Feeding the World with Intelligent Agriculture Solutions

July 28, 2021

In the past decade, the agricultural industry has been faced with increasing pressure to produce food with higher requirements on safety and quality while remaining affordable and available in order to feed the world’s growing population. In the years before the COVID-19 pandemic, the industry was under pressure to reduce its use of pesticides while managing a diminishing labor supply.

Figure 1: Employment in agriculture (%) VS World Population (billion) Sources: World Bank; Worldometer

Then COVID-19 occurred, and the unavailability of farm labor went from being a serious concern, to being a critical issue. Today, we are facing food shortages, which drives up prices while unused produce is left to rot on the fields.

The Season for Action

The time is ripe for a disruption to the global food production process. Agricultural technologies have reached a stage where they can provide practical solutions to resolve this major global challenge. Autonomous vehicles, be they ground or air, once were a dream. Now they are a reality, with sophisticated geographical positioning systems integrated with precision sensing and artificial intelligence.

Starting at the cultivar selection stage, deployment of hyperspectral vision systems with statistical machine learning has paved the way for an upcoming field known as digital phenotyping. Sensors, in other words, can identify plants based on the electromagnetic signature they create –– much like the way humans identify plants by their size, shape or structure.

This technology helps accelerate breeding programs that assist plant physiologists in identifying cultivars with disease –– and/or climate-resilient genes. Selected cultivars are then grown with smart sensing systems that record and monitor their growth. When used with digital twin technology, it is able to simulate how different business decisions affect the crop yield without actual physical “test and see” seasons.

The growing stage of the plants is the most labor-intensive part of the cultivation process. There are numerous manual tasks currently performed, such as de-leafing, spraying, pruning, weeding and layering. The summer season generally requires more labor, and yet this is when the working conditions are the harshest.

A farm worker manually de-leafs tomato plants in a hot and humid glasshouse. This process is crucial to ensure the best size and quantity of harvest. Imagine hundreds of thousands of plants to de-leaf every week during the summer season, and the shortage of labor means that thousands of plants are left unattended, leading to lower supply of food down the line.

Sensing, Working and Learning

The wide availability of new robotics and automation technologies can be deployed to perform these manual and monotonous farming tasks. Imagine autonomous vehicles that drive around the rows of farmland while pruning and weeding, simultaneously performing digital evaluations of plant growth.

Environmental and physical data of the farmland and produce can then be used, for example, to indicate which fertilizer works best or how much pruning needs to be done, or what irrigation cycle provides higher crop yield. Not only can this data be visualized on a computer screen for farm owners, the autonomous system can order components of the system to take action.

Figure 2: Types of technologies used at different stages of food production

The culmination point is at harvest, and there is little wiggle room with regards to the optimal picking time. The travel restrictions brought by COVID-19are severely disrupting the typical influx of back-packers and seasonal workers.

Smart technologies can work together to perform autonomous harvesting and hence help alleviate the impact of worker shortage.

Harvesting a Food Revolution

In the video above, we see this system in action. An intelligent vision system powered by machine learning algorithms first locates the fruit. Once a fruit is located, it sends coordinates to the arm, which deploys a bespoke end-effector to the given position. Another smart sensor unit sits on the end-effector to estimate fruit size and ripeness. If determined to be harvest-ready, the end-effector goes on to pick up the fruit, and then drops it off either into a crate or conveyor belt depending on the design of the harvester system. Many developments in control systems and robotics have managed to capture the essence of human motion for fruit-picking such as cutting, plucking, rolling, pinching, breaking and twisting. The beauty of this robot evolution is the consistency in robot decision-making which leads to crop consistency, higher product yield and better quality of produce.

Smart systems are now increasingly prevalent in the farmlands and through them, we can expect to achieve higher food security and sustainability, and ultimately Zero Hunger by 2030.

Figure 3: New agricultural technologies paving the way towards Zero Hunger

ABOUT OUR AUTHOR
Melanie Ooi is an IEEE Senior member and member of the IEEE Instrumentation & Measurement Society, 2016-17 Distinguished Lecturer for the IEEE Instrumentation and Measurement Society, and the 2014 Outstanding Young Engineer Award from the IEEE Instrumentation and Measurement Society.

Shen Hin Lim is an IEEE member and a member of the IEEE Robotics & Automation Society.

If you wanted to build an automated greenhouse that balanced water and energy needs, where would you start? This paper maps out the decision matrix for building a greenhouse powered by renewable energy.