I led the development of unsupervised UAV flight controls for a competitive, tournament-style robotics challenge: spanning simulation, estimation, control, perception, and onboard planning.

I established the build by creating a Simulink digital twin of our under-actuated quadrotor. Using Runge–Kutta to propagate the simulation, I was able to model flight dynamics, noisy IMU/camera/GNSS sensors, and closed-loop feedback control to enable high-fidelity software-in-the-loop (SITL) testing. 

Employing sensor fusion, I derived attitude through Wahba’s problem and initialized an Unscented Kalman Filter to estimate vehicle state under realistic noise and bias conditions. Subsequent work entailed tuning PID controllers to achieve stable, responsive flight across core maneuvers.

Building on this validated framework, I worked with a team to develop a C++ codebase implementing real-time computer vision, obstacle avoidance, and A*-based path planning with environmental awareness (wind and visibility).