Examination of Quadrotor Inverse Simulation Problem Using Trust-Region Dogleg Solution Method

In this paper, the particular solution technique for inverse simulation applied to the quadrotor maneuvering flight is investigated. The trust-region dogleg (DL) technique which is proposed alleviates the weakness of Newton’s method used for numerical differentiation of system states in the solution process. The proposed technique emphasizes global convergence solution to the inverse simulation problem. This algorithm is evaluated by calculating the control inputs necessary to enable the quadrotor to follow a specified trajectory including climb-hover and cruise-hover maneuvers. The trajectory is generated by the direct simulation using a linear optimal control developed for the quadrotor. The model of rotors for the quadrotor is a nonlinear model developed based on blade element theory (BET), linear aerodynamics, and non uniform inflow over the rotor disc. The results show that the control inputs obtained from the inverse simulation are in good agreement with control inputs estimated by direct simulation. The results also confirm that the maximum difference between the prescribed trajectory and the trajectory generated by the direct simulation is less than 0.02%, and thus the potential application of the inverse simulation with the trust-region dogleg optimization is evident.