EMERGING TRENDS AND APPLICATIONS OF DRONE TECHNOLOGY IN AGRICULTURE AND PLANTATION MANAGEMENT

Abstract

This critical narrative review examines whether unmanned aerial vehicle (UAV) sensing, artificial intelligence and spray application improve management decisions and field outcomes in oil palm, paddy, rubber and tropical fruit, with particular attention to Malaysian conditions. Seven purposively selected primary studies were charted by crop, sensor, method, validation design and outcome, then graded against stated evidence criteria. Palm detection had the strongest image-level evidence, with F1-scores near or above 0.90 in two independent studies, although one model differed by eleven percentage points between plantations. Detection accuracy did not establish inventory accuracy. Paddy weed mapping and Ganoderma mapping were validated only within single sites, and lightly infected palms could not be separated from healthy palms. Two studies compared field outcomes with conventional practice. In maize, UAV-derived prescription maps yielded estimated herbicide savings of 14–39.2% without reducing crop biomass. In replanted oil palm, UAV spraying matched pre-planting mist-blower control and used 37% fewer working hours and 91% less water than knapsack spraying, but produced weaker selective grass control. Modelled costs favoured UAV spraying over knapsack spraying from approximately 3,000 ha under the study’s assumptions. Durian pollination was evaluated only in simulation. These findings concern a purposive evidence set rather than the whole literature. A sense–interpret–decide–act–verify framework is proposed to connect aerial observations with auditable outcomes. Priorities include cross-site validation, understorey weed detection, contract-specific costing and multi-season comparative trials. Practical adoption should therefore distinguish image classification performance from agronomic effectiveness and economic feasibility. Independent ground checks, explicit decision thresholds and comparisons with existing management are needed before promising results can support wider operational deployment. The review distinguishes predictive performance from management effectiveness and identifies the additional validation needed before UAV outputs can support routine operational decisions. Its findings should not be interpreted as a systematic estimate of the effectiveness of technology across the agricultural sector.