<- Portfolio

UAV ESKF, LQR, and LQG Simulation

A fixed-wing UAV simulation combining 6-DOF dynamics, Error-State Kalman Filtering, LQR control, and LQG closed-loop validation, with my main contribution focused on Kalman filter implementation.

Intro

This project belongs in the portfolio because it connects directly to the estimation and controls work behind TARAF. Instead of pedestrian motion, the system modeled a fixed-wing UAV and combined state estimation with feedback control in simulation.

My main contribution was implementing the Kalman filter side of the project.

Purpose

The goal was to understand how modeling, estimation, and control interact in a closed-loop system. The project used an Error-State Kalman Filter for state estimation and LQR/LQG control for stabilizing and guiding the simulated aircraft. Working through UAV dynamics in simulation reinforced concepts, especially quaternion math, error-state representation, and validation discipline, that carry directly into TARAF’s foot-mounted navigation work.

Implementation

  • 6-DOF fixed-wing UAV dynamics
  • Quaternion-based state representation
  • 20-state Error-State Kalman Filter
  • LQR control synthesis
  • LQG closed-loop simulation
  • Unit testing and simulation validation

Design decisions

DecisionSelected directionWhy it fit the course projectTradeoff / next step
EstimatorError-State Kalman FilterHandles nonlinear attitude representation while keeping correction states manageableMore implementation work than a simpler complementary filter
Attitude mathQuaternion stateAvoids singularities and matches aerospace estimation practiceRequires careful normalization and error-state handling
ControlLQR with LQG integrationConnects estimation quality to closed-loop tracking behaviorSimulation-only validation
ValidationUnit-test harness and scenario checksMade estimator behavior easier to trust under coursework pressureMore flight cases would improve confidence

Results

  • Completed ECE 263 course project
  • Implemented the Kalman filter portion of the simulation
  • Passed the full unit-test harness for the estimator work
  • Demonstrated ESKF and controls concepts in a UAV simulation context
  • Reinforced quaternion math, state estimation, and validation discipline later applied in TARAF