Luke Rooney

Skills

Python, Scheme

Description

This project implements an interpreter for the Scheme programming language written in Python. The interpreter is powered by 'Read-Eval-Print' loop which parses Scheme inputs commands and computes their output. The 'Read' step tokenizes the input buffer and converts it to an expression represented by a tree data structure. The 'Eval' step then recursively evaluates the expression given its tree form. The interpreter also supports built-in and user-defined Scheme functions.

Skills

MSP430 Microcontroller, PCA, Filter Circuits, Op-amps, K-means Clustering, PID Control

Description

This is a simple voice-controller car which is powered by two 9V batteries, controlled by a MSP430 microcontroller, and actuated by two small DC motors. The steering PID controller has inputs which are derived from voice commands. The voice commands are picked up by a microphone, fed through a custom amplification and filtering circuit, then input to a word detection algorithm. The algorithm performs Principle Component Analysis (PCA) on the voice command. This serves the purpose of reducing the data dimensionality of the time-domain voice command into two dimensions so word matches could then be identified by a k-means clustering algorithm.

Skills

Logisim, Computer Architecture, RISC Processor, Digital Logic Design, Pipelining

Description

I implemented parts of a RISC-V processor in Logisim, a digital logic simulator. Some aspects of the processor that I created included the ALU, support for 11 registers, over 30 instructions, a two-stage pipeline, and ROM-based control logic.

Skills

C++, Arduino, PID Controller, Motors, Power Systems

Description

I have spent over 200 hours designing, building, and testing a 450mm quadcopter. I purchased each part individually and wrote all of the Arduino-based flight controller code from scratch. During the design phase I performed power budget analysis, wrote a dynamics simulation, and added fault protection as a means of verifying that the system was going to work as intended. This is by far the largest physical project that I have ever worked on and it truly taught me the power of research and learning with a particular application in mind.

Click here to check out a presentation I made on this project.

Skills

MATLAB, Dynamics Simulation, Control Theory

Description

A simulation designed to visualize the dynamics of a quadcopter with customizable parameters. The final simulation reproduces the results obtained in Andrew Gibiansky's research titled Quadcopter Dynamics, Simulation, and Control.

Skills

Operating Systems, Syscalls, Scheduling, File Systems, Synchronization, C, GDB

Description

I was part of a four person group who implemented features on top of a base implementation of the Pintos OS. The features that I worked on specifically include the following:
(1) Setting up the user stack to accomodate arguments passed to user programs.
(2) Implemented file operation syscalls in the kernel to give users the ability to interact with files.
(3) A priority scheduler which supports donation to protect against priority inversion.
(4) An inode-based file system (like the Unix FFS file system) and modified the Pintos base functionality to support an inode structure which supports direct, indirect, and doubly-indirect disk block pointers.
(5) Subdirectory support.
In general, simultaneous reads and writes by different threads to separate disk sectors were allowed, but operations on the same disk sector were synchronized to ensure data consistency.

Skills

Python, Reinforcement Learning, A* Graph Search, Dijkstra's Algorithm, Markov Decision Processes (MDP), Bayes Net and Inference, Particle Filtering

Description

Implemented various uninformed and informed pathfinding algorithms. Modeled Pacman as an adversarial agent which can collect all food in an environment while avoiding ghosts.

Skills

PBC Design, Soldering

Description

Designed 10 PCBs in KiCad implementing analog synthesizer circuitry sound generation, mixing, filtering, envelope generation, and note step sequencing. Assembled over 30 PCBs (many SMT-based) to populate modules in a 3U-by-43cm modular synthesizer.

Skills

Description

In a group of three, my team and I built a line-following car based on the Beaglebone Blue robotics computer. The car consists of a linescan camera which takes 128-pixel samples of the path in front of it. The analog camera data is fed into a robust line detection algorithm whose output informs steering and velocity controllers. Our group was able to achieve speeds of 2 m/s on a non-trivial track, but due to COVID-19 we were unable to add improvements to the physical car. Instead, we turned our attention towards a simulator which simulates the dynamics of the physical car. From here, I took the lead on implementing a learning algorithm which maps the track during an initial lap to be used as feedforward input in the steering and velocity controllers in later laps. Ultimately we fine-tuned the car to achieve an average lap speed of 3.5 m/s on the simulated track.

Skills

Go (Programming Language), Symmetric Encryption, Public Key Encryption, RSA Digital Signatures, Message Authentication Codes (MAC/HMAC)

Description

With the help of a partner, our task was to design a file sharing application which remains secure against a malicious actor who has read/write access to the file database. Our implementation preserves confidentiality via encryption for both the files and the user data. Additionally, we support integrity in the form of digital signatures or message authentication codes (MAC) in order detect if any data is tampered with.
The application supports efficient append operations to the files. We devised a pointer data structure with the flavor of a linked-list in order for users to make modifications to a file linear in the size of the data being appended, rather than linear in the size of the file.
Each user has the ability to securely share access to a file, as well as revoke access from a file they have ownership over.
This project was written in the Go programming language.

Skills

Java, Priority Queues, K-d Trees

Description

Bear Maps was a large project given in my data structures class. The project was intended for students to use various data structures to replicate a simplified, efficient version of Google Maps and its underlying algorithms for location searching and tile rendering.

Skills

Java, Minimum Spanning Trees, Dijkstra's Algorithm, Disjoint Sets, Software Engineering

Description

This project implements a simple adventure game in 16-bit style. A world consists of rectangular shaped rooms connected by halls and each map is pseudo-randomly generated based off of a 32-bit input key. The goal of the game is to collect keys randomly distributed throughout the world in order to open a chest. Zombies use a shortest-path algorithm in order to locate and chase the user. World states can be saved and loaded at a later time if desired.