research

My research sits between data-driven methods and flight control — building estimators and controllers that stay reliable under faults, uncertainty, and the loss of GPS, and carrying them onto real flight hardware. Four threads run through most of my work.

1 · Data-driven modeling & control

I learn vehicle dynamics directly from data using sparse identification of nonlinear dynamics (SINDy), the Koopman operator, and Gaussian-process regression, then build controllers on top: nonlinear disturbance observers, model-predictive control, and model-reference adaptive control. The goal is control design that adapts to the vehicle it actually flies on, rather than the model we assumed.

2 · Fault diagnosis & fault-tolerant control

For multirotors, eVTOL, and urban air mobility, I develop methods to detect, isolate, and accommodate actuator and sensor faults in real time — including control allocation for over-actuated airframes and redundant, voting-based flight-control architectures. Much of this targets the reliability that certifiable air mobility will require.

3 · GNSS-denied navigation

When satellite navigation is jammed or unavailable, I estimate a vehicle’s position by fusing quantum magnetometers and magnetic-anomaly maps through factor graph optimization, alongside INS/GNSS integration and multi-sensor navigation. This connects estimation theory to real quantum-sensing hardware for passive, GPS-independent positioning.

4 · GNC systems & hardware integration

I care about closing the loop between algorithm and airframe: triple-redundant flight-control computers, hierarchical voting, hardware-in-the-loop simulation, and flight campaigns on platforms from quadrotors to hoverbikes. The algorithms have to survive contact with real hardware, not just simulation.

For the full record, see my publications and CV.