Evaluating the navigation performance of multi-information integration based on low-end inertial sensors for precision agriculture

The main objective of this research was to evaluate the navigation performance of multi-information integration based on a low-end inertial measurement unit (IMU) in precision agriculture by utilizing different auxiliary information (i.e., GNSS real-time kinematic (RTK), non-holonomic constraints (NHC) and dual antenna GNSS). A series of experiments with different operation scenes (e.g., open sky in wet and dry soils) were carried out for quantitative analysis. For the position drift error during a 20-s GNSS outage, the dual-antenna GNSS-assisted approach did not provide a reduction, and the NHC reduced the maximum error in the lateral and vertical directions by over 80% in the dry soil test, but only by approximately 30% in the wet soil test. The heading error with continuous GNSS assistance can be less than 0.03° and be reduced by more than 90% with the aid of dual-antenna GNSS. Additionally, the NHC reduced the heading error from 0.54° to 0.21° and from 0.34° to 0.25° in the dry and wet soil tests respectively. The results suggested that the multi-information integration improved the positioning and orientation reliability. Moreover, the lateral positioning accuracy required for the control of agriculture autonomous vehicles was achieved at approximately 3.0 mm with over a 60% accuracy improvement brought by the dual-antenna GNSS assistance. In contrast to the vulnerability of a single system, multi-information integration can provide comprehensive navigation information with higher reliability and lower costs. Hence, multi-information fusion will be a great opportunity for agriculture to meet the high-accuracy and high-reliability requirements of precision agriculture.