Vision-based Fractional Order Sliding Mode Control for Autonomous Vehicle Tracking by a Quadrotor UAV

This paper proposes a vision-based sliding mode control technique for autonomous tracking of a moving vehicle by a quadrotor. The proposed vision algorithm estimates the quadrotor's position relative to moving vehicle using an on-board monocular camera. The relative position is provided as an input to a Fractional Order Sliding mode Controller (FOSMC) which ensures the convergence of the relative position between the moving vehicle and the quadrotor thereby enabling it to track the vehicle effectively. In addition, the proposed controller guarantees robustness towards bounded external disturbances and modelling uncertainties. The proposed vision-based control scheme is implemented using numerical simulations and validated in real-time on the DJI Matrice 100. Theses validations help in gaining into the maximum allowable speed of the moving target for the quadrotor to successfully track the object. This plays a vital role in surveillance operations and intruder chase.