Amber Miller

Construction + Prototype + "Final Product"

Demo Video

GitHub Repository

Motivation:

The current market for cat toys contains very few options for a toy which both serves live video and allows the user to remotely play with their cat using a laser pointer. Among these devices, it is typical for the video feed to have a 1500-2000 ms delay due to network latency, video encoding, decoding, and compression, and slow webhosting. Additionally, such devices typically cost upwards of $100 and can require a subscription service to access the video feed. The features on these devices are also very limited. As such, I set out to create a low-cost semi-automatic cat toy robot that could stream live video with minimal latency and allow for remote laser pointer play with my cat, but also enable custom features such as customizable laser patterns, including a modular random trajectory generator, the ability to record and playback laser trajectories, and record and save video, enabling the cat toy to double as a security camera.

Project Overview

• Built a semi-automatic cat toy robot with pan-tilt laser targeting controlled via a web interface
• Developed low-latency MJPEG streaming from a Raspberry Pi camera, with live video flip and zoom controls
• Implemented servo motion control with adjustable speed, angle limits, and smooth movement interpolation
• Added random walk and manual targeting modes for interactive play, with recording and replay functionality
• Integrated lidar sensing for distance measurement and real-time telemetry over WebSocket
• Created a full-stack Python + Flask + WebSocket system with authentication, user management, and persistent configuration

Full Report

GitHub Repository

Overview:

A real-time browser ecosystem simulation where six species across four trophic levels hunt, breed, and die inside a procedurally-generated cosmic canvas. Each creature runs its own drive hierarchy and moves independently every frame. Multi-tab sync is handled via Firebase Realtime Database. Population dynamics were modeled with an extended Lotka–Volterra ODE system and validated against live simulation data.

Key Features

• Six species (jellyfish, manta, seahorse, shark, anglerfish, leviathan) with heritable traits, Q-tables, and Lévy-flight wandering
• Procedural cosmic background: 90 stars, 8 planets, 6 suns, 4 galaxies, comets, nebulae, and a day/night cycle driving food bloom spawns
• Firebase Realtime Database sync — multiple browser tabs share one live simulation with automatic master election and offline fallback
• Priority-ordered drive hierarchy per creature: flee > hunt/feed > seek mate > wander, with spiral mate-search and contact-range kill checks
• God Mode panel for live parameter tuning: food multiplier, predator aggression, mutation rate, day speed
• Extended Lotka–Volterra formulation with spatial encounter saturation, energy-gated reproduction, and circadian forcing — linearized stability analysis predicts ~300 s oscillation period, confirmed by FFT of simulation data
• "Chaos UI" easter egg: mouse-trail sparkle effects, cycling neon headers, and a Zerg Rush mini-game where zerglings attack the portfolio cards

Project Overview + Demo Video

Full Report

GitHub Repository

Contributions

• Designed and Tuned LQR Controller and Observer offline using data collected from MoCap system as well as parameters of Crazyflie Drone with MoCap and LED Deck attachments
• Created Pygame Interface for User input to allow for mouse cursor drawings
• Converted cursor-drawn image to numpy array containing pixel location and timestamp data
• Sampled from pixel data to map a set of smooth bezier curves to the image with constraints on curvature to enable dynamically feasible flight trajectory
• Auto-calibrated flight trajectory to drone location and capped maximum velocity by time scaling

Full Report

GitHub Repository

Role: Stability & Control, Cost

Stability and Control:

• Aircraft sizing and configuration optimization using contour plots
• Horizontal tail sizing via scissor plots (static margin & tail stall)
• Trim drag minimization for most fuel-demanding mission segment
• Static and dynamic stability analysis using Roskam & Nelson methods
• Crosswind trim triangle with asymmetric ordnance and fuel loading
• Validation using AVL and aerodynamic drag buildup

Cost:

• Compared Modified DAPCA IV appended with recent cost models for considerations for titanium and composite usage and Roskam cost models to ensure flyaway cost and inform design decisions including material usage, aircraft configuration, and control surface selection
• Ensured flyaway cost < $25M
• Lifecycle and unit cost variation with production quantity
• Titanium trade study replacing up to 2400 lb of aluminum with titanium to ensure flyaway cost while meeting instantaneous turn rate requirments

Aircraft Dynamic Stability Analysis

RDTE Cost Analysis

Operating Cost Analysis

Aircraft Design

Stability & Control

Cost Modeling

Demo Video

Full Report

Dynamical System Derivation

GitHub Repository

Project Overview

• Derived dynamical system from basic drone geometry
• Designed and Tuned LQR Controller and Observer in simulation using Condynsate python package
• Motivated Drone to move through obstacle course by dynamically manipulating the equilibrium

Demo Video

GitHub Repository

Motivation:

While Working at IMEG, the current state of completing fluid volume and pressure drop calcuations involved a lengthy excel spreadsheet where one would measure every individual twist and turn of the pipe and input that along with the pipe dimensions. For a multi-story building, these calcuations could take hours or even days. In addition, older buildings that needed mechanical system renovations often only had 2D blueprints available, and converting these blueprints into a modern BIM model would be a lengthy and expensive process. As such, a method or software was needed that could expedite these calculations, while working on both BIM models and old blueprints. This software must also be intuitive to use and easy to distrubte so as to effectively replace the ubiquitous master excel spreadsheet.

Project Overview

• Developed an interactive PDF-based takeoff and annotation tool tailored for MEP piping layouts
• Implemented CAD-style drawing, snapping, zooming, and multi-page navigation using a custom PyQt workflow
• Integrated pipe sizing logic, scale calibration, and volume calculations directly into the drawing process
• Automated persistence and Excel export to generate system-level quantities and summaries for estimating and design workflows