Smart(er) Agriculture: Robotics, Sensing, and Autonomy

The U.S. agricultural enterprise is very resource intensive and consumptive.  It uses 50% of U.S. land, 10% of the nation’s annual energy budget, 70-80% of fresh water, and more than 20 million tons per year of fertilizer, as well as other chemicals. Furthermore, aside from conservation land and repurposed land for food production (e.g., converting forestland to farmland), we are rapidly approaching the global limit for arable land.  Yet, the global demand for food will only increase as the planet is expected to reach nine billion inhabitants by 2050.  It is estimated that food production globally will need to double in the next 35 years to meet a growing population, and one that demands a higher-protein diet, mostly from animal sources.  While inter-decadal increases in agricultural productivity of 8-12% might allow the U.S. to achieve a doubling of food production in this timeframe, it is less certain whether this target can be attained globally.  This means that future agriculture will need to become dramatically more productive while reducing resource use, including land area.

Along the way to feeding, clothing, sheltering, and fueling more than nine billion people, there are some constraints that must be negotiated.  Primary among these are the uncertainty and variability inherent in changes to our global climate.  Climate change impacts are already being felt in agriculture, and will continue to exacerbate annual variability in food production, both in amount and location.  Second, much agricultural production is currently dependent on irrigation, either from annual snowpack or ground water, which are either unavailable or depleted.  Third, near population centers, highly productive farmland is being lost to urbanization.  Fourth, there is a growing awareness of, and commitment to, the importance of reducing agriculture’s environmental track record and its ecological footprint.  And, finally, a growing middle class globally demands a higher-protein diet and also expects positive health outcomes from these dietary changes.  Taken together, it’s clear that the road ahead will require significant changes in how agriculture operates.

When considering “smarter” agriculture, the following conceptualization applies, “develop new or improved engineered devices, products, or systems that sense, “reason,” and respond: just-in-time (and -place), unsupervised, dynamically, and precisely to improve the profitability, productivity, and/or efficiency of agriculture-related operations of all sizes, and benefit consumers and society.” Just-in-time and -place agriculture relies on the application of tightly coupled and highly dense information, sensing, and actuation spaces representative of the physical environment in time and space—throughout production, processing, handling, storage, transportation, and consumption.  These spaces should be available for interrogation, manipulation, or control by humans or other intelligent agents.  In addition to consuming large amounts of land, water, and petro-chemicals, the agricultural enterprise must also deal with the availability and cost of farm labor, which creates an economic disadvantage for many agricultural industries in the U.S. as they try to compete in the global marketplace.  Using robotics to eliminate unskilled, unsafe, and low-wage jobs will create new business opportunities, with higher-wage, technically demanding jobs, which can lead to more viable and resilient rural economies.  Smarter agriculture can help operations be more productive and efficient, and reduce their footprint in consuming resources and generating waste.  Taken together, these advanced technologies can address all four elements of sustainable agriculture: productivity, environmental quality, economic viability, and social acceptance (quality of life). 

Key R&D areas that are currently at the forefront of federal and private-sector investments in smarter agriculture are: (1) new or improved sensors, sensor systems, and sensor fusion (sensitivity, modalities, speed, repeatability, reliability); (2) individualized (or generalized) manipulators and end-effectors to facilitate human-level actuation; and (3) collaborative autonomy that will result from distributed intelligence and fault tolerance, enabling high-level task completion despite failure of one or more agents or temporary loss of human attention.  Of course, none of this can be realized without a skilled and motivated workforce.  Creating this new “agriculture” will require commitments by two- and four-year institutions of higher education, federal and state government policies and incentives, in-class and after-school programs, and private-sector partnerships.  Nineteenth-century agriculture is long forgotten and twentieth-century agriculture, while dramatically increasing food production, has lacked the sustainability that is required to support the global population in 2050.  Building a smarter agriculture is a necessary and inevitable component of the future trajectory of sustainable livelihoods.