Answers are grounded in this paper and cite their sources.
Auto-guidance systems for agricultural tractors have traditionally relied on geometric control and external positioning infrastructure, limiting robustness and adoption in resource-constrained production contexts. This systematic review analyzes 64 studies and introduces the Perception-Driven Autonomy Framework (PDAF), which relates perceptive capability to functional autonomy. The reviewed literature shows a progressive transition from GNSS/RTK-centered architectures toward vision, sensor fusion, and artificial intelligence. Classical controllers remain widely used, but increasingly operate within hybrid architectures supported by richer perceptive models. The results indicate that perception is a principal enabler of advanced autonomy: systems with explicit sensor fusion and semantic interpretation (PDAF P3) show the most consistent potential to operate under environmental variability and partially unstructured conditions, whereas predictive perception (P4) remains largely prospective. A substantial gap persists between reported performance and real-world operation. Although 86% of the studies report real-field validation, only 11 of 64 satisfy the strict operational-realism criteria applied in this review, including meaningful test scale, adverse conditions, and obstacle exposure. In addition, 59% depend on critical infrastructure, mainly RTK, restricting applicability in environments with limited connectivity or technical support. The literature also lacks homogeneous criteria for evaluating autonomy and operational feasibility. PDAF provides a common analytical language for relating perception, architecture, functional autonomy, and robustness. The findings suggest that future progress depends less on further improvements in geometric precision than on perceptually robust, scalable systems validated under representative field conditions.
No takes yet. Share an insight, caveat, or question.
Chávez et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: