Zhang

Posted on January 1, 2025

Self-learning anything can be challenging. Our device, the PocketPro, is an embedded device that can be attached to a glove to allow table tennis enthusiasts and beginners to conveniently train their technique in practice by utilizing motion classification to classify and compare a user’s actions with professional data. The PocketPro... [Read More]

DuoGuard: Advanced Elderly Care Fall Detection System

Not every fall is the same, and some falls are more critical than others. This wearable can distinguish them and raise an appropriate alarm.

By Vishal, Aniruddhan, and Sai Shuba

Posted on January 1, 2025

DuoGuard represents a significant advancement in elderly care monitoring systems, combining audio and motion detection through a dual-board TinyML implementation. This paper presents a comprehensive overview of our system, which achieves 90% detection accuracy while maintaining strict privacy standards through on-device processing. The solution addresses the critical need for non-... [Read More]