Archives of Current Research International

Digital and automated technologies have been promoted as a route to higher productivity, lower input use and improved resilience in both field agriculture and horticulture, yet the evidence underpinning these expectations remains uneven and, in several domains, contested. This review critically examines developments in smart agriculture and horticulture across six interdependent technological domains: proximal and remote sensing, machine-learning analytics, field and glasshouse robotics, controlled-environment production, postharvest quality assessment and chain integrity, and the integration architectures that connect them. Literature was identified through structured searching of open scholarly databases and indexes, supplemented by backward and forward citation tracking and by targeted retrieval from institutional sources, with critical appraisal focused on methodological adequacy, ecological validity and the correspondence between reported performance and operational conditions. Three findings dominate the synthesis. First, the technical performance of sensing and recognition systems is frequently established under conditions that differ materially from commercial practice, and the resulting laboratory-to-field generalisation deficit is the single most consistent methodological weakness across the literature. Second, horticultural systems impose constraints that arable-derived technologies do not resolve automatically, including three-dimensional canopy occlusion, delicate and heterogeneous produce, high crop value per unit area and dense, frequently repeated operations; the most convincing advances involve co-design of the crop system and the machine rather than the machine alone. Third, evidence on economic and environmental outcomes is considerably weaker than evidence on technical feasibility, with profitability strongly conditional on within-field variability, holding size, labour costs and energy prices, and with sustainability claims for controlled-environment production remaining sensitive to assumptions about electricity supply and system boundaries. Substantial gaps persist in multi-season field validation, transparent reporting of failure modes, distributional analysis of who benefits from digitalisation, and governance of agricultural data. Confidence in current conclusions is constrained by geographical concentration of evidence and by the rapid obsolescence of technology-specific findings.