Autonomous Vehicle System Based on A* Pathfinding and improved YOLOv11

This paper presents an autonomous vehicle system that integrates the A* algorithm for route planning, YOLOv11 for road obstacle detection and Semi-Global Block Matching for distance measurement. The proposed system is evaluated in a simulated driving environment using AirSim, which provides realistic scenarios for testing and validation. The integration of these advanced algorithms enhances the navigation capabilities of autonomous vehicles by ensuring efficient path planning and robust obstacle detection. Based on YOLOv11 algorithm, this paper improves the optimizer function and post-training quantization technique to study obstacle recognition and localization in autonomous driving. The improved YOLOv11 model achieved an average accuracy of 92.3% and an average recall of 85.0%, while the recognition speed is 64.4 frames per second.