GPS-Aided Odometry Navigation for IARs: Comparison Between Loosely and Tightly Coupled Integrations Under Restricted Satellite Visibility Conditions

In modern times, the navigation of agricultural vehicles is predominantly reliant on Real Time Kinematics (RTK)-based Global Navigation Satellite Systems (GNSS), particularly the Global Positioning System (GPS). However, when precise positioning is required, as is often the case in Precision Agriculture (PA), this technology becomes financially unfeasible for small and medium-sized farmers, who are the majority of farmers in Brazil. On the other hand, odometry can serve as a cost-effective alternative (or complementary navigation technology) to GPS. Despite its well-known error accumulation over the traveled distance, odometry provides satisfactory short-term positioning solutions. Thus, the objective of this study is to evaluate the performance of a low-cost Single-Frequency (SF) GPS-aided odometry navigation solution, aiming at enhancing positioning accuracy for Intelligent Agricultural Robots (IARs). The primary contribution of this research involves extending previous research about two integration techniques commonly used in sensor fusion applications, namely the Loosely Coupled (LC) and Tightly Coupled (TC) topologies, in the specific scenario of GNSS signals becoming eventually blocked, i.e., when Satellite Visibility (SV) is gradually impaired. Experimental results are provided, which highlight the superiority of the TC solution, even when only 2 satellites are in view.