Model Predictive Control for Visual-Servo Tracking in Unmanned Aerial Vehicles: A Scoping Review

Abstract

Model predictive control (MPC) has become a central receding-horizon strategy for robotic systems subject to constraints, uncertainty and multiple objectives. In unmanned aerial vehicles (UAVs), its integration with visual servoing makes it possible to formulate tracking problems directly in the image space, keep targets inside the field of view, anticipate vehicle dynamics and coordinate safety, actuation and perception constraints. This scoping review characterizes the available evidence on MPC applied to visual-servo tracking in UAVs. The process identified 20 core studies, 11 strong complementary studies, 12 borderline studies and 18 non-ideal records for the defined scope. The core literature reveals four dominant lines: tracking of targets and visual features, landing or docking on moving platforms, perception-aware control with visibility constraints, and robust, distributed or learning-augmented MPC extensions. The findings show a recent consolidation of methodological directions, although experimental maturity remains uneven across the field, with a predominance of nonlinear MPC, predictive image-based visual servoing, field-of-view constraints and validation in simulation or UAV platforms. Persistent gaps remain in reproducibility, standardized comparison, visual latency, depth uncertainty, outdoor validation and formalization of perceptual costs.