Research & publications

Lightweight autonomy for real robots.

My graduate research focused on improving the reliability of visual features for real-time GPS-denied quadrotor navigation.

IEEE International Conference on Robotics and Automation (ICRA) · 2026

Lightweight Learning-Based Feature Selection for Real-Time Optical Flow Navigation on a Quadrotor Platform

Ali Abosaad, Jinjun Shan

A compact CNN prunes unreliable visual features, while adaptive thresholding, Lucas–Kanade optical flow, IMU/range compensation and EKF fusion enable efficient GPS-free quadrotor state estimation.

Read paper ↗Related case study →

ICRA2026 publication

Master's thesis · York University · 2026

Lightweight Learning Based Feature Selection for Real Time Optical Flow Navigation

The thesis expands the work into system architecture, sensor calibration, multiple trajectories, lighting and velocity robustness studies, feature-distribution analysis and computational evaluation.