Project Portfolio

👋 Hi, I’m Torin Hamilton

I’m a UCLA graduate with a B.S. in Electrical Engineering. Below you can find some of my past projects.

🚀 Projects

🔷 V.R.O.O.M. – Voice-Responsive Obstacle Overcoming Maze Navigator

A senior capstone project focused on real-time robotics, embedded systems, and intelligent human-computer interaction.

V.R.O.O.M. is a maze navigation game built around a custom-designed RC car (the Maze Navigator) that explores a physical maze with hidden, virtual walls. The player must guide the car using either voice commands or a custom-built physical controller, discovering walls through trial and error. The game features a PyQt5 desktop interface that visualizes the maze, tracks the car’s position, and offers an omniscient view of the game.

🧠 System Overview

Maze Navigator: A Raspberry Pi–based robotic car with onboard camera and gyroscope, capable of local sensor-based navigation and autonomous turning/forward movement.
Maze Program: A desktop application responsible for maze generation, visualization, and communication across all subsystems.
Controller: A handheld interface built with physical buttons that sends commands via a network connection.

🔧 Tech Stack Breakdown

Subsystem	Tech & Tools
Maze Navigator	Raspberry Pi 4, BerryIMU, PiCamera2, L298N, DC Motors, Python, OpenCV, TCP socket server, IMU gyroscope turning, ROI-based line detection
Maze Program	PyQt5 GUI, recursive backtracking maze algorithm, Python, OpenAI Whisper, `difflib` fuzzy matching, TCP socket client
Controller	Raspberry Pi GPIO, Python, physical buttons, TCP socket server

🗣️ Key Features

Natural Language Voice Control: Enabled flexible user commands via offline speech processing using OpenAI Whisper and difflib fuzzy matching.
Automated Forward Navigation: Guided the robot’s forward movement using line detection implemented with OpenCV (grayscale conversion, binary thresholding).
Precise Automated Turning: Achieved consistent rotational accuracy using closed-loop control based on BerryIMU gyroscope data.
Real-time Visualization Dashboard: Provided live feedback on robot position, detected walls, and game state via a custom PyQt5 GUI.
Distributed System Architecture: Employed a modular design using TCP sockets for robust, real-time communication between two Raspberry Pis and the central program.
Simplified Startup: Automated the launch sequence of all system components using systemd, enabling a simple power-on-and-play user experience.

📹 Demo & Presentation

👨‍💻 My Contributions

Custom Maze Navigator Designed and built the full hardware stack for the Maze Navigator, including the Raspberry Pi 4, L298N Motor Driver, two TT motors, BerryIMU, camera module, and battery pack.
Voice Recognition System Designed and implemented the full voice control interface using OpenAI Whisper with fuzzy keyword matching via Python’s difflib module to support flexible natural language commands.
Gyroscope-Based Turning Developed the IMU-based turning algorithm using BerryIMU. Tuned parameters and resolved sampling issues to achieve reliable 90° rotations on command.
Custom Controller Interface Built a handheld controller using Raspberry Pi GPIO and Python. Implemented TCP server communication to maintain stable control signal transmission to the Maze Program.
Computer Vision Integration Implemented OpenCV-based line detection using Region of Interest (ROI) analysis and binary thresholding for automated forward movement as the player guides the Maze Navigator through the maze.
Maze Program & GUI Collaborated on the development of the main application, including the PyQt5 GUI (real-time visualization) and the integration of game logic.
Networking Architecture Helped implement and debug TCP socket communication across all subsystems for stable real-time data exchange and reduced system latency.

🔮 Future Improvements

Design custom enclosures for the Maze Navigator and controller
Implement automatic maze reset functionality
Improve motion precision with finer motor control ranges
Add encoders for motor feedback and localized error correction

🔗 View GitHub Repo

🔷 Line-Following Robot

Languages/Tech: Arduino, C++, PID Control

Programmed an autonomous robot using C++ on an Arduino platform to navigate black-line tracks using infrared sensor feedback. The core challenge was implementing and tuning a PID control algorithm to manage motor speeds for curve handling and maintain course. The robot successfully demonstrates functional path following, staying consistently on the line as intended.

📹 Watch Demo Video

Note on Video: This video shows the successful implementation of the PID controller and line-following logic. While the robot reliably stays on track, the motion reflects the initial tuning stage and further parameter adjustments could enhance movement smoothness.