改变游戏规则：农业中的数字双胞胎

DIGITAL TWINSIN AGRICULTURE Digital twins represent a significantopportunity throughout theagriculture supply chain to model, predict,and optimize crop yields and othermetrics —providing a worthy return oninvestment while positively impactingthe bottom line. Given the aggressivedemands on agriculture over the nextdecades for increased crop productivity,the application of technologies suchas IoT, big data, analytics, computervision, artificial intelligence, machine learning, and others provides aneffective set of tools that can be unitedto create digital twins of increasingrobustness and maturity over time. Digital twins represent perhaps thebest opportunity to meet modernagriculture challenges by enablingthe greatest optimizations of farmsand the associated supply chain. WHAT ARE DIGITAL TWINS ANDHOW THEY APPLY TO AGRICULTURE Although there are many challenges within the agriculture domain, a central problemthat must be solved is how toincrease crop yields by up to 70% per hectare by 2050tosupport food consumption for a projected population of approximately10 billion people.Achieving this significant yield increase will be extremely challenging. The amount of arableland is essentially constant, and yield increases must be achieved while nutrient lossesand greenhouse gas emissions must decrease dramatically to ensure sustainability. While it is a worthy pursuit to optimize fertilizer, moisture, and other variables as singulartrial-and-error solutions, the totality of a farm and its components represents a complexinter-dependent web of natural elements. It seems reasonable to expect that creatingrealistic models that produce validated predictions of component interactions wouldprovide measurable advantages to farmers, and other parts of the supply chain, from“farm to fork.” However, to sustain the projected human population going forward, it is necessary toholistically optimize all aspects of farming techniques radically and locally, even to theindividual plant level. Fortunately, modern technology advances provide realistic methodsto achieve these outcomes. EnterDigital Twins. Digital twins are virtual (computer-based) replicas of real-life assets created totest, predict, and understand the behavior of theirphysicalcounterparts. In farming, digital twins are virtualrepresentationsof real objects (sensors,measuring tools, actuators, weeding robots, IoT devices, etc.). The current stateof these objects can bereproduced,and future statessimulated,with highfidelity. These can be used to monitor, analyze, and createactionablescenariosfor farmers. Scenarios can be represented through graphical visualizations ofthe digital twin and data related to its elements. This empowers farmers toobserve, analyze, and remediate unexpected crop and equipment deviations. Clearly, pests and diseases can adversely affect crop production. In affected areas of afarm, visual and other information that human senses can’t directly observe can begathered and combined, leveraging various sensors and satellite data. This combined datacan be displayed on a mobile application that serves as a farmer’s digital counterpart(digital twin). Further, captured images of affected plants, provided with descriptions ofobserved problems, can be utilized to compare with a virtual or augmented reality replicaof that exact plant—a digital twin providing immediate diagnosis—enabling or enactingappropriate treatment plans. By identifying crop or equipment problems in advance, scheduling predictive maintenanceat the right time, and providing real-time prescriptive solutions, digital twins can ultimatelyenable higher-quality yields and a shorter time-to-market, from crop and livestockproduction to final product sale. Examples include: •Field monitoring systems report the overall crop state for better and informedmanagement decisions. Remote monitoring systems report the state of farmmachinery, like planters, tractors, sprayers, etc., providing early detection and diagnosisof equipment issues. •Monitoring systems enable measurement and management of livestock, physicallyand remotely, to optimize feed and activity schedules. Users typically interact with digital twins on laptops, tablets, or phones. These providehuman-friendly ways to access and visualize historical farm-related data and even predicttheir future states. Users can perform “what if” scenario analysis and even utilize the digitaltwin to manifest real-world actions on physical components, such as irrigation systems. Ideally, a digital twin presents a purpose-driven interface of appropriate detail, showingonly essential information, abstracting unnecessary complexity from the user. Mostimportantly, a well-constructed, data-driven, mature digital twin can enable a farmerto analyze and holistically drive optimal crop production, equipment maintenance, andmore thorough well-designed, user-friendly “command and control” style interfaces. CHARACTERISTICS, CONTEXTS, AND TYPES To relate