Projects
Adaptive Flight Control of Fixed-Wing UAVs (PCAC)
Research & Flight TestsFlight-test demonstration of Predictive Cost Adaptive Control (PCAC) implemented on an autonomous fixed-wing aircraft without prior aerodynamic modeling. Combines recursive least squares (RLS) with variable-rate forgetting (VRF) for online identification and receding-horizon MPC solved via dual-gradient projection. Deployed in ROS2 on an NVIDIA Jetson companion computer interfaced with ArduPilot for real-time trajectory tracking.
Velocity-Independent Wheel Slip Control
Vehicle Dynamics & ControlFormulated a velocity-independent traction control strategy using model-based wheel angular acceleration thresholds derived from a 5-DoF longitudinal half-car model coupled with a nonlinear Pacejka tire curve. Developed a three-stage discrete filtering pipeline (Kinematic Kalman Filter, Causal Savitzky-Golay polynomial, and Statistical Shrinkage) to eliminate sensor differentiation noise with minimal phase lag. Implemented an asymmetric, slew-rate-limited PI torque controller that suppresses deep four-wheel slip and driveline chatter across RWD and AWD configurations.
Camera GNSS Geotagger & Telemetry Module
Embedded SystemsDesigned a low-power external GNSS geotagger and OLED display module that wakes automatically on camera power-on via an RTC hardware interrupt and enters deep sleep 10 seconds after power-down. Uses an ESP32 to read a SparkFun u-blox NEO-M8U receiver over I2C (UBX protocol at 3 Hz), synthesize checksummed GPGGA/GPRMC NMEA sentences at 4800 baud for direct Nikon camera geotagging, and filter battery voltage via a 10-tap FIR filter. Implemented high-side peripheral power gating using an FQP27P06 P-channel MOSFET and a 22kΩ pull-up resistor to isolate peripheral quiescent draw during deep sleep.