Agricultural Automation and Food Processing — Part 2 of 2
In Part 1 of this article we discussed how selective breeding ushered in a new era of crop science, and how machinery has evolved along side our knowledge of plant biology over the last century to change the face of agriculture as we know it. In this portion of the article I’ll discuss the future of hybridized plants, the new flow of data in the fields, and how the airspace above the fields is being put to productive use.
Hybridized Plants and Harvesting
When designing and running automation equipment, uniformity is king. If the material being handled is all the same size, shape, weight, texture, firmness, it’s much easier to pick, move, separate, inspect, etc. This remains true in agriculture; and dealing with living, growing biological shapes presents interesting challenges. With increasing automation taking place in agriculture, it makes sense that farmers might again look to selective breeding and GMO crops to solve some of these challenges in the modern age. But now, instead of just optimizing for raw yield, a farmer might choose a variety of crop that is known to grow more uniformly with fruit of a certain size; or stalks of a particular height that can more easily be harvested by a tractor with a mechanical picker; or maybe a variety that all comes ripe at the same time so a whole field can be harvested at one time. So in this modern age, many more parameters need to be optimized for than just yield or taste. A blueberry that tastes amazing, but is very delicate and must be hand-picked, will not be as desirable to a farmer as a blueberry with a tougher skin that can be machine-harvested and survive through the conveyor belts of the washing and packaging process. And in fact the price the end customer pays will directly reflect the ease of harvesting, packing, transport logistics, and spoilage. Of course, the intent is to maximize all of these parameters simultaneously, including taste and quality; but food engineering, as will all engineering, is truly the art of managing tradeoffs. For this reason, oddly-shaped heirloom tomatoes that must be hand-picked and hand-packed will always be more expensive.
If aspects of the plant can be made more homogenous, then the machine to handle that crop can be simpler, which often means less expensive and more reliable. But even if the plant cannot be made more homogenous, there are still ways to automate its handling; it’s just that the system complexity may have to scale up to handle the irregularities. More sensors can be added, or more degrees of motion can be built into the actuators. Machines can be built to handle even the most odd-ball of shapes. But this doesn’t come free, the complexity of such a machine often goes up by orders of magnitude, and with it the cost to purchase and maintain. In agriculture, where volume is high and prices are low, it’s very important to keep equipment costs down and robustness/reliability up. The equipment has to work or it won’t get used. In farming and food production there is often a narrow window of time to harvest, process, and deliver the product before it goes bad. So a large-scale farmer is unlikely to have time to futz around with a finicky piece of equipment. For this reason, the more reliable tool will generally win out in the field and in the packing house. Human hands and eyes are very easy to train, and then very reliable for short timespans of a couple of hours. Cameras and end effectors (technical name for grippers) are much harder to train to do the job right. So ideally the crop itself would not require complex end effectors to be developed to handle it. Better if the crop is known to be uniform enough that simple, robust metal rails, sieves, combs, or some other simple mechanism can be used to perform a function over and over and over again with no babysitting and no maintenance. For this reason, crop science, genetics, informatics, and machine automation all go hand-in-glove to streamline this process. So I believe that the advances in food production that we will see over the coming decade will be due in equal part from advances in machine automation as well as from efforts made in the life-sciences to engineer better plant hybrids.
The benefits of using hybridized plants coupled with customized machinery have already been brought to bear on certain crops such as blueberries, grapes, strawberries and tomatoes.
In one example of this, BBC Technologies, which was acquired by TOMRA foods for $67m in 2018, has built a blueberry sorter and packager whose artificial intelligence and image processing can “discern a stem hole from a bird peck and will reject punctured fruit that is likely to rot” and “is learning to detect dozens of different varieties and subtle seasonal differences, so fruit grown at the beginning of a season will be sorted and packed differently than fruit that comes at the end of it” (NY Times).
Mechanical grape harvesting has become the norm in California, where 80% of wine grapes are harvested by machine. One machine can harvest 15 to 20 tons of grapes per hour, work that would normally take 30 or more human pickers, at less than half the cost of picking by hand (CNBC).
Labor costs in the domestic strawberry industry are roughly $1b per year and it is becoming harder to find people to perform the backbreaking task of picking the berries from the knee-height bushes. Reducing the number of hands that touch strawberries on their way to the end consumer has cost and safety benefits also, while alleviating the problem of finding people to do the work. Strawberry picking robots can pick 8 acres of strawberry plants in one day and can replace the work of 30 human pickers (CNBC). Harvest Croo Robotics and Agrobot are two startups competing to serve this expanding market..
Other than corn and grain, tomato harvesting is perhaps the longest standing, most impactful example of harvesting automation that we have. Automated harvesting of processing tomatoes (those found in canned tomatoes and sauces, soups, juices, salsa and Ketchup) was introduced in the early 1960s, and within 5 years 99.9% of the industry was using mechanical harvesters (UC Davis). In this case, adoption of automation was also catalyzed by two things: 1) a labor crisis caused by the end of the Bracero Program in 1964, which brought seasonal farm workers to the US, largely from Mexico and 2) the introduction of the VF-145 tomato in 1961, a hybrid with tougher skin and an easy ability to be destemmed, both of which allowed for it to be handled by automatic harvesters. California’s production of tomatoes saw a more than five fold increase from 1.3 million tons to over 7 million tons from 1954 to 1975 (New York Times) a result of combining automation technologies with new farming techniques. In 2015, 14.3 million tons of processing tomatoes were harvested, almost entirely by machine (UC Davis).
Dealing with non-homogeneity
While many of the crops grown in the world today are bound for a processing facility to be crushed, shredded, milled, or otherwise processed, the plant produce we are most familiar with in our daily lives are the whole fruits and vegetables we hand-select at the grocery store or farmers market. Because of the need to maintain visual appeal, these crops are often carefully picked, and then sent to a packing house to be cleaned, sorted, and packed for shipment. These products pass through several different supply chains on their way to you, and many different hands. While these facilities do uphold the highest hygiene standards to protect the consumer, the workers often need to work shoulder to shoulder on the processing floor for long hours; which in this time of COVID-related restrictions has proven to be problematic for both workers and food processing companies.
Nowhere is this more on display today than in the meat processing industry, which has had a number of high profile closures due to corona-virus outbreaks among the workers. Robots may not be suited for every task needed in meat processing, and certainly not for every type of animal carcass, but there are companies such as Scott Technologies Unlimited that are attempting the effort. They use X-ray analysis on every single carcass, which is then conveyed into a process flow of intelligent robotic saw blades which use the X-ray data to make customized cuts on whole lamb carcasses, often more accurately and consistently than a human worker could provide. The future of automated meat processing is underway and may be ripe for investment and innovation; at least until the cultured-meat industry matures and begins to address more of the demand.
Data in the fields
Using data to optimize crop inputs/outputs and track their progress throughout the growth lifecycle is known as precision agriculture. According to MIT, “Precision agriculture augments a farmer’s decision-making ability by integrating advances in our understanding of crop growth, sensor technology and wireless connectivity” (MIT). Drones are a key component of precision agriculture since they provide farmers with valuable data that allows for greater efficiency with inputs like water and fertilizer, resulting in greater quality and yield. Drones also allow for constant monitoring of crops so that problems, such as improper irrigation or disease, can be quickly identified and their impact minimized.
In order to derive maximum benefits from precision agriculture, a suite of IoT devices is needed for monitoring soil, crops, water, and climate conditions, among other things. Agrilinks, a knowledge sharing platform provided by the US Agency for International Development sums up the necessity of IoT sensors and the value they provide:
“An essential component of precision agriculture is low-cost, connected sensors that can measure important parameters related to farming. Data from these sensors are transmitted via wireless networks, aggregated in web-based data storage, analyzed, packaged into a recommendation, and fed back to help farmers make decisions.” (Agrilinks)
One low hanging fruit use case for IoT is in crop irrigation. A prevalent method of watering crops is via Subsurface Drip Irrigation (SDI), which applies water to the crop root zone via underground tubes, instead of directly on the soil surface, where most weeds live. By coupling SDI with IoT sensors to monitor moisture levels, water can be autonomously applied directly to the plant roots at the optimal time.
There are myriad use cases for the IoT data that farms will generate and being able to tie data points like light levels, irrigation, air quality, soil conditions and weather together will create whole new ways for farms to optimize cultivation. As the graph below from Business Insider Intelligence shows, we are still in the early days of farms generating data from the IoT:
In 2014, connected farms were generating just 190,000 data points per day per farm; by 2050, Business Insider is estimating that they will generate 4.1 million data points per day from an increasing array of different sensor types. The number of installed agricultural IoT devices has more than doubled from 2015 (30 million) to 2020 (75 million). Harnessing all of this new data from farms will allow us to maximize yields efficiently and allow for crop production to keep pace with population growth.
Agricultural Aviation
Tractors and ground-based equipment will always be indispensable for farming. But when it comes to accessing large, dense fields covered in delicate cash-crops, sometimes the airspace above is the way to go. While most companies working on autonomous aviation for precision agriculture are working on use cases around optical surveillance and data collection, there are a few companies developing hardware solutions that involve physical interaction with the ground, such as applying chemicals, pollen, or seeds.
Crop dusting, or aerial application, is the process of spraying crops with substances like pesticides and fertilizer using aircraft. In the past, there had to be a human operating the aircraft, which was both dangerous for the pilot, who was being exposed to potentially harmful chemicals, and for the environment and surrounding population because of spray drift. This all changed in 2015, when
“the Federal Aviation Administration approved the Yamaha RMAX as the first drone weighing more than 55 pounds to carry tanks of fertilizers and pesticides in order to spray crops. Drones such as this are capable of spraying crops with far more precision than a traditional tractor. This helps reduce costs and potential pesticide exposure to workers who would have needed to spray those crops manually.” (Business Insider)
Drones can get closer to the actual crops they are spraying, leading to greater precision and less drift. While this could also be accomplished by spraying from a tractor, drones can do the job between 40 and 60 times faster, according to a separate Business Insider article.
Drones can also provide farmers with data that helps them determine where to plant seeds. Once a drone gathers this data, another drone can be used to plant the seeds in this location. MIT Technology Review outlines this process: “These systems shoot pods with seeds and plant nutrients into the soil, providing the plant all the nutrients necessary to sustain life.” And notes that some of these systems can achieve an uptake rate of 75% and decrease planting costs by 85%.
Of the drones that are working on applications that involve interaction with the ground, 95% are employing multi-rotor copter-style drone aircraft. The advantage of these is that they require almost no runway whatsoever (only requiring a small landing pad), are very precise and can handle obstacles well. However, these copter-style drones inherently have short range, very limited payload, and slow airspeed. For these reasons, they will never be suitable for longer range beyond-line-of-sight cargo delivery, and their area coverage rate for crop spray applications will be limited by airspeed and wind conditions.
Representing the other 5% of drones in this category are fixed wing aircraft. Fixed wing aircraft provide greater payload capacity, speed, and range than their copter-style counterparts. This is the area where we are most excited and have been choosing to invest. Our portfolio company Pyka is building autonomous planes designed to make agricultural chemical application safe, fast and precise. The company’s self-flying, fixed wing aircraft systems use autonomous flight controllers and onboard sensors to fly precise paths and land safely on rough service roads. The craft can update flight patterns in real time while detecting the right path and timing for spraying the target areas to compensate for changing wind conditions and reduce chemical drift into unwanted areas. This enables cultivators to spray agricultural chemicals and dust crops over hills and challenging terrain, while using less chemical per acre and decreasing accidental exposure to workers. Further still, these autonomous aircraft may one day be able to fly at night, using LIDAR and other non-imaging sensors to perform aerial application when the wind naturally dies down after sunset and when no fieldworkers are present; a task that no human pilot could safely do without illuminating the entire field.
Conclusion
We are very excited about investing in the future of agricultural technology. We see this as a field of massive growth and guaranteed continuous need as we move to feed the 9.8 billion people projected to be alive in 2050. Further, I’m personally passionate about freeing human minds from the drudgery of repetitive task-work such as picking and sorting on a processing line. Doing the same exact action over and over for hours on end is machine work, which they are perfectly suited for; freeing the human mind to dream, create, build, and fix. This is what we are perfectly suited for.
Prime Movers Lab invests in breakthrough scientific startups founded by Prime Movers, the inventors who transform billions of lives. We invest in seed-stage companies reinventing energy, transportation, infrastructure, manufacturing, human augmentation and computing
Sign up here if you are not already subscribed to our blog.
