Integrated YOLOP Perception and Lyapunov-based Control for Autonomous Mobile Robot Navigation on Track
Mo Chen
TL;DR
The paper tackles GPS-denied autonomous track navigation by integrating a real-time, multi-task perception module (YOLOP) with a Lyapunov-stability-based controller for a nonholonomic differential-drive robot. Lane centerlines are reconstructed via 2D-to-3D projection, arc-length resampling, and cubic polynomial fitting solved with QR least-squares, producing smooth centerlines for control. The Lyapunov-based controller guarantees bounded error dynamics and asymptotic convergence of position and heading, enabling stable tracking of moving targets using three look-ahead points. Real-world experiments on embedded hardware validate real-time performance, trajectory smoothness, and closed-loop stability, demonstrating robustness in perception-driven and preset-path navigation. The framework offers a practical, map-free approach for autonomous navigation on tracks with potential extensions to adaptive gains and enhanced perception for obstacle-rich environments.
Abstract
This work presents a real-time autonomous track navigation framework for nonholonomic differential-drive mobile robots by jointly integrating multi-task visual perception and a provably stable tracking controller. The perception pipeline reconstructs lane centerlines using 2D-to-3D camera projection, arc-length based uniform point resampling, and cubic polynomial fitting solved via robust QR least-squares optimization. The controller regulates robot linear and angular velocities through a Lyapunov-stability grounded design, ensuring bounded error dynamics and asymptotic convergence of position and heading deviations even in dynamic and partially perceived lane scenarios, without relying on HD prior maps or global satellite localization. Real-world experiments on embedded platforms verify system fidelity, real-time execution, trajectory smoothness, and closed-loop stability for reliable autonomous navigation.