Robots on the field, the future of Agriculture
Sesame Financial Group
Sesame and its in-house research team offer investment solutions, in a strong spirit of partnership with wealth managers
New technologies in global agricultural production are crucial to overcome modern challenges such as food needs, sustainability and arable land availability. Through robotics and artificial intelligence, automation seems to be the main tool for this needed modernization.
Now facing many challenges of how we can meet the world’s food needs and preserve the environment, by taking into account that population is set to reach 9.7 billion by 2050 and arable land decreases by 100,000 hectares per year,?global agricultural production needs to double in the next 30 years(1).?
Over the last century, agriculture transformed from a labor-intensive industry towards mechanization and power-intensive production systems, while over the last 15 years agricultural industry has started to digitize. Through this transformation there has been a continuous labor outflow from agriculture, mainly from standardized tasks within production process.?Robots and artificial intelligence can now be used to conduct non-standardized tasks (e.g. fruit picking, selective weeding, crop sensing) previously allocated to human workers and at economically feasible costs.
A few technological advances in recent years have led to the development of agricultural robotics, such as machine vision technology to avoid hazards, identify crops, and even determine if they are ready to be harvested.??Satellite location systems like GPS have also made many of the advances in agricultural robotics possible. Robotic farm equipment often relies on GPS information to position and locate themselves on farms. Autonomous field plowing, seeding, or navigating tractors and equipment may use a combination of computer vision sensors and GPS to navigate and act as the driver in robotic plowing trucks. Another technology employed in agricultural robots is machine learning. Machine learning provides an advanced method of identifying collision paths that can help autonomous vehicles learn to adapt and avoid new or unexpected hazards in their paths.
1. Autonomous Tractors
In the case of autonomous farm equipment, machine vision and movement sensors work hand in hand to avoid obstacles while navigating the field. The robots create a virtual 3D model of the surface, and with the help of high-resolution cameras, they’re able to navigate freely. The tractor should imitate a human in its ability to observe spatial position and make decisions. The technology for the driverless tractors has been evolving since its beginning in the 1940s. There are now two main approaches to building and programming such tractors: full autonomy or supervised autonomy.
Full Autonomy
Currently, the majority of fully autonomous tractors navigate using lasers that bounce signals off several mobile transponders located around the field. Instead of drivers, the tractors have controllers, people that supervise it without being inside. These controllers can supervise multiple tractors on multiple fields from one location.
Another fully autonomous tractor technology involves using the native electrical system of the tractor to send commands. Using GPS positioning and radio feedback, automation software manages the vehicle’s path and control farming implements.
A company that we have historically included in our Actively Managed Certificate on Agriculture technologies is CNH Industrial (ticker CNHI US). The Italian-American company??unveiled an autonomous tractor through its Case IH brand in 2016, which farmers can supervise via a connected tablet.
Supervised Autonomy
Tractors that function with supervised autonomy (automated technology, but with a supervising operator present) use vehicle-to-vehicle technology and communication.?
2. Fruit Harvesters
No robotic harvesters are widely spread and used but some initiatives seem to be close to commercialization. In fact, historically, a limitation of fruit picking robots was their low accuracy at spotting fruit among foliage, however modern computer vision techniques, which combine different sensors and/or different visual cues, and which are robust to lighting variation, are overcoming this limitation.?
For fresh fruit such as citrus, apples, pears, peaches and cherries, pickers are fully automated yet designed to be easy to use and easy to maintain. Unlike shake-and-collect methods, robotic fresh fruit picker can precisely and gently pick 10 times more usable fruit compared with a laborer(2).?
Most grippers used in agriculture today are either two, three or four-fingered grippers. These fingers are in silicon and are able to pick up to 2kg depending on the shape, softness and friction of items to be handled. Many sensors are embedded in those fingers in order to supervise forces applied to the item.?
3. Agricultural Drones
Spraying drones
Due to the rapid development of computer vision and artificial intelligence, robotic sprayers feature novel intelligence systems that enable selective spraying, compared to conventional uniform spraying across the crop.
In countries with advanced agriculture, aerial spraying by drone completes the precision farming virtuous circle. This begins with remote crop scouting targeting treatment areas that are followed by applications on a pre-programmed route. And this, can not only be achieved remotely, but also truly autonomously.?At the same time drones improve application timeliness, reduce the need for skilled labor and cut hand-held sprayer operators’ exposure to harmful pesticides.
Drone use to apply spray treatments is already widespread in south-east Asia, with South Korea using drones for approximately 30% of their agriculture spraying(3). Drone sprayers are able to navigate very hard to reach areas, such as steep tea fields at high elevations. Drone sprayers delivery very fine spray applications that can be targeted to specific areas to maximize efficiency and save on chemical costs.
VoloDrone: Volocopter and John Deere team up??to shape the future of drone logistics
Surveillance drones
Drone imaging is capable of producing accurate image location to the millimeter. This means that after planting, areas with stand gaps can be spotted and replanted as needed, and disease or pest problems can be detected and treated for right away. Drone field monitoring is also being used to monitor the health of soil and field conditions. Drones can provide accurate field mapping including elevation information that allow growers to find any irregularities in the field. Having information on field elevation is useful in determining drainage patterns and wet/dry spots which allow for more efficient watering techniques. Some agricultural drone retailers and service providers also offer nitrogen level monitoring in soil using enhanced sensors. This allows for precise application of fertilizers, eliminating poor growing spots and improving soil health for years to come.
In the past, a company that we used to include in our AMCs was unmanned aerial vehicles (UAVs) manufacturer AeroVironment (ticker: AVAV US). The company developed a drone called Quantix, which featured autonomous flight capabilities and a hybrid design that allowed it to launch vertically and transition to horizontal flight. It was able to fly up to 400 acres per 45 minute flights, with a cruising speed of 40 mph. Quantix’s sensors captured high-resolution color RGB and multispectral NDVI imagery that provided farmers with instant, actionable intelligent insights on their field.
4. Automated Weeders
Weed control is a daily struggle for vegetable growers, particularly hand weeding. Hand weeding is a major expense for crops like lettuce, and accounts for about 50% of the weed control costs for growers(4).
Automation for weed control has two components – detection and actuation (how the weed is killed). Automatic detection most commonly uses two-dimensional image processing to differentiate the plant from soil by color or light reflectance to detect the crop from the weeds by size differences and crop row pattern.
Several things still need to happen to overcome challenges in weed automation:
5. Sorting Machines
In recent years, application of image processing and machine vision has increased in the fruit industries. By using these techniques, we can now sort and grade the fruits more easily. These sorting machines often use multimodal image processing, meaning that images from different canals are being analyzed (laser, video, high-resolution image…) for segmentation and feature extraction (external quality, internal quality, weight, size, color and curvature).?
Examples of image processing for quality assessment
6. Farming Exoskeletons
Recent advancements in exoskeleton technologies offer new approaches with the potential to enhance performance and reduce MSD (Musculoskeletal disorders) risks in agriculture. Lifting and carrying heavy loads, operating hand tools, and climbing equipment were highlighted as candidate tasks that were most likely to benefit from exoskeleton adoption.?
However, even if the advantages of exoskeletons are now proved by many studies, there is still a need for commercial offer toward farmers and farm workers.
7. Automated Pruning
Automated pruning includes two systems: a robotic one that cuts the branch and a software that analyzes which branch to cut. The main challenge in this domain is the algorithmic part, not the robotic one. In fact, those systems use regular anthropomorphic robots with a cutting end-effector. The algorithmic method used to determine where to prune is way more difficult and still remains quite a challenge for scientists.
No technology is mature yet and lots of work is still to do in AI/machine learning in its domain, and specifically on the plant’s model.?However, even if fully automated trimming systems are currently being used, robotic pruning is a major sector of modern agriculture and is bound to grow in coming years.?
8. Milking Robots
Voluntary milking system (VMS) allows the cow to decide her own milking time and interval, rather than being milked as part of a group at set milking times.?
The innovative core of the automatic milking system (AMS) is the robotic manipulator in the milking unit. This robotic arm automates the tasks of teat cleaning and milking attachment and removes the final elements of manual labor from the milking process. Careful design of the robot arm and associated sensors and controls allows robust unsupervised performance, such that the farmer is only required to attend the cows for condition inspection and when a cow has not attended for milking.
What about human-robot interaction?
Human-robot collaboration
Robots closely collaborating with humans (so-called cobots) are delivering real step changes in many industrial sectors, and are anticipated to be vital to automation in agriculture. Use cases range from farm in-field logistics (transportation), where efficient and safe hand-over of goods and produce needs to be facilitated, to applications enhancing animal and crop welfare by means of integrated monitoring and intervention delivery.
As the technology matures, and in particular for safety-critical tasks, various levels of shared autonomy will be seen, where the human operator guides the high-level execution, while the robotic system performs the required sensorimotor coordination on the ground. The fan-out, or number of robots a human can control simultaneously, will help drive the mixture of human supervisors and robot agents in such a paradigm.
Safe human-robot interaction?
Human supervision will be a vital safety factor for most agri-robotic systems for the foreseeable future, while the technology develops towards higher levels of autonomy. Robotic systems will also be learning and adapting to task and farm-specific constraints. Human and robot collaborators will therefore likely be mutually adapting to each other, in order to maximize performance. Approaches to safe physical Human Robot Interaction (pHRI) include supervisory systems to monitor the interaction and adjust the behaviour of the robot if an unsafe situation is identified. This typically involves slowing, or completely stopping the robot, to prevent accidents. However, this approach can significantly reduce productivity as the robot is not working to its full potential. Current research aims to improve on this approach by allowing robots to identify and predict unsafe situations, and then to adapt and adjust their operation to continue the task in a manner that allows both productivity and safety to be maintained. A further approach to ensuring safe pHRI is to design robot systems which are inherently safe, meaning that if collisions occur between human and machine, injury will not result. This requires a change away from heavy, rigid and high inertia robots to systems which are more akin to biological creatures. Again, this is a challenge that the new field of soft robotics may be able to address.
References: (1) Robotic Industries Association (2017),?Robotics in Agriculture: Types and Applications - (2) Robotic Business Review,?FFrobotics & Fully automated fresh fruit harvester - (3) Future Farming (2020),?Drone spraying takes off as regulations relax worldwide - (4) University of California, Davis - Department of Plant Sciences (2019),?Automated Weeders are Attracting More Interest