According to the USDA, agriculture accounts for almost two percent of total energy consumption in the United States. Researchers at University of Colorado Boulder are developing sensors that will monitor soil, environment and crop conditions so that inputs like water and fertilizer may be precisely matched to crop needs. New Ag International Editor Janet Kanters speaks with Gregory L. Whiting, mechanical engineering associate professor with University of Colorado Boulder, and leader of the project.
New Ag International JUN/JUL 2020
Gregory L. Whiting
What do you aim to achieve with these crop sensors? That is, what will they do specifically? Our primary aim with these sensors is to be able to measure agricultural inputs directly in soil at high spatial resolution. We’ll start lower of course, but our goal is ultimately to create a sensor that would be economic to use on the scale of tens or even hundreds of sensors per hectare if needed. In
the short term, we’re focused on the measurement of soil nitrate concentration and soil moisture content, but could expand to other analytes of interest in the future. Ultimately the goal is to use these sensors to improve input use efficiency, crop yield and farm profitability while reducing energy demands for food, fuel and feed crops.
What will the project cost, and where is the money coming from? This project is funded by the Advanced Research Projects Agency - Energy (ARPA-E), which is part of the Department of Energy. ARPA-E has an impressive portfolio of projects related to agriculture, so it’s really great to be working with them, as we get some very valuable support and perspective in addition to the funding. Project funding from
ARPA-E is about US$1.6 million, and the project is planned to run for three years. We are currently about one-third of the way though the project.
What is the project called, and what is the goal? The project is called Precision Agriculture using Networks of Degradable Analytical Sensors, which gives the handy acronym PANDAS. The goal is to develop and demonstrate low-cost, low-profile, no-maintenance sensors for soil moisture and nitrogen status that are manufactured using printing techniques and biodegradable materials. Fabrication from biodegradable materials allows the sensors to simply be left in the field at the end of the season without any need for collection or maintenance.
The illustration depicts the vision for the developed sensor network. It shows a field with a number of sensor nodes distributed around the field (at a spacing on the order of 10s of meters). When probed from a remote RF source, the sensor nodes report information that can aid decision making for growers, including the primary measurements of soil moisture, soil nitrate concentration and soil temperature, as well as other information such as the sensor node number (which is related to the position of the measurement), the time of measurement and the strength of the received signal (which is related to the sensor position and the measurement reliability).
What do current sensors lack? Commercially available soil sensors can often require additional labour, and typically have costs that preclude their use at high spatial density. As such, spatiotemporal variability of the soil is often not completely captured. For example, nitrogen content in soils can vary over relatively small distances (on the range of 30 metres) and can change rapidly under certain environmental conditions. We want to be able to capture these changes to help assist growers in making input use decisions. Having large numbers of sensors and high data density also enables the use of sophisticated data analytics techniques that can provide useful insights and would be less applicable with lower data density.
How will the project and resulting new sensors solve this problem? Our hope is that through the use of additive manufacturing techniques to completely fabricate the devices, they can be kept relatively low-cost, with an eventual target of about $1 per sensor, allowing them to be used in large numbers in the field. Biodegradability of the sensors offers benefits that address existing issues as well. For example, since the sensors are only intended to be used for a single growing season, it frees up our options for materials and approaches for devices and packaging. Materials and devices
that would not have been suitable for long-term operation can be used here, and this allows us to simplify the design and lower the eventual cost. This has allowed us to develop new printed moisture and nitrate transducers that may not be suitable for long-term operation, but are suitable for short-term use. In addition, degradability means that the devices don’t have to be maintained or recovered from the field, meaning they can be used at high-spatial-density without causing excess waste.
Photo: Gregory L. Whiting
What is the power source for the sensors? How do they transmit information and over what distance? There are a number of options for power, depending on the specific application and environment. One option is to use passive RFID-like transmission where there is no power onboard the sensor itself, and the sensors receive their power and are interrogated from an incident RF-field supplied by a reader mounted on farm equipment (such as irrigation equipment or vehicles), or for example on drones that move around the field. This approach keeps devices simple, but ultimately limits temporal read-out frequency to the times that a reader is passing over the sensor. In many cases this approach is fine, but we are also looking at combinations of printed biodegradable electrochemical power sources that we are developing, with new RF communication approaches that allow for low-power, long-range (often up to one kilometre or more), low-bandwidth communication.
How would farmers benefit from the new sensors? I.e. in terms of helping growers make input decisions, what are the results from the prototypes? We're fairly early on in this project at the moment, so we're focused primarily on developing the printed sensors and other devices themselves. We still have a lot of work to do on the specifics of how data will be displayed to growers. We expect that growers would see, in real time, a map of the field with current and historical concentrations of moisture of nitrate, and could then act appropriately to optimize those inputs. This would be particularly enabled in settings where variable rate irrigation systems are present.
Discussions that we've had with growers has suggested these sensors would also be useful for validation of inputs, looking at the differences in soil concentrations just before and after application to ensure appropriate amounts are used in appropriate places.
How many sensors would be required per hectare, i.e. what sort of concentration do they require? I think this will depend a lot on the needs of the specific grower and environment, but our intention is to develop sensors that could be used economically up to the scale of 100s per hectare if needed. Of course, a smaller number could be used as well if desired.
The sensors are touted as being biodegradable. Why is that important, and how long before they degrade into the earth? Biodegradability and transient function give a number of benefits. It allows us to make devices that are relatively simple, and low-profile. Since they don't require long-term operation, packaging and additional power and electronic systems can be minimized. It also enables sensor modalities and materials to be used that might not be suitable for long term use due to drift or degradation, which ultimately allows us to make the devices using low-cost materials and manufacturing techniques like printing. In addition, biodegradability and engineered short-term lifetimes means these sensor nodes can be dispersed at high concentration in the field, and won't require collection or maintenance if there are problems, and won't litter the field with electronic waste. We think of time to degrade in two parts: first is the functional lifetime, which is the
amount of time for the sensor to become non-functional or at least no longer reliable; the second is a physical lifetime, which is the amount of time until the sensor node object itself completely degrades away. Our first target is for a functional lifetime of greater than about four months and a physical lifetime of around one year. We're working on approaches at the moment to control these lifetimes accurately without introducing issues of sensor drift and believe that ultimately the biodegradation time will be able to be tuned to particular needs and environments.
Who else is working with you on this project? In addition to ARPA-E, we are working on this project with Prof. Raj Khosla and his research group in the Department of Soil and Crop Sciences at Colorado State University, and Prof Ana Claudia Arias and her research group in the Department of Electrical Engineering and Computer Science at the University of California Berkeley.
Are you working on any other soil sensing projects? In addition to the described project, we are also currently working on other projects related to using printed electronic sensors for monitoring of soil properties. Two of these projects are part of a joint US/UK effort as part of the Signals in the Soil program. One of these projects (funded by the National Science Foundation on the U.S. side) is called Phytoelectronics and is led by my colleague in Electrical Engineering and Materials Science at CU, Robert McLeod. For this project, we are also collaborating with groups at the USDA and at the University of Cambridge. Here, we
are investigating the use of a type of device called an organic electrochemical transistor (OECT) that can be manufactured by printing and can be implanted directly into plant tissue in order to monitor analytes of interest within the plant (and therefore, indirectly, in soil). Our other project is, on the U.S. side, funded by the USDA and is led by Jason Neff, who is the Director of the Sustainability Innovation Laboratory at Colorado (SILC) at CU Boulder, and carried out in collaboration with groups at the Universities of Lancaster and Manchester. This project focuses on the use of large numbers of both conventional and novel sensors coupled with machine learning approaches to understand and monitor soil health in real-time. In general, and particularly after discussions with various stakeholders, we think there is a significant opportunity for
distributed printed electronics to address issues in agriculture and soil science, and our research group at the Boulder Experimental Manufacturing and Electronics (BEEM) Lab is putting a lot of effort into developing devices to be used in this area.
Whiting, foreground, has partnered with other groups on the biodegradable sensor project. Photo: Gregory L. Whiting
Airbus has launched AgNeo, its new integrated precision farming solution for commodity and permanent crops. Delivering in-season actionable information based on satellite imagery and premium agronomic analytics, AgNeo is expected to help customers save time, optimize inputs and increase yields.
According to a press release from Airbus, AgNeo leverages the company’s 30-year experience in the agriculture market and the long-standing collaboration with its customers to propose a wide range of crop analytics. Based on imagery-derived health maps from SPOT, Pléiades, Sentinel-2 and Landsat 8 data, they support a wide range of capabilities: from alerting features that direct agronomists to scout areas in their fields that require immediate attention, to the creation of management zones and variable rate application maps. AgNeo customers can also derive absolute agronomic indicators to use as input to their own models for building nitrogen recommendation or establishing water budgets.
AgNeo also provides enhanced analytics for high value permanent crops such as almonds, apples and grapes. Using 50cm resolution data from Pléiades, AgNeo creates a mask of all the tree crowns to remove noise from background vegetation and soil, improving the results and bringing analytics to the tree level. These analytics include a stand count inventory eliminating time-consuming and error-prone manual tree counts. The product will be able to utilize 30cm data from Airbus’ new Pléiades Neo satellites due to be launched later this year.
The AgNeo solution is available with a full user interface and will also be accessible via APIs.
Permanent Crop Analytics Report as provided by AgNeo.
An Israeli tech company says it has developed a machine that can replicate the crop pollination process normally performed by bees, seeking to address concerns over the insect’s dwindling numbers.
As reported by Reuters, Edete Precision Technologies says it is already operating its invention at an almond orchard in Tel Arad in the southern Negev desert region.
“We see a crisis in 15 years where we don’t have enough insects in the world to actually do pollination and most of our vitamins and fruits are gone,” Eylam Ran, CEO of Edete Precision Technologies for Agriculture, told the Reuters news agency.
Edete has developed an end-to-end artificial pollination service comprised of two steps that mirrors the action of honeybees – collecting and distributing pollen. In step one, Edete mechanically harvests flowers; separates the pollen from the flowers; stores the pollen for more than one year using its proprietary method, overcoming the problem of desynchronization of different cultivars’ blooms; and ensures fertilization and fruit setup by matching the best pollinizer with each commercial variety regardless of the timing of their bloom, guaranteeing, and even increasing, crop yield.
In step two, during blooming, the stored pollen is loaded into Edete's proprietary high-efficiency artificial pollinator which disperses dry pollen on the trees. The pollinator uses LIDAR sensing to algorithmically reach as near as required to each tree contour and uses electrostatic deposition onto the targeted flowers. Edete's mechanical pollinator units can operate day and night, rapidly and thoroughly covering any open flower in its range and are not limited by daylight or low temperatures conditions.
Edete has been working on a small-scale trial in several orchards in Israel and Australia and has agreements to do the same in the United States. The company hopes to scale up and be ready to sell its products on the market in 2023.
Edete’s proprietary high-efficiency artificial pollinator disperses dry pollen on the trees. Photo: Edete Precision Technologies
University of Nebraska-Lincoln engineers have developed a penetrometer prototype that, when thrust beneath the soil, can shine some visible and near-infrared light on the agriculturally and environmentally vital properties of the subsoil lying beneath topsoil.
By characterizing that subsoil on the fly and in the field, University of Nebraska–Lincoln engineers Yufeng Ge and Nuwan Wijewardane believe the prototype could emerge as a time- and cost-saving tool that informs precise irrigation and fertilizer application. And they envision it allowing more farmers to participate in a burgeoning, climate-friendly marketplace that incentivizes farmers to capture carbon in their soils.
That potential stems from a simple principle: all substances, including the organic matter and minerals present in soil, reflect light differently. More to the point, every substance reflects different wavelengths of light, including the visible and near-infrared regions of the electromagnetic spectrum.
Nuwan Wijewardane, postdoctoral researcher in biological systems engineering, prepares to hydraulically plunge a penetrometer prototype into soil. Photo: Greg Nathan, University of Nebraska-Lincoln
Knowing that, Ge, Wijewardane and their colleagues embedded the prototype with a broad-spectrum halogen light that pours through a quartz aperture, nestling it below a parabolic mirror and fiber-optic cable that collect whatever wavelengths bounce back.
Another device, connected to the penetrometer, then measures the intensity of about 2,100 different wavelengths across the visible and near-infrared spectra. The intensities of certain wavelengths in that spectral signature correlate with the presence of certain substances and types of soils. Carbon- and nitrogen-rich organic matter, for instance, contributes to darker soils that reflect relatively few visible wavelengths. Soils with less organic matter or lots of iron, by contrast, will often reflect yellows or reds.
After being hydraulically plunged several feet into a given patch of soil, the prototype can take spectral readings from every one-inch cross-section that resides between the prototype’s tip and the soil’s surface. The five- to eight-minute process eliminates the need to dig soil pits or extract soil cores, which are traditionally sent to the lab for costly analyses that can take weeks yet still examine far fewer cross-sections.
Ge said the prototype also represents an improvement over most portable soil-sensing technologies, which typically take readings no more than six-inches deep.
“That’s OK, because that topsoil is the most significant,” said Ge, associate professor of biological systems engineering. “But if you look at something like the corn production system, the roots go deep. For some properties, like water uptake or nutrient uptake by a crop, the surface level of soil is just part of the story. You really want to consider the entire root zone.”
The team, which includes Cristine Morgan of the Soil Health Institute, Jason Ackerson of Purdue University and Sarah Hetrick of Texas A&M University, wasn’t satisfied with estimating just the composition of subsoil. The researchers wanted their prototype to measure how tightly packed the soil is, too, as a means of discerning how well that soil might retain water and share it with crops. So, they included a force-measuring load cell near the prototype’s tip, along with an ultrasonic sensor that measures how deeply the prototype has delved, to estimate the soil’s density.
A readout of spectral signatures produced by the team's prototype, which collects the visible and near-infrared wavelengths that bounce back from soils. Photo: Greg Nathan, University of Nebraska-Lincoln
According to Ge, the texture really determines the water-holding capacity. If you have too sandy soil, it just moves through the root zone very quickly. “But if you have a too clay(-like) soil, it’s going to hold the water very tightly, and the roots can't fully extract it.”
With the prototype built, the researchers sought to compare its spectral signatures with a library of about 20,000 signatures the U.S. Department of Agriculture collected from soil samples throughout the country. Because the USDA also reported the actual concentrations of carbon and certain minerals in those samples, comparing the new signatures against the USDA’s would allow the team to better calibrate the model it was using to estimate concentrations in its own samples.
There was, naturally, just one problem. The USDA had collected its spectral signatures after drying its soil samples in the lab, meaning they contained none of the moisture that virtually all field samples do. And given that water interacts with light, the lab-drying seriously altered those signatures.
Fortunately, an existing algorithm helped the researchers minimize the statistical noise drummed up by the water, transforming their spectral signatures into a form that more closely approximated the USDA’s. To test its corrections, the team took readings from 11 total fields across Nebraska, Illinois, Iowa and South Dakota. As expected, the team found that the prototype’s estimates of carbon and nitrogen levels hewed closer to actual levels after applying the algorithm than they did before.
Guiding light While conceding that the prototype’s accuracy could stand to improve — and that he expects it to — Ge said that even the current version might aid farmers looking to use irrigation and fertilizer more strategically.
Most forms of precision agriculture, Ge said, involve dividing a field into a grid and sampling the soil from a certain number of its cells. The relative expense of lab-based measurements might limit that number, he noted, making in-the-field estimates of subsoil composition an appealing alternative on a hypothetical farm of 160 acres.
“Let's say you have the resources to go out and collect five very accurate measurements,” Ge said. “You take the average, and you get the standard deviation, and you think, ‘well, that’s the mean and variance for the sample of soils in that field.’
“I would disagree with that sort of mentality, because I would argue (that) five isn’t enough. No matter how carefully you place those five locations, you’re not going to capture the entire view of the field. My argument would be: You really have to do this many times. Even though your measurements may not be as accurate as a lab-based measurement, you can still get a really good estimate and potentially be more useful than the first scenario.”
Ge expressed just as much enthusiasm for a less-obvious but promising application of the prototype — carbon sequestration — that could ultimately help farmers diversify their revenue streams while addressing the leading cause of global warming. The world’s soils stash more than three times the amount of carbon that currently resides in the atmosphere, even as atmospheric carbon dioxide has reached levels unseen in the last three million years.
In 2019, agricultural technology startup Indigo Agriculture launched an initiative that offers money to farmers who engage in practices — planting cover crops, rotating crops, limiting tillage — that encourage soils to capture and store the carbon left over when crop roots and leaves begin decaying.
“That is additional income for the farm economy,” Ge said. “I don't think that we have figured everything out yet, because you have to verify that you have actually sequestered that carbon on the farm. And that's why measurement becomes really important.”
Though the team would first need to up the accuracy of its prototype’s carbon estimates, Ge said he sees the design as a cost-effective way of helping farmers verify the success of those sequestration efforts.
“I really think that this technology can get a foothold in that carbon market,” said Ge, who cited some initial interest from General Mills. “We have been talking about this for a long time. I'm hoping that, in a few years, this can expand to a much larger scale — to the point of providing additional income for a farm — and people can realize the importance of being able to manage soil in a way that it can store this carbon.”
In the meantime, Ge and Wijewardane are examining whether it’s possible to shrink the diameter of their prototype, which could make it more feasible for users to ditch the hydraulics and manually plunge the penetrometer into subsoils.
Several companies have joined forces on SlugBot, the world’s first robotic monitoring and bio-molluscicide treatment system for slugs – the precision spraying SlugBot robot prototype is anticipated to enter early field trials in summer 2021.
The initiative is led by Dr. Jenna Ross from UK Agri-Tech Innovation Centre, Crop Health and Protection (CHAP), in collaboration with the Small Robot Company (SRC), a British agritech start-up for sustainable farming; COSMONiO, a British artificial intelligence start-up; and Devon based farming enterprise, AV and N Lee.
Current slug control methods rely on traditional chemical pellets, containing either metaldehyde or iron (ferric) phosphate. Bio-molluscicides are also available in the form of the nematode products, however these are not economical for use in arable crops.
Phase one of the SlugBot project focuses on developing the artificial intelligence slug detection capability. Phase two will look to deliver slug detection using Small Robot Company’s ‘Tom’ robot, with mobile imaging of slugs and field-surface materials in glasshouse conditions anticipated by early 2021. Phase three will then focus on precision spraying, delivering an in-field slug treatment solution with SRC’s precision spraying ‘Dick’ crop-care robot prototype for autumn 2021.
Sumi Agro Europe Ltd. (SAE) and NIK group of companies have entered into a partnership agreement, aiming for expansion and implementation of precision agriculture in the Balkans (specifically Bulgaria and Romania) and broadly in Europe and CIS (Ukraine, Russia, Poland, Czech Republic, Slovakia, Hungary, France and Germany).
Synergy and innovation are expected by combining both company’s core competencies: NIK’s advanced experience and operational know-how in the precision agriculture space, and SAE’s wide-ranging business platforms and assets for distribution of agricultural related materials.
Sumi Agro is an official international representative of over 40 producers of plant protection products, fertilizers, biostimulants, and seeds for conventional and organic production. NIK represents a group of companies operating in Bulgaria and Romania with over 18 years of experience in developing and implementing precision agriculture technologies focused on offering solutions for farm automation and work optimization.
Yanmar R&D Europe (YRE) is in the midst of a two-year, four million euros ‘SMASH’ project being carried out in cooperation with 10 technology partners to develop a mobile agricultural “eco-system” to monitor, analyze and manage agricultural crops.
SMASH robot
‘SMASH’ stands for Smart Machine for Agricultural Solutions Hightech. The project consists of the development of a modular robotic platform that employs the latest information communications technology to examine crops and soils, analyze gathered information and provide clear, actionable information to farmers to support crop management.
One of Yanmar’s many roles was to develop control systems for the multipurpose robotic arm for mobile manipulation (including precision spraying), sensor integration for positioning technologies, and autonomous navigation and software development for the control of the system’s mobile base (in collaboration with other partners).
In its European research facility nestled in the hills above Florence, Italy, YRE’s modelling and control engineer Manuel Pencelli developed a prototype agro-bot that could be used to monitor and control crops, take soil samples for analysis and accurately target agricultural chemicals for precision application. He noted the project required many different areas of expertise.
“There have been many partners involved throughout,” said Pencelli. “We needed mechanical expertise for developing the structure of the vehicle, and many ‘communications’ experts because we have a lot of devices that need to ‘talk’ to each other. Our starting point was in fact a tracked vehicle that was originally built for moving along a beach and cleaning the shoreline.”
There are two working SMASH prototypes – one for grapevines and the other for spinach – to cover the two different types of crops that were originally slated for research. The former has already undergone significant testing at a vineyard farm in the Pisa province, where Pencelli has been instrumental in demonstrating the possibilities that this robotic eco-system could offer farmers.
“SMASH is not a single machine, but a series of different devices including a robot, base station, drones and field sensors that together provide vital information to help farmers. A farmer could program the task that he wants SMASH to carry out, and while he is involved in other activities, this machine could move autonomously, monitoring crops, detecting and treating diseases, and saving the farmer or his workers significant time out in the fields manually checking crops.”
SMASH consists of a mobile base, a robotic arm featuring manipulators and vision systems, a drone and an ancillary ground station. The system is designed to function across a range of precision agriculture technologies, offering specific insights on geomatics, robotics, data mining, machine learning, etc., while taking into account the environmental and social issues facing farmers.
For Pencelli, the possibilities for SMASH are endless. “In addition to all the functions that can be performed by the robotic arm, we also have some attachments that can be mounted on the back of the vehicle for mechanical weeding or working the soil as it moves. This work can be done simultaneously, together with the monitoring and detection.”
Yanmar’s expertise has been in the software development for the agro-bot and the integration and installation of all of the other parties’ components. It’s a complicated mass of electronics, with wires, sensors, cameras, GPS receivers and multiple electric motors competing for space.
“The sensor fusion was one of the most challenging aspects of this project,” said Pencelli. “Because we have a very particular environment within fields, where a number of variables can change, such as the infrastructure, soil, shape of the fields and even other workers moving around the agro-bot. So, the localization of the vehicle, improving the robustness of it and understanding its physical constraints were interesting – such as speed, steering angle, the positioning of and communication between the mounted on-board devices – all these aspects can affect the motion of the vehicle.”
YRE also joined forces with Florence University’s Agriculture Department in order to further advance research activities in the field. The university has significant experience in sustainable crop management, having recently completed the EU-funded Rhea project that looked at improving crop quality, health and safety for humans, and reducing production costs by using a fleet of small, heterogeneous robots – ground and aerial – equipped with advanced sensors, enhanced end-effectors and improved decision control algorithms.
