Our project utilizes machine learning to create a non-intrusive home appliance monitor. When attached to the main power coming into a house, the device will predict the appliances that are in operation at any given time to help the user evaluate their power consumption.

Team Members

Kevin Aguilar, Kai Cheu, Pulkit Mahajan, Travis Mongoven, Santhosh Saravanan, Joey Tombari, Brent Zhuang

MLP-Shaker is a model that adapts to the image classification task at hand, learning all that it needs from the data itself. Complex convolutional neural networks and vision transformers that dominate the image classification field rely on inductive biases for vision tasks. Meanwhile, our architecture starts with a fully connected multi-layer perceptron (MLP) – called MLP-Mixer – and prunes the unneeded weights. Pruning weights is important to reduce classification times and improves the model's generalization to various images.

The goal of this project is to create a physical security system that uses nanofabrication to make keys that are irreproducible and work in parallel with other forms of authentication. The optical transmission spectra of the nanofabricated keys contain identifying information used to authenticate the key-owner. Few companies or individuals have access to tools that can fabricate keys with nanometer-scale thin films. Furthermore, uncontrollable and random fabrication variations create uniqueness in the keys, meaning that a key cannot be reproduced by an attacker.

In this project, we designed a compact near-field antenna measurement system for experimental purposes. Our system comprises of a positioner with a probe antenna, an antenna under test (AUT), a vector network analyzer (VNA), and a PC. The PC sends position commands to the positioner over USB/UART. The PC also sends measurement commands to the VNA. The VNA sends an analog RF signal through a wire to the probe antenna which transmits the same signal wirelessly to the AUT. The VNA measures the signal received by the AUT. The VNA then sends the measurement data to the PC.

In the shadow of the COVID-19 pandemic, the ability to quickly and accurately test for disease has never been more important. Our Senior Design team developed a portable silicon-based biosensor for biomarker detection. Our system utilizes a photonic chip that can detect the presence of biomarkers associated with any type of disease. Intended to act as a ‘lab on a chip’, our system aims to enable users to rapidly test for disease even in remote areas where access to testing equipment is limited.

COVID-19 pandemic has burdened today’s healthcare, supply chain, and socio-economic systems, testing the world’s ability to make informed public health policy decisions. In the current data-driven age, can we better gauge the severity of COVID-19 infections?

This project is motivated by the difficulty in standard WiFi localization due to poor signal propagation within buildings due to environmental blockage and other interfering signals. Indoor localization is critical for for 6G applications such as compliant manufacturing, indoor navigation, and autonomous vehicles. The project uses machine learning and chirp signal positioning approaches to get the centimeter accuracy of location of user equipment.

Team Members

Gagan Kaushik, Jean Lee, Brian Menezes, Matthew Qin, Justin Swinney

whispr is an assistive wearable device that uses facial EMG signals to interpret mouthed speech. Our device collects data from six EMG sensors placed around the face and neck and sends the data in real-time to our software suite over Bluetooth. Users are able to silently communicate from a 14 word dictionary with a high level of accuracy.

Team Members

Adithya Ashok, Christian Boswell, Sai Koukuntla, Lindia Tjuatja, Logan Winger

We created a software tool that researchers and students alike can use to learn more
about the constructs and behaviors of chemical linkages. Several features were developed such as an intuitive user interface, the ability to create and remove vertices, fix/unfix vertices, auto-merge joints, label points, and a color key, etc. Our team created this tool with the intention of providing a starting point for further research and development in the field of chemical linkages.

Team Members

ERCOT manages the electricity infrastructure and market in Texas. This senior design team created an interactive web dashboard that improves upon the existing ERCOT dashboard. The dashboard features data visualizations and easy-to-access explanations for the important features of the ERCOT data. The current ERCOT data available on the web dashboard are System-Wide Demand, Generation by Fuel Types, System Frequency, Wind and Photovoltaic Power Generation, and Market Prices.