Homography-Based Predictive Control for Smooth Visual Path Following in Autonomous Robots

This paper presents a novel Nonlinear Visual Model Predictive Control (NVMPC) framework for visual path-following in mobile robots. The proposed method overcomes critical challenges such as velocity discontinuities during transitions between reference images and dependency on global localization systems. By integrating a homography-based predictive model, the controller ensures dynamic trajectory tracking with smooth velocity profiles and high robustness to abrupt path changes. Furthermore, it incorporates constraints to handle kinematic limitations, improving stability and navigation precision. Experimental results demonstrate the superiority of the proposed approach in achieving accurate trajectory tracking and consistent speeds under real-world conditions, including curved and straight trajectories. This work contributes to advancing robust and efficient visual navigation systems for autonomous robots.