Coursework and research from my master's at NYU and my degree in Auckland — perception, optimal control, reinforcement learning and mechanical design. It predates my work at Eyebot, but it is where most of this started.

Perception

LoFTR in TensorFlow

2023 / New York University

A PyTorch state-of-the-art feature matcher re-implemented in TensorFlow, with a custom training loop, custom loss, and a transformer stage for coarse-to-fine matching.

Perception

Optimized Image Stitching

2023 / New York University

Feature matches drive a homography, the second frame is warped onto the first, and dynamic weighted fusion dissolves the seam line that pasting leaves behind.

Optimal control

iLQR Quadrotor Control

2022 / New York University

LQR and iLQR controllers flying a 2D quadrotor through acrobatics — circular tracking, a vertical hold, and a full 360° flip.

Reinforcement learning

Q-Learning Pendulum

2022 / New York University

Learning a swing-up policy for an inverted pendulum with no model of the system — 6000 episodes from flailing to balanced.

State estimation

Vision-Based Pose & Velocity Estimation

2022 / New York University

Pose recovered by decomposing a planar homography over an April Tag grid; velocity from KLT feature tracks, cleaned up with RANSAC outlier rejection.

Perception

Augmented Reality in Python

2022 / New York University

Calibrate against an Aruco grid, recover the camera pose relative to the tag, then project a 3D cube back onto the image plane in real time.

Robotics

Intelligent Line Following Robot

2022 / New York University

Line following and path planning around an obstacle course — first on infrared alone, then with vision added to see further ahead.

Mechanical design

Dexterous Manipulator

2020 / The University of Auckland

A compliant, under-actuated gripper that adapts to what it grasps — lightweight, rapid-prototyped, and forgiving of uncertainty.