On the Application of Adaptive Online Learning based Control on Single Axis Tilting Thrust Vectored Quadcopter

Traditional quadcopter designs entail 4 independent thrusters that can control the dynamics of 6 DoF quad-copter resulting in an under-actuated system. As a consequence of this, traditional quadcopters can not track a desired reference trajectory with arbitrary Euler angles and require two virtual control inputs. Single Tilt axis thrust vectored quadcopter designs were recently proposed for higher operational flexibility rendering it the capability to hover at a point or follow any desired reference trajectory with arbitrary Euler angles. In this paper, online adaptive control was formulated for a single axis tilt thrust vectored quadcopter to control all the states. Adaptive sliding mode control was synthesized for the position control loop and Adaptive online neural network control was used for the attitude control loop. Physical parameters of the thrust vectored quadcopter were varied in real time during simulation and the control formulation was able to adapt to these variations in real-time and successfully track the desired trajectories. Simulation results provided at the end establish the effectiveness of the approach.