Real-time model predictive control for keeping a quadrotor visible on the camera field-of-view of a ground robot

This paper considers a cooperative control design for an aerial/ground robot system, and addresses the problem of maintaining visibility of a quadrotor within the camera field-of-view of a ground robot in the presence of external disturbances. The quadrotor needs to be tracked by the ground robot with a monocular camera, and hence its motion should facilitate the ground vision-based tracking process by remaining in the effective camera sensing area. We design a model predictive controller (MPC) strategy where the visibility constraints of the camera and the control input constraints of the quadrotor are encoded into the cost function via barrier functions, and we adopt a fast MPC solver that is able to solve the optimization problem in real time. We also propose a method to enhance the robustness of the algorithm by suitably defining a restart method for the MPC solver. The applicability of the proposed algorithm is demonstrated through simulations and experimental results on real setups.