分析了目前农业领域中几种常用的自主车辆导航方式, 并说明了多传感器融合的必要性。着重探讨了卡尔曼滤波、模糊逻辑推理以及人工神经网络3 种融合方法, 分析了未来的发展方向。 叙词: 车辆 导航 传感器 Abstract In th is paper, som e general navigat ion m ethods of au tonomou s veh icle u sed in the agricu ltu re w ere analyzed concisely, and the essen t iality of m u lt i2sen so r fu sion w as derived. Then som e m ethods of m u lt i2sen so r fu sion w ere in t roduced, and the at ten t ion w as paid to the m ethod of Kalm an f ilter, fuzzy logic inference and art if icial neu ral netwo rk. Kalm an f ilter can p rovide an op t im al stat ist ical resu lt, and is a po ten t ialm ean s in the f ield of m u lt i2sen so r fu sion. Bo th fuzzy logic inference and art if icial neu ral netwo rk can successfu lly estab lish a comp lex non2linear m ap betw een the sen so r data and the data needed. It show ed that all of them cou ld imp rove the p recision and reliab ility of the system by review ing m any navigat ion p ract ices in recen t years. F inally, the p ro spect of m u lt i2sen so r fu sion techno logy in the navigat ion of au tonomou s veh icle w as discu ssed. Key words V eh icles, N avigat ion, Sen so rs