LifeLineBot is an autonomous home-care robot designed to improve safety and independence for stroke patients, who often require 24/7 monitoring that caregivers cannot always provide. Built as part of my senior design capstone at UNLV, our three-person team took home the Grand Prize at the Senior Design Competition for this project.

The robot integrates four core systems: wearable vital monitoring via Apple Watch over Bluetooth Low Energy, LiDAR-based autonomous navigation for obstacle detection and room mapping, automated medication dispensing triggered by elevated vitals, and emergency video recording using OpenCV for patient detection. The entire system runs on a Raspberry Pi 5 with a custom-designed power PCB for low-voltage distribution and battery management.

We validated the design through torque calculations, motor testing, and linear actuator analysis to ensure reliable real-world performance. LifeLineBot represents a preventive approach to home healthcare, combining embedded hardware, robotics, and real-time health monitoring into one affordable system.

Diabetes affects millions of people worldwide, yet most existing devices focus solely on systemic glucose control without addressing the localized complications that come with it, such as impaired circulation, poor wound healing, and increased risk of amputation. This project set out to change that.

Working alongside a research partner under the mentorship of Dr. Shengjie Zhai at UNLV, I co-developed a portable, two-in-one biofeedback device that combines non-invasive glucose monitoring with Electronic Muscle Stimulation (EMS) therapy in a single affordable system.

The glucose monitoring system uses near-infrared (NIR) spectroscopy to measure skin light absorption through the index finger, processing the data through a linear regression model to estimate blood glucose levels in real time. Results were validated against invasive glucose measurement methods to verify accuracy.

The EMS therapy component delivers controlled pulse width modulation (PWM) signals at 40-50 Hz with adjustable strength ranging from 100-400 microseconds to stimulate the calf muscle, improving localized blood flow and circulation for diabetic patients. The system was validated using a logic gate analyzer to confirm signal accuracy.

This project was presented as a research poster at UNLV and represents a meaningful step toward accessible, home-based medical technology for the diabetic community.

As medical technology becomes increasingly reliant on imaging for diagnostics, the need for high-resolution, AI-integrated imaging systems has never been greater. This project explored the design and implementation of a CMOS/CCD-based imaging system tailored specifically for healthcare applications such as skin abnormality detection, wound healing monitoring, and early disease diagnosis.

The system architecture follows a pipeline approach: a CMOS or CCD image sensor captures raw image data and passes it to an Image Signal Processor (ISP) for noise reduction and feature enhancement. The processed images are then analyzed by an AI module using deep learning algorithms to detect healthcare abnormalities. A microcontroller manages data flow and system operations, while a wireless communication module transmits results to a user interface for visualization.

For hardware, I evaluated and selected components including the Arducam IMX219 CMOS sensor, Sony ICX285 CCD sensor, Intel FPGA for signal processing, Analog Devices ADV7280A, and NVIDIA Jetson Nano for AI processing, with an ESP32 serving as the microcontroller.

Implementation was planned around MATLAB simulation for initial image capture validation, TensorFlow for training the AI model on medical image datasets, and iterative prototyping through to physical testing. This project deepened my understanding of imaging hardware, AI integration, and the unique challenges of building reliable diagnostic tools for healthcare environments.