Abstract: Abstract: Unmanned aerial systems can widely be expected to realize the demand for "machine for man" in the field of precision agriculture. Particularly, the workforce is aging in most industrial countries, together with ever-increasing consumption, unreasonable exploitation of resources, and gradually deteriorated ecological environment. Correspondingly, these issues have posed a great challenge to sustainable agriculture in the future, as rapid information technology (IT) is emerging. Nevertheless, current farmland is distributed mainly on a small to medium scale after decades of land use regulation. By contrast, modern agricultural production requires a large scale, precision, and low cost in the direction of automation. Furthermore, most advanced technologies generally present relatively high cost, complex composition, and limited application, such as satellite, visual, and radar navigation. Therefore, it is highly necessary to explore a detection system with low-cost, high-precision, and easy-to-use navigation parameters. Fortunately, ultra-wideband communication (UWB) has received extensive attention in the field of wireless communication transmission in recent years. UWB-based positioning is one of the most promising high-precision technologies, providing convenience and new ideas for autonomous navigation in agriculture and industry. An important new initiative is also currently to vigorously promote land rectification and construction of high-standard farmland at this stage of agricultural production in China. This scheme has laid a solid foundation to realizing autonomous navigation for large-field agricultural machinery in the agronomic aspects. However, the current national large-field agriculture is still dominated by the individual management system and the family joint production contract responsibility system. Specifically, 90% of large-field agriculture is characterized by regional production on a small scale. Fragmented distribution is still the main body with a concentrated planting pattern of small fields at the 100-metre level, especially in the southern paddy fields. At present, there are two types of UWB positioning: short-range and long-range. The long-range module can reach more than 300m, up to 1 200 m, and the optimal positioning accuracy can be known within 5cm. It indicates that the long-range UWB wireless range sensor can meet the demand for autonomous navigation and the driving of farm machinery in large fields in terms of range and detection accuracy. Furthermore, there is also local, moderately small-scale and scattered production, particularly on the large-field cultivation patterns in the southern paddy fields of China. The main factors are confined to the development of satellite, visual and radar navigation, including the high investment, complicated structure, limited applicable environment, and susceptibility to environmental interference. In this study, an improved signal angle of arrival (AOA) model was proposed to detect the navigation parameters under the reciprocal operation patterns of agricultural machinery in an unmanned environment in a large field. The UWB-based station tag was adopted as the detection sensor. Two arrangements were also designed to achieve fast and accurate detection of navigation parameters during the unmanned operation of agricultural machinery, including the double base station-body longitudinal double tag (TBZ) and double base station-body transverse double tag (TBH). The static test results show that the detection accuracy of navigation parameters improved significantly for the TBZ and TBH arrangements, as the tag or base station spacing increased. Specifically, the detection error of lateral deviation was ≤8 cm, and the heading deviation tended to be close to 0, but not greater than 1°. An orthogonal test was combined with the analysis of variance (ANOVA), thereby clarifying the significance of key parameters in the TBZ and TBH arrangements on the detection of lateral and heading deviation, and finally to determine the main and secondary factors for the optimal combination of parameters. The dynamic test results show that the detection accuracy of lateral and heading deviation decreased significantly, as the speed of the vehicle increased, where the error of lateral deviation was below 10 cm and the error of heading deviation was less than 3°, and the coefficient of variation was less than 10%. The finding can provide a sound reference for the development of unmanned systems with low cost, high precision, easy operation in modern mechanized agriculture.