Publication

Abstract : This paper presents an integrated framework for real-time pothole detection and autonomous path planning addressing critical road safety challenges in autonomous vehicles. The objective is to develop a unified system combining YOLOv8 computer vision with multi-algorithm path planning including A∗, RRT, and DWA for effective hazard avoidance. The methodology implements dynamic algorithm selection based on environmental conditions and threat severity assessment. Experimental validation using 348 real-world images demonstrates 87.4% detection accuracy with 55 ms processing latency, identifying 1,354 potholes with 78.9% high-confidence detections. The integrated system achieves 97% avoidance success rate with total response time under 200 ms, meeting real-time autonomous vehicle requirements. Statistical analysis validates system reliability across diverse road conditions and traffic scenarios. The significance lies in bridging the gap between isolated detection systems and comprehensive autonomous safety frameworks, providing empirically validated integration of perception and planning modules for practical deployment in intelligent transportation systems.