Research Advances in YOLO-Based Obstacle Detection Algorithms

School of Computer Science, Hubei University, Wuhan, Hubei province, 430064, China

^* Corresponding author: 202231116020148@stu.hubu.edu.cn

Abstract

With recent breakthroughs in artificial intelligence and computer vision, obstacle detection has become an essential capability for enhancing safety and enabling automation across various industrial sectors. Among the range of methods available, YOLO-series algorithms have gained widespread adoption in practical applications due to their effective balance between inference speed and detection accuracy. This paper presents a systematic literature review of studies on YOLO-based obstacle detection published between 2023 and early 2025. It focuses on ten representative works that highlight significant advances in areas such as multimodal sensor fusion, lightweight model deployment, and scenario-specific optimizations—including applications in underground mining, agricultural robotics, and low-altitude UAV missions. By summarizing the technological evolution and comparing model performance across different constraint conditions, this study addresses the current absence of comprehensive survey work in this rapidly evolving field. Furthermore, it provides a structured analysis of improvements in network architectures, training strategies, and evaluation protocols, delivering valuable insights for researchers and practitioners working toward efficient and robust obstacle detection systems.