Precision Agriculture UAS Swarm Technology

Precision Agriculture UAS Swarm Technology

Raymond J. DeMarco III

Embry Riddle Aeronautical University Worldwide

Abstract

Unmanned system technology continues to grow in leaps and bounds.  The decreased cost of sensor packages for visual cameras, infrared vision, and other types of computer vision along with miniaturization have created greater capabilities for unmanned vehicles.  Unmanned aerial, ground, and maritime surface vehicles can intelligently work together by the command and control of human operators.  A high level of autonomy is required for Unmanned Aircraft Systems (UASs) to execute and collect various data for precision agriculture.  A swarm UAS can safely and efficiently capture data that is significant for farmer analysis of animal grazing and crop cultivation.  Swarm technology is needed to increase the total coverage, quickness, and efficiency of aerial collect data. For the same mission, a swarm of UAVs versus a single UAV is more robust, flexible, and can cover a greater area. This paper discusses the need for unmanned swarm technology in agricultural applications and describes the need for the enhancement of precision agriculture.  This paper describes the sensor integration, human interface, and UAS architecture required for navigating, capturing, and processing data for analysis.  The advantages of the unmanned system over its manned counterpart will explain the appropriateness of swarm technology. Discussed here are the needs, benefits, limitations, and constraints of unmanned sensor selection, processing, and system architecture for precision agriculture UAS swarm technology.

Precision Agriculture UAS Swarm Technology

            At this day and age, an Unmanned Aerial Vehicle (UAV) has evolved far from its beginnings as a hobbyist activity.  Its fundamental purpose in the year 2017 is to gather important information.  A mining expedition in an untouched Yukon terrain will require visual 3D information about the topography and decide where to take mineral samples for a potential location to begin mining.  A UAV can gather information about deadly radiation by remote sensing in an area where humans cannot enter such as a nuclear disaster that occurred at Fukushima, Japan.  When it is required to obtain a detailed image of a specified area, whether it is by infrared or HD video, an UAV can provide the aerial sensors to create a 3D map, or orthogram for analysis. 

The possibility to deploy a cooperative swarm of UAVs programmed to gather crop and field information as an alternative to a single UAV will provide faster and more efficient data for analysis; thus, giving farmers a wealth of information very quickly to make decisions on crop production.

            The potential speed and coverage that is supplied in an aerial sensor network of a UAV swarm working together is far superior to a single UAV working alone. This paper discusses the possibility of a cooperative UAV swarm in an aerial sensor network; multiple UAVs consisting of actor and master nodes that are deployed at the operator’s command to follow a predetermined path and gather the requested data through various sensors and processing software. Data is collected and stored as a historical set to observe patterns as well as current field conditions.

Problem Statement

            The agricultural industry can benefit greatly from the use of cooperative UAVs in an aerial sensor network that will cover a greater area of data collection, provide faster, more accurate and up to date crop information.  Cutting edge technology such as UAV camera size, capability and weight is becoming more affordable for farmers; unmanned systems are low cost, high-tech, and are delivered fully assembled (McNeil, 2017). UAS have also been used in a variety of 3D-virtualization and aerial service applications such as disaster response and search and rescue; the time has come for this technology to flourish in precision agriculture industry.

Significance of the Problem

With thousands of acres of farmland, crop monitoring is one of farming’s greatest challenges.  Unmanned Aircraft Systems (UAS) equipped with specialized sensors for agriculture present a major advantage to farmers for soil and field analysis, crop spraying, crop monitoring, irrigation and health assessment.  Factors such as water quality and quantity, climate change, glyphosate-resistant weeds, and soil quantity to name a few are farmers’ increasingly complex concerns (McKinnon, 2016).  Agriculture UAS can produce precise 3-D images for soil analysis useful in planning seed planting patters and data for irrigation and nitrogen level management.  UAVs have been around for decades, but what is exciting in the UAS industry is the cost and size of the sensor equipment has decreased substantially. 

Modern technology has upgraded the banner of precision agriculture.  GPS equipped machinery such as tractors and UAVs navigate through field zones at the control of the farmer.  Unmanned aerial vehicles equipped with multispectral cameras sense and filter various light from frequencies within and beyond the visible light range; image data is collected and processed to create a 3D orthogram displaying critical information about crops such as ripeness and temperature that give knowledge to the farmer about crop health and livestock movement.

Figure 1: Field Images.  A: NGRDI, C:RGB, D: WDI (Hoffman, 2016)

The daily irrigation of crops requires billions of gallons of freshwater.  Much of the water is wasted on already ripe or dying plants (Kornei, 2017).  Since not all areas will require water, infrared and optical imaging from a UAV swarm can quickly gather and process that information for analysis, ultimately saving an absurd amount of water worldwide.  As an alternative to satellite and manned aircraft, unmanned platforms provide collection of remotely sensed land surface temperatures (LSTs) with high resolution thermal images from partially vegetated fields (Hoffmann, 2016). Researchers have investigated the Water Deficit Index (WDI) based on UAV captured imagery, and created accurate crop water stress maps.  From crop water stress maps researchers have measured the greenness and temperature of barley plants through color images: the normalized green-red difference index (NGRDI) indicates the surface greenness.  Remote sensing of temperature from thermal imaging reflects the water content, transpirations, and crop water stress (Hoffmann, 2016).  It has been proven that airborne observation with multispectral images can reliably differentiate between ripe and unripe crops.  This method of remote sensing in precision agriculture enables farmers to create water stress maps to pinpoint the soil that is in more need of irrigation, thus minimizing water use and pollutant runoff (Kornei, 2017).

The United Nations (UN) Food and Agriculture Organization predicts in the next 40 years food production will need to increase by 70 percent.  Other challenges include climate change, droughts, floods, decreases in resources and political shifts (Swan, 2012).  Noelle Swan, writing for seedstock.com has identified 5 major challenges to the future of agriculture and food security.

Resource Depletion

Since North American agriculture has become more industrialized, it relies heavily on fossil fuels and topsoil resources. John Gerber, food and farming professor at the University of Massachusetts says, “we have an industrial agricultural system that’s totally dependent on the assumption that cheap fossil fuels will last forever (Swan, 2012).”  The U.S. Geological Society has found the amount of ground water irrigated has tripled since the 1950s.  With some resiliency to heavy rainfall and periods of draught, this is an added challenge to the industry.  The EPA states that the leading cause of water pollution is from agricultural runoff.  Farmers need more efficient use of nitrogen and other additives.  Topsoil supply is becoming a greater concern; it is effected by “poor agricultural practices such as overcultivation, overgrazing and overuse of water” says Leo Horrigan at the Johns Hopkins School of Public Health (Swan, 2012).

Land Management

            Continually planting the same crop in one location will drain soil of nutritional value.  Rotating and interspersing cover crops prevents weed propagation and helps to with pest control since this method attracts predator insects.  The cover crops add organic material, increase water holding, and replenish the soil (Swam, 2012).  Regarding livestock management, overgrazing creates dry soil, destroys grasses, promotes weed growth and leads to desertification.  Good practice of land and cattle management holds potential for better food production.

Food Waste

            The UN states that roughly one-third of the world’s food is wasted doing production, handling, processing or distribution.  Supermarkets and restaurants purchase an excess amount of produce that drives up prices and increases risk of spoilage since consumer expectations expect produce to be without blemishes, and have the usual shape, size and color (Swan, 2012).

In the scope of this paper only production is discussed but these are factors that place a burden on the agriculture industry.

Demographic Changes

            There is an increasing disconnect between people and farming communities.  80% of United States citizens live in urban and suburban communities.  Increasing populations of metropolitan areas create a barrier between farming communities and consumers. Some areas known as “food deserts” have no fresh food for sale.  The USDA reports that 20 million Americans reside in these supposed food deserts. As urban areas grow farmers have pressure to sell their land since they cannot afford to not sell it.  This is an effect on urban-fringe farming.

Political Shift

            The North American Free Trade Agreement (NAFTA) has burdened agriculture in North America and blamed for the loss of 2 million farm jobs in Mexico due to the flood of U.S corn imports and other agribusiness subsidies.  The U.S. has had major shortages of immigrant workers due to tightened immigration policies while Americans have little desire or connection to farm labor. 

What the agriculture industry is lacking is complete knowledge, control, and efficiency of farm production from the technology that is available.  Unmanned systems are already having an impact in precision agriculture but the billion-dollar industry has more to reap.  Unmanned tractors perform autonomously through set waypoints.  When cows are ready for milking, they step onto a platform where robotic arms wash the teats, extract the milk, and disinfect the suction cups.

Advancements in agriculture processes and machinery are allowing dull, dirty, and dangerous tasks to become fewer.  A fixed wing or multi-rotor UAV is set to replace manned flights over crop fields.   It is possible that in the very near future farmers will be able to send out a swarm of UAVs to autonomously gather the information, and have it downloaded and processed for analysis.  The remainder of this paper suggests modern unmanned system technology should be fitted to the farmers needs so they can maximize production, increase efficiency, and reduce resource requirement and waste.

Alternative Actions

The use of UAVs in North America is governed by the FAA.  UAS operators who want to fly outside the requirements of the Special Rule for Model Aircraft 14 CFR part 107 will need to request a waiver or Certificate of Authorization (COA) (Request a Waiver/​Airspace, 2017).  Eventually UAVs claimed their place on the map for precision agriculture.  Today they are deployed to observe livestock grazing and crops for various types of data through different spectrums of light as mentioned previously.  The most common use of UAVs is for spot-checking to quickly gather data on perimeters or a certain field area.  A single UAV is quick to deploy and an acceptable way to gather information from the air but for multiple large fields it cannot accurately or timely map and capture field-level data with the speed or precision a cooperative swarm of Micro UAVs.  There is even a UAV that can spray nutrients or pesticide.  The quickness and accuracy of deployed UAS swarm to gather precision agriculture data will far surpass the limited capacity of an unaccompanied UAV.

Basic system architecture states that the first step is identifying the stakeholder needs.  Utilizing a UAV swarm in contrast to a single manned aircraft or UAV is operator friendly, fast to deploy and acquires information for a differential analysis. The second step is to identify how to maximize performance at the mission level by maximizing the probability of accomplishing all tasks effectively.  Complex algorithms, autonomous UAVs, and a Human-Swarm Interface (HSI) will be required for farmers to command and monitor operations remotely.  System architecture should tell us ‘what the system is’ or the tangible elements that the system is composed of (Qin, 2014).  It should also tell us ‘what the system does’ to provide value to the stakeholders (Qin, 2014).

Concept of Swarm Agriculture

Figure 2: Master - actor swarm structure (Say, 2016)

            UAVs have become an excellent sensor platform for the enhancement of precision agriculture. For faster and more efficient coverage of a large field, an autonomous wirelessly linked swarm of UAVs can be deployed to gather field-level information.  Through Radio Frequency (RF) communication the operator is at the helm of the Mission Control Center (MCC).  A UAV swarm is configured to gather data in a fully autonomous mode permitting aircraft to move in a structured formation.   UAVs are autonomous agents of a master/actor protocol similar to the illustration in figure 2.  Autonomy is necessary for fulfilling overall mission objectives (Rohde, 2011).  Communication algorithms facilitate the UAVs to be aware of each other’s position and movement while receiving commands and relaying data back to the MCC for analysis. 

Why Swarm?  The deployment of Micro UAVs (MUAVs) working cooperatively with a master UAV from a technical perspective provide benefits such as enhanced sensing and detection capabilities, operation in complex environments, higher spatial coverage, high mobility and economy of time, and the ability to carry out long time monotonous missions autonomously (Rohde, 2011).  The enhanced sensing and detection capability provides a more accurate 3D map data for analysis than a single UAS.  Terrain and crop information will be significantly more accurate with multiple cameras and angles.  Total coverage and efficiency of exploration is increased, visiting more cell grids but avoiding revisits of already explored areas (Rohde, 2011).

Sensor Integration

            Remote sensing in agriculture relies on the integration of visual and multispectral cameras mounted on an UAS.  It should be noted that UAS platforms collect “remotely sensed temperatures with higher spatial and temporal resolution than those collected by satellites and manned aircraft (Hoffman, 2016).”  Thermal and Red-Green-Blue (RGB) cameras can collect the necessary data to “determine the greenness and composite land surface temperatures (LST)” as well as the spectral Vegetation Index (VI) (Hoffman, 2016).  Maps

            The French-based manufacturer Parrot has produced a lightweight multi-spectral camera called the Sequoia at a price of $3500 (McNeil, 2017).  It has a 16 megapixel (MP) RGB sensor and four 1.2 MP sensors that collect near-infrared, red-edge, red, and green data (McNeil, 2017).  It weighs less than 4 ounces and contains internal storage plus SD card slot. 

            Plant stomata are pores on the epidermis of plant organs that help facilitate gas exchange.  Stomata are the openings for which gas is exchanged in photosynthesis and respiration (Franks, 2007).  Water vapor leaving the plant through the stomata is a process called transpiration.  Regarding crop irrigation, the stomata will be closed to reduce water loss through transpiration when water supply is insufficient.  This leads to stored plant energy and thus higher canopy temperatures (Guilioni, 2008).  Canopy temperature is positively related to crop water stress and negatively associated to soil moisture and transpiration.  Calculating the crop water stress helps farmers calculate irrigation needs.  UAV imagery will provide the accurate crop water stress maps at different stages of crop growth.  As discussed earlier, WDI maps determine accurate water stress values and provide interpretation of ripeness and crops that will not need large volumes of water (Hoffman, 2016)

Aerial Sensor Network

            Autonomous wirelessly connected UAVs are a viable sensor platform for precision agriculture. Agent based mobility algorithms make possible the fusion of measurements and data from different sensors more timely and accurately than any single UAV, satellite or manned aircraft.   Not yet brought into the agriculture industry, aerial sensor networks have gained significance in systematic environmental monitoring and are applicable in a variety of areas for 3D virtualization (Daniel, 2011). 

            Air to Air links.  Radio Frequency (RF) relaying is carried out through Air-to-Air (A2A) links in an aerial sensor network.  Relaying RF transmission of telemetry and sensor data enables real time calculation of crop field conditions.  Relaying within the network also compensates for losses of air-to-ground links.  An aerial mesh network transmits the telemetry and payload data and allows the UAS to swarm and cover a large area.  “Each node adapts its motion in order to keep the mesh network alive and prevent interruptions (Daniel, 2011).”

Figure 3:  3 different scenarios of sector divisions (Say, 2016)

            Within the master-actor protocol the main UAV receives, processes and stores all data packets sent from actor UAVs (Say, 2016).  The Received Signal Strength Indicator (RSSI) is significant for the master UAV to be aware of actor UAVs’ location and distance.  Lower RSSI indicated a UAV is further away while higher RSSI indicates a UAV is closer.  Within the swarm, UAVs will continuously broadcast their current GPS location and determine RSSI referring to neighboring actor UAVs (Daniel, 2011).   Figure 3 is an example of a network allocation illustration of N-1 through N-4 where N-4 has the lowest RSSI.  The actor UAVs are separated into 3, 6 or 8 sectors with the master UAV positioned at the core.  For alignment, the UAVs steer based on the direction of the master and for avoidance the actor UAVs steer to avoid collision maintaining an exact an appropriate distance from each other and the main UAV (Say, 2016). Actors UAVs are equipped with the selected sensor for its role and a multidirectional array antenna.  The master UAV acts as the remote data control center and gateway for data dissemination with larger data storage and antenna to send and receive data to and from the ground control center.

Human-Swarm Interface and Processing Software

            Teleoperating individual UAVs within a swarm protocol for this application would be extremely difficult and unnecessary.  The Human-Swarm Interface (HSI) is ideal when a wide area is to be monitored with autonomous aerial vehicles.  A human pilot will take control of the swarm of robots, autonomy will reduce the effort required of the operator by setting waypoints or selecting the field area for data to be collected.  In a Single Operator Multi UAV (SOMU) context robots are required to have a high level of autonomy to decrease the workload of the human operator (Recchiuto, 2016).  The HSI controls robot behavior by user commands.  A decentralized protocol exists where UAVs keep their position in formation while avoiding obstacles and inter-UAV collision while following inputs.

DroneDeploy. A 3D orthogram image is produced from the data gathering swarm.  A San Francisco based company called DroneDeploy has created an app that uses cloud based servers to process sensor data while the UAS is still in flight.  DroneDeploy has multiple cloud servers that receive and process sensor data as quickly as possible; as the images are being taken, they are uploaded to the cloud through Wi-Fi or cellular LTE service (Rees, 2015).  Therefore, data is uploaded for processing during flight and the orthogram map is available immediately for analysis. 

DroneDeploy currently is contracted with DJI but can process data for any type of aircraft if it has GPS Exchangeable Image File (EXIF) metadata (Rees, 2015).  GPS enables geotagging while EXIF provides date and time, shutter speed, exposure compensation, and white-balance depending on the camera type (Pic2map, 2017).

Having instantaneous data is especially advantageous for agriculture. Speed is important to farmers so they can make the necessary assessments and conclusions from crop information.  The major uses of DroneDeploy for agriculture are to keep track of livestock and grazing patterns, detect crop parasites and fungi, plan drainage and irrigation, and assess crop damage (DroneDeploy, 2017).

Limitations and Recommendations

            Limitations currently are mostly due lack of swarm technology that is not military based.  The cost of multispectral cameras is expensive.  Not every UAV in a swarm may be outfitted with multispectral but they can be staggered with actor UAVs for different roles.  The concept of a HSI is relatively new and will require constant upgrades.  Fixed-wing UAVs will be able to achieve similar data gathering results as Vertical Takeoff and Landing (VTOL) copters but hovering in one area will require constant circling where a quadcopter may hover with ease.  VTOL aircraft also provide user friendly operation by ease of launch and recovery.  Fixed-wing UAVs will fly for a longer duration and will cover more area but launch and recovery will be a manual process.

Development of a swarm algorithms specifically for data collection for precision agriculture or UASs will be required.  A human-machine interface or HSI should be developed with farmers’ recommendations for layout of controls and displays for feedback.  Processing software to analyze crop field data needs to be user friend and designed to allow farmers to create a historical library to compare previous data sets.  It should be customized to agriculture so that different levels of greenness reflect ripeness, areas requiring greater or lesser irrigation are highlighted, et cetera.  Areas requiring attention should be highlighted and put in queue to be evaluated.

Conclusion

            In this paper, I have discussed the perceived need for an UAS for the augmentation of precision agriculture by providing more and faster information for farmers.  This concept of a swarm UAS is faster, more accurate, and covers a greater area for data analysis than a satellite, and manned or single unmanned aircraft.  The requirement specification for UAS swarm is included in the system architecture, sensor selection, HSI and processing software.  The aerial sensor network is structured in a master/actor protocol collecting multispectral data, using RF for communication and data link between UAVs.  The swarm operates autonomously, receiving GPS waypoints to collect 3D map data.  Data is downloaded for analysis.  Farmers can make important decisions on crop health, irrigation requirements, and monitor cattle grazing.  Currently, a solo operating UAS can perform some of these tasks in smaller fields in selected area for spot monitoring.  The use of swarm UAS will greatly improve agriculture efficiency in the future.

References

Daniel, K., Rohde, S., Goddemeier, N., & Wietfeld, C. (2011). Cognitive agent mobility for aerial sensor networks. IEEE Sensors Journal, 11(11), 2671-2682. doi:10.1109/JSEN.2011.2159489

Franks, P. J., & Farquhar, G. D. (2007;2006;). The mechanical diversity of stomata and its significance in gas-exchange control. Plant Physiology, 143(1), 78-87. doi:10.1104/pp.106.089367

Guilioni, L., Jones, H. G., Leinonen, I., & Lhomme, J. P. (2008). On the relationships between stomatal resistance and leaf temperatures in thermography. Agricultural and Forest Meteorology, 148(11), 1908-1912. doi:10.1016/j.agrformet.2008.07.009

Hoffmann, H., Jensen, R., Thomsen, A., Nieto, H., Rasmussen, J., & Friborg, T. (2016). Crop water stress maps for entire growing seasons from visible and thermal UAV imagery. Biogeosciences Discussions, 1-30. http://dx.doi.org/10.5194/bg-2016-316

Kornei, K. (2017). How drones could become a farmer’s best friend. Science | AAAS. Retrieved 26 April 2017, from http://www.sciencemag.org/news/2017/01/how-drones-could-become-farmer-s-best-friend

McKinnon, T. (2016). Agricultural Drones: What Farmers Need to Know | Agribotix. Agribotix.com. Retrieved 26 April 2017, from http://agribotix.com/whitepapers/farmers-need-know-agricultural-drones/

McNeil, B. (2017). New UAV cameras smaller, lighter and more capable. Directions Magazine. Retrieved 14 May 2017, from http://www.directionsmag.com/entry/new-uav-cameras-smaller-lighter-and-more-capable/463778

Pic2Map Photo Location Viewer. (2017). Pic2map.com. Retrieved 20 April 2017, from https://www.pic2map.com/

Powerful Drone & UAV Mapping Software | DroneDeploy. (2017). Dronedeploy.com. Retrieved 20 April 2017, from https://www.dronedeploy.com/

Recchiuto, C. T., Sgorbissa, A., & Zaccaria, R. (2016). Visual feedback with multiple cameras in a UAVs Human–Swarm interface. Robotics and Autonomous Systems, 80, 43-54. doi:10.1016/j.robot.2016.03.006

Rees, E. v. (2015). Creating Aerial Drone Maps Fast. Emmeloord:

Request a Waiver/​Airspace Authorization – Small Unmanned Aircraft System (sUAS). (2017). Faa.gov. Retrieved 27 April 2017, from https://www.faa.gov/uas/request_waiver/

Say, S., Inata, H., & Shimamoto, S. (2016). A hybrid collision coordination-based multiple access scheme for super dense aerial sensor networks. Paper presented at the 1-6. doi:10.1109/WCNC.2016.7565148

Summary of Small Unmanned Aircraft Rule (PART 107). (2016). FAA News. Retrieved 14 May 2017, from https://www.faa.gov/uas/media/Part_107_Summary.pdf

Swan, N. (2012). Five Major Challenges Facing North American Agriculture. Seedstock.com. Retrieved 26 April 2017, from http://seedstock.com/2012/04/18/five-major-challenges-facing-north-american-agriculture/

Qin, D., Li, Z., Yang, F., Wang, W., & He, L. (2014). Modeling and optimization of multiple unmanned aerial vehicles system architecture alternatives. The Scientific World Journal, 2014, 189679. doi:10.1155/2014/189679