Image-Based Visual Servoing Control of Quadrotor with MsQL Method

In this paper, a multi-step Q-learning (MsQL) method is employed to design an image-based visual servoing controller of the quadrotor, where image moment of the target image is used as the image feature for information extraction. The virtual camera and virtual image plane are introduced to construct the image dynamics model to realize the decoupling of the quadrotor. Combined with the quadrotor dynamics model, the quadrotor-image dynamics is proposed. The MsQL method is used to design the optimal control law for the horizontal motion of quadrotor. The control policy is updated through the dataset, such that the control gain can be solved and the desired trajectory of roll angle and pitch angle can be calculated. Simulation studies are conducted on the Matlab platform and good results are obtained.