This research introduces a robust framework that dramatically boosts the efficiency and effectiveness of autonomous systems especially in constrained environments with special emphasis on lane-keeping as lanes are a logical obstacle and do not have any physical dimensions. By integrating cutting-edge visual technologies with advanced machine learning, and utilizing ROS 2, we achieve real-time adaptation to obstacles-working better than traditional LiDAR systems in streamlined outdoor navigation. Our technology excels in real-time identification of lane markers, using advanced reconstruction techniques to navigate through traffic and pedestrian congestion quickly and effectively. Our compact, high-performance navigation model for autonomous robots, equipped with an adaptive costmap, ensures seamless maneuvering through complex environments. Fast, powerful, and highly effective, our system is setting new standards in real-time navigation and lane detection, making it a game-changer in the field of autonomous systems.