Why Agriculture Will Automate Before On-Highway

By  PETER BLEDAY

Danfoss’ top autonomy executive says automation will help overcome labor and technological challenges that would otherwise leave billions of dollars’ worth of crops rotting in fields.

A World Wildlife Fund report estimates that global food waste on farms amounts to 1.2 billion tonnes per year – about 15.3% of the world’s annual food production. Total value of that loss? About $370 billion. Crops left unharvested due to lack of labor are not included in these figures. While aggregate numbers for the U.S. are hard to come by, last year the British farming union NFU estimated that overall food loss in the U.K. due to workforce shortages was £60 million – and that was just the first half of the year.

Farm-stage crop loss can be attributed to a number of factors: market conditions, poor technology or infrastructure, agricultural practices, pests and disease, natural disasters and weather, and labor availability and cost. The impact of the problem can already be seen today in price inflation at supermarkets. While the issue may seem transient, others – such as food supply and climate change – are not. Reducing food waste, as the WWF report states, will play a significant role in improving global food security and reducing greenhouse gas emissions.

While technological advancements alone will not solve this challenge, solutions such as autonomous machinery can go a long way in overcoming the human and technology factors of farm-stage food waste. This is why agricultural vehicles will progress down the road to autonomy faster than on-highway vehicles.

Three factors impacting on- and off-highway autonomy
The mainstream media is highly focused on autonomous passenger cars. From self-driving capabilities to robo-taxis, stories abound about the coming revolution in vehicle autonomy. But in on-highway, there’s really not a need for these capabilities, aside from perhaps long-haul trucking. Passenger cars are utility vehicles; they move people from one place to another. Automating a car does not remove the need for a human to be in it. And there’s not much value in replacing human taxi drivers, which are currently plentiful.

This contrasts greatly with agriculture, where there is not just a shortage of workers, but a shortage of experienced workers. According to the USDA, the number of H-2A temporary farm worker positions requested and approved increased more than seven times over the past 17 years, from about 48,000 positions in 2005 to about 371,000 in 2022 – a clear indicator of the scarcity of farm labor. These laborers don’t stay employed at farms year-round; they arrive for the harvest then move on. The following cycle tends to bring in new workers, so the experience doesn’t remain at farms.

How is it possible to address food waste at the same time as the labor challenge? The answer lies in improving machine and operator productivity, a key benefit of autonomy. This is critical for crops that require hand-picking, which include all 10 of the country’s most popular fruits and seven of the top 10 vegetables. Autonomy enables workers to focus on the critical task of picking while machines handle processes such as collection and transportation of the harvest through the field or orchard.

Such technology has already been commercialized. Vinergy, a California-based startup, developed an autonomous cart that eliminates the need for workers to push wheelbarrows full of grapes down long rows to a central collection point. Instead, crews are freed to pick the grapes, lightening the load for seasonal employees while driving up productivity by as much as 30%.

This compelling market need is the first reason agriculture will automate before on-highway. The second is lower technological complexity. In agriculture, the speed at which vehicles travel is much slower than in on-highway, allowing OEMs to leverage existing technologies for safety. The environment also presents fewer variables. In a field, there are no crosswalks or cyclists and far fewer vehicles; the environment is more defined. It does not mean autonomy is easy, but it is certainly easier.

The third reason is greater adoption. For motorists, the value proposition of autonomous functionality tends to be low. They may also perceive the technology to be high risk. Self-driving capability is not typically a requirement for consumers in the market for a new vehicle, at least not today. Conversely, farmers are actively seeking out features that make machines more productive and precise because it drives faster return on investment. For a machine in operation eight to 16 hours or more each day, significant payback is available in a matter of years.

The regulatory environment also comes into play. With on-highway vehicles, the individual regulations of cities, states and countries will force the autonomy gap to widen, whereas some agriculture standards already exist as the sector moves toward a defined set of regulations.

How agriculture will automate
Discussions of autonomy in mainstream media tend to focus on self-driving capabilities. The reality is more nuanced than that. In agriculture, as in other industries, there are five defined levels of autonomy (similar to but different than the recognized SAE levels for on-road vehicles):

• Level 1: Driver still controls most functions. Specific functions, such as steering, can be undertaken by the machine upon the operator’s direction.
• Level 2: In certain circumstances, the machine can steer, accelerate and brake. The operator is still directing and monitoring the action of the vehicle. Row following in a sprayer is an example.
• Level 3: More complex actions are controlled and executed by the machine. In the right conditions, a machine can manage most driving aspects, including monitoring the environment. For example, in addition to row following, a sprayer can turn and find the next row without operator intervention.
• Level 4: Machine can operate without human input or oversight and multiple autonomous machines can work together on one site, but supervision is required for edge cases. This is known as supervised autonomy.
• Level 5: Machine can operate without human input or oversight and be controlled off-site. This is known as unsupervised autonomy.

Most autonomous-machine development efforts in agriculture currently focus on Level 3; it’s expected that numerous Level 3 vehicles will emerge in the near term. There will even be Level 4 vehicles in agriculture before on-highway. By 2027, agriculture likely will have a well-advanced autonomy offering.

Autonomous machines improve safety, boost productivity and enable greater precision. In agriculture, this leads to greater yields and less food loss. In terms of safety, there are several applications in which the operator focuses on what the machine’s implement is doing while auto-guidance drives the vehicle, reducing his or her situational awareness. If a person or animal wanders into the machine’s path, the machine can warn the operator, navigate around the object or stop the machine automatically.

As for precision and productivity, a range of operations associated with planting, spraying and harvesting can be automated. The result is increased speed and accuracy, regardless of the operator’s experience. This reduces the impact of labor shortages that lead to food loss. Autonomy also enables easier navigation in tight rows or alleys, which prevents damage due to collisions and extends the life of equipment.

Reliability and repeatability are further benefits of autonomy. Cattle feeders, for example, require precise operation to ensure the feed is mixed and dispensed properly, with cattle fed the right amount at the right time. An autonomous vehicle can dispense food evenly and at the correct distance from the cattle every time.

Engineering challenges
Despite the huge potential, automating a machine is complex, with considerations for software and hardware. Autonomy requires perception sensors such as lidar and radar, global positioning, high-power processors and other technologies that may be new to design engineers. These technologies are not plug-and-play, and the sensors are not one-size-fits-all. Each application and environment has unique needs that demand different products, and there’s a significant amount of engineering work to integrate these devices and services into a system.

With their in-house engineering teams, including software expertise, large OEMs have the advantage. Many small and mid-sized OEMs, particularly those developing machines for specialty agriculture where the labor challenge is often the greatest, don’t have those same in-house software engineering resources.

Rather than spending the time and expense to develop this expertise, OEMs can look to suppliers and integrators that have done the work for them. There are solutions on the market now, such as Danfoss Autonomy, that tie together hardware, software and engineering services to help OEMs develop autonomous machinery. Such companies will often work with OEMs from concept to production, supporting full vehicle development.