ERA Banner
Download Add to Cart Share
More Like This
  • Nonlinear State Estimation and Modeling of a Helicopter UAV
  • Barczyk, Martin
  • English
  • Helicopter
    Kalman Filter
    Aided Navigation
  • Jan 25, 2012 12:16 PM
  • Thesis
  • English
  • Adobe PDF
  • 3775941 bytes
  • Experimentally-validated nonlinear flight control of a helicopter UAV has two necessary conditions: an estimate of the vehicle’s states from noisy multirate output measurements, and a nonlinear dynamics model with minimum complexity, physically controllable inputs and experimentally identified parameter values. This thesis addresses both these objectives for the Applied Nonlinear Controls Lab (ANCL)’s helicopter UAV project. A magnetometer-plus-GPS aided Inertial Navigation System (INS) for outdoor flight as well as an Attitude and Heading Reference System (AHRS) for indoor testing are designed, implemented and experimentally validated employing an Extended Kalman Filter (EKF), using a novel calibration technique for the magnetometer aiding sensor added to remove the limitations of an earlier GPS-only aiding design. Next the recently-developed nonlinear observer design methodology of invariant observers is adapted to the aided INS and AHRS examples, employing a rotation matrix representation for the state manifold to obtain designs amenable to global stability analysis, obtaining a direct nonlinear design for gains of the AHRS observer, modifying the previously-proposed Invariant EKF systematic method for computing gains, and culminating in simulation and experimental validation of the observers. Lastly a nonlinear control-oriented model of the helicopter UAV is derived from first principles, using a rigid-body dynamics formulation augmented with models of the on-board subsystems: main rotor forces and blade flapping dynamics, the Bell-Hiller system and flybar flapping dynamics, tail rotor forces, tail gyro unit, engine and rotor speed, servo operation, fuselage drag, and tail stabilizer forces. The parameter values in the resulting models are identified experimentally. Using these the model is further simplified to be tractable for model-based control design.
  • Doctoral
  • Doctor of Philosophy
  • Department of Electrical and Computer Engineering
  • Control Systems
  • Spring 2012
  • Lynch, Alan (Electrical and Computer Engineering)
  • Chen, Tongwen (Electrical and Computer Engineering)
    Tavakoli, Mahdi (Electrical and Computer Engineering)
    Jagersand, Martin (Computing Science)
    Tayebi, Abdelhamid (Electrical Engineering, Lakehead University)