Algoritmos de posicionamento por ponto preciso em tempo real para veículos agrícolas conectados

Abstract

Global Navigation Satellite Systems (GNSS) play a critical role in Precision Agriculture (PA), which has revolutionized agricultural practices by enabling more efficient use of resources and improved decision-making. The successful application of GNSS in PA has offered significant benefits in terms of improved efficiency, productivity, and sustainability. A few decades after the first Global Navigation Satellite System (GNSS) was made available to civilians, it has already become the main navigation technology due to its global coverage, low cost, and versatility in different applications. Real-Time Precise Point Positioning (RT-PPP) is a technique that improves positioning accuracy by correcting common mode errors present in GNSS observables by means of products made available in real time by specialized agencies, such as the International GNSS Service (IGS) and the Faculty of Astronomical and Geophysical Sciences (FCAG) of the Argentine University of La Plata (UNLP). The Real-Time Service (RTS) of IGS provides various correction streams for RT-PPP deployment, whose availability is beyond 95% for the Global Positioning System (GPS) and 90% for the GLObal NAvigation Satellite System (GLONASS). To take full advantage of PPP and Real-Time Kinematic (RTK) and overcome the limitations of both techniques, PPP–RTK, which can be considered an extension of the PPP model, allows users to obtain absolute positioning using a single receiver, with greater accuracy and convergence speed by applying ambiguity resolution. The main objective of this work, which is the first in a series of studies and research, is the design, implementation, analysis, and validation of a prototype that performs RT-PPP, without however resolving the entire ambiguities (PPP-RTK), but focused on the use of single-frequency receivers (low cost) and only GPS pseudorange and Doppler observables. As the main contribution of this work, a system composed of two GPS RT-PPP software modules is developed, which have the functionality to collect observables and ephemeris data from the receiver (rover), collect the necessary corrections in real time, and apply the studied techniques in the processing of this data, the first using as processing strategy the Weighted and Iterated Least Squares (WILS) and the second, an Extended Kalman Filter (EKF). During epochs when the correction streams become momentarily unavailable or when communication problems/latency occur, the equipment user has no other option than to use outdated products, which may degrade position accuracy. One of the objectives of this work is to evaluate the impact of using outdated RT-PPP products on GPS positioning accuracy in Brazilian territory. As a main contribution, we show that the RT-PPP GPS positioning accuracy is not significantly degraded when outdated products up to 25 minutes are used, being able to comply with the Society of Automotive Engineers (SAE) J2945 specification, which stipulates a maximum horizontal position error of 1.5 meters, at 68% probability, aiming at Connected Autonomous Vehicle (CAV) applications. Results from experimental tests conducted in a stationary environment validate these findings. In addition to the communication latency/interruption problem in obtaining IGS products for corrections provided by RT-PPP, this work also investigates the existence of residual errors that remain in the GNSS observables, which need to be addressed to further improve positioning accuracy. Among these errors, the tropospheric error stands out, which is a Common Mode Error (CME) for which no RT-PPP products are currently provided, and also the multipath error, which is a Non-Common Mode Error (NCME) whose effect is associated with the reception of multiple GNSS signals reflected from the surrounding environment. In this work, we propose to estimate the multipath residual errors as first-order Gauss-Markov (GM) processes, which are augmented to the Extended Kalman Filter (EKF) state vector, after having their correlation times and driven noise Power Spectral Densities (PSDs) suitably identified. For the tropospheric error, we estimate and compensate for it via the University of New Brunswick 3 (UNB3) empirical model, as well as augmenting the EKF with one additional state to account for the residual wet tropospheric delay, which is modeled according to Niell’s mapping function. Therefore, the second objective of this work is to evaluate the effectiveness of the aforementioned techniques, and as a main contribution, we show that the use of the proposed residual error mitigation techniques, besides bringing increased complexity, does not yield significant accuracy improvements when compared to the standard single-frequency code-based RT-PPP solution. Results of experimental tests conducted in a dynamic environment validate these findings.