关键词: 机器视觉；智能车竞赛；多传感器融合；车道线检测；交通标志识别；自主泊车与 导航

Abstract: With the rapid advancement of intelligent driving technology, university-level intelligent vehicle competitions have become an important platform bridging theoretical education and engineering practice, as well as promoting technological innovation and talent development. Based on the 2025 Shanxi Provincial University Intelligent Vehicle Competition, the article systematically elaborates on the application of a multi-sensor fusion-based machine vision system in tasks such as lane detection, traffic sign recognition, parking triggering, and navigation. The system adopts a "perception–decision–control" hierarchical architecture and implements real-time image processing and closed-loop control on the Jetson Nano embedded platform. To address challenges such as varying illumination and track reflection in the competition environment, innovative strategies including virtual lane compensation, hybrid event triggering based on spatiotemporal consistency, and ultrasonic-visual fusion for parking determination are introduced, thereby enhancing the system's environmental adaptability and robustness. Experiments demonstrate that the system maintains high robustness and real-time performance under complex conditions such as varying illumination, providing a reproducible engineering reference for machine vision applications in intelligent vehicle competitions.

Key words: machine vision; intelligent vehicle competition; multi-sensor fusion; lane detection; traffic sign recognition; self-parking and navigation