Integrated YOLOP Perception and Lyapunov-based Control for Autonomous Mobile Robot Navigation on Track

2025-12-01

AI Summary

Autonomous lane-following navigation for non-holonomic differential-drive robots operating without global localization or high-definition maps in dynamic, partially observable lane environments remains highly challenging. Method: This paper proposes a real-time end-to-end visual navigation framework integrating YOLOP-based multi-task perception, 2D-to-3D lane-line reconstruction, arc-length-uniform sampling, and robust cubic polynomial fitting via QR decomposition; it further introduces a Lyapunov-based nonlinear controller ensuring strict stability of the perception–planning–control loop. Results: The system runs in real time (>30 FPS) on embedded hardware, generates smooth trajectories, and guarantees asymptotic convergence of both lateral position and heading errors. Experimental evaluation demonstrates significantly enhanced robustness and adaptability in dynamic scenarios, marking the first end-to-end visual navigation framework with provable closed-loop stability under partial observability and motion constraints.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Vision for Robotics & Autonomous Driving
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Browser and end-point security

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.

Problem

• Develops a real-time autonomous track navigation system for differential-drive robots.
• Integrates visual lane perception with a stable Lyapunov-based control algorithm.
• Ensures reliable navigation without HD maps or global GPS localization.

Innovation

• YOLOP perception with 2D-to-3D lane reconstruction
• Lyapunov-based stable tracking controller design
• No reliance on HD maps or global localization