Thrust Mixing Saturation and BodyRate Control for Accurate
Thrust Mixing, Saturation, and Body-Rate Control for Accurate Aggressive Quadrotor Flight Matthias Faessler, Davide Falanga, and Davide Scaramuzza IEEE Robotics and Automation Letters (RA-L)
System Overview High-Level Part Position Controller Low Level Part Body Rates Collective Thrust Body Rate Body Torques Collective Thrust Controller Mixer Matthias Faessler – Robotics and Perception Group – University of Zurich Cmds Motors 2
Dynamical System Ø Body-rate dynamics Ø Additionally consider single rotor thrust dynamics Ø Resulting body-torque dynamics Ø Dynamical system with body rates and body torques as states Matthias Faessler – Robotics and Perception Group – University of Zurich 3
LQR Control Design Ø Dynamical system Ø Design infinite horizon LQR controller Ø Resulting LQR control law with feed forward terms Matthias Faessler – Robotics and Perception Group – University of Zurich 4
Controller Trajectory Tracking Performance Matthias Faessler – Robotics and Perception Group – University of Zurich 5
Controller Disturbance Rejection Matthias Faessler – Robotics and Perception Group – University of Zurich 6
Body Torque Estimation Ø Integrate first-order thrust dynamics Ø Load-cell step-input experiments Matthias Faessler – Robotics and Perception Group – University of Zurich 7
Idea Behind Proposed Thrust Mixing Ø Matthias Faessler – Robotics and Perception Group – University of Zurich 8
Iterative Thrust Mixing Ø Initialize iteration Ø Using rotor thrust and drag torque model Ø Iterate Matthias Faessler – Robotics and Perception Group – University of Zurich 9
Yaw Control Performance Matthias Faessler – Robotics and Perception Group – University of Zurich 10
Idea Behind Handling Motor Saturations Ø Recall thrust mixing Ø Example: one motor saturates Ø Proposed prioritizing saturation Matthias Faessler – Robotics and Perception Group – University of Zurich 11
Prioritizing Motor Saturation Matthias Faessler – Robotics and Perception Group – University of Zurich 12
Computation Times Matthias Faessler – Robotics and Perception Group – University of Zurich 13
Video Matthias Faessler – Robotics and Perception Group – University of Zurich 14
- Slides: 14