Robust, Fiducial Markers Based Predictive Control Scheme for Infrastructure Inspection with Unmanned Aerial Vehicles
Over the last decades a lot of research effort has been put into the development of new Unmanned Aerial Vehicles (UAVs) and mobile robots for the inspection of critical infrastructure, such as bridges, roads, and dams, in remote locations or dangerous conditions. But despite the increased interest, the application of such autonomous platforms is hindered by the lack of sufficiently accurate localisation methods, especially in GPS-denied environments. Employing fiducial markers to aid such a localisation is an efficient solution that necessitates the use of subsidiary control laws that will allow the autonomous platform to attain a predefined, desired pose with respect to the marker of interest for localisation purposes. In this paper, we utilize Nonlinear Model Predictive Control (NMPC) in a vision-based framework to accurately control a drone in proximity to its target. This approach takes into account the Field of View (FoV) constraints, control input saturation, uncertainties, and external disturbances during the control design phase. Furthermore, the stability and convergence of the closed-loop cascade system have been examined. The efficacy and robustness of the proposed vision-based control strategy are validated through real-time experiments on a compact, custom-build UAV platform.