Robots in Agriculture This is Parkway North High School's (team #07-0183) entry for the 2007 National Research and Design Website Challenge. You are viewing the printer-friendly version of our entry.

Current state of Robotics in Agriculture      In today's society, technology has made huge breakthroughs where our life is made easier. One place most people do not think about in the category is agriculture. Agriculture today has several advancements that help farmers obtain cheaper and better food. Furthermore, in the near future, new products will make their work even more productive.
     In wineries around the world, farmers prefer to use the best grapes to make their wine the best it can possibly be. In order to acquire these grapes, farmers in the past had to examine the entire harvest to weed out the unworthy. Today, Fresno State students and faculty are working to create "a sophisticated wine grape harvesting system that uses GPS, GIS, near-infrared sensors and automated machinery to harvest only the most desirable fruit in a vineyard" [13]. This is a significant step forward in the field, because it reduces the need for human inspection in order to ascertain quality.
     A popular name in the agricultural robot industry is 'AgBot’. AgBot is a multipurpose robot for use in agriculture that has a variety of possible uses. One of these uses is weeding. The AgBot uses image recognition software to recognize different leaf patterns and base the type of plant off of that. When AgBot finds a weed, it cuts it down, and applies a specific amount and type of herbicide to the root. Although still in prototype phases, it shows great promises in the near future, as it uses a small fraction of the herbicide used today.
     While AgBot may appear simple, it is not. It is equipped with two cameras that measure depth perception and includes a version of Windows XP onboard, an 80 gigabyte hard drive, and a wireless connection to the internet. The amount of information that can be collected is very large, so the possibilities are endless.
     Armed with these and many other high tech robots, a slightly delayed robotic agriculture revolution is underway, helping farmers do their jobs more efficiently and helping feed more of our population.

Saving the World By Design Beginning problem: The human population has passed six billion and exhibits "a J-shaped growth curve, and is accelerating" [1]. More people require more food. However "Healthy food needs a healthy eco-system balance within an unpolluted air, water and soil system" [2]. Current industrial agriculture practices, dependent on massive chemical use and intensive land conversion, do not balance output with sustainability.

To feed a growing world, there are two options:
1. Divert more arable land to agriculture.
2. Increase the efficiency of current agricultural land.

Out of earth's 57 million square miles of land, only about 12 million square miles of land are arable1. In other words, there is a limited amount of arable land. Much of this is not currently used for agriculture, but conversion would have considerable cost to the environment as forests, grasslands, and other natural ecosystems would have to be destroyed to make way for agriculture. Affecting natural ecosystems disrupts cycles of nutrients, water, and energy, upon which agriculture itself depends. Therefore, the second option is more viable than the first.

The problem thus becomes: how can the efficiency of current agriculture be increased in a sustainable way?

Since the beginnings of agriculture, farmers have devised new ways of increasing the productivity of the land with better farming practices, equipment, and even engineering the plants themselves. For much of history, humans farmed on small scales with a variety of crops. Because of the small scale, agriculture largely depended on natural cycling of nutrients and energy with the surrounding ecosystems [3]. Although yields were modest, they were stable. However, with the advent of modern technology in farming, namely machines, fertilizers, and pesticides, the past farming practices were replaced by the more efficient modern practices. Unfortunately, the modern farm gains its efficiency and productivity at the cost of stability and environmental sustainability.

Modern agricultural practices favor large, mechanized farms with specialized production and crop monocultures [3]. A monoculture is a large area dominated by single species with a homogenous genotype. The cost to stability largely results from the specialization of farms in vast monocultures. Large farms specialize because each crop requires a different set of machines, so it is more economical to have one set of machines and one type of crop. The workings of a combine could dizzy most people. However, if an insect or infection affects one plant in a monoculture, it can affect the entire monoculture. Only the applications of vast amounts of pesticides over the entire monoculture keep the monoculture stable. Different species give and take different amounts of nutrients. A monoculture, with its single specie, will drain certain nutrients from the soil without the input of nutrients from other species. Massive use of fertilizer has compensated for this with negative impacts on surrounding, particularly aquatic, ecosystems. A natural community has a closed cycle of nutrients while a monoculture has an open, wasteful cycle of nutrients.

The Irish Potato Famine of the mid 1800s is a prime example of the instability of monoculture agriculture. At first, the introduction of the potato to Ireland proved a boon for the peasant farmer. Potatoes are cheap, easy to grow, ideally suited to the climate, and rich in essential nutrients. Potatoes were such an efficient crop that the majority of Irish agriculture was devoted to its cultivation. However, when the fungus phytophthora infestans arrived in Ireland, it decimated the entire crop and millions of people starved [4]. Hence, by example, diversification of crops is essential to the continued productivity of agriculture. However it is difficult to diversify the land due to the high degree of specialty of current farm machinery. With the problems of stability and the secondary effects of maintaining stability in a monoculture, we've decided to focus on solving the problem of turning monoculture into polyculture using robotics. Polyculture will not just stabilize agricultural production but will also increase production. A recent study, published in Science found "Biofuels derived from low-input high-diversity (LIHD) mixtures of native grassland perennials can provide more usable energy, greater greenhouse gas reductions, and less agrichemical pollution per hectare than can corn grain ethanol or soybean biodiesel. High-diversity grasslands had increasingly higher bioenergy yields that were 238% greater than monoculture yields after a decade" [5]. If low maintenance and high yields can be achieved with diversity in the biofuel area of agriculture, why not in the rest of agriculture?

Saving the World By Design: Solution
Part 1: The Field While robotics in industry dramatically increases efficiency by automating every process, robotics could affect agriculture by allowing for the complete redesign of agricultural practices. Why conform robots to current practices? Why not conform the current practices to the potential capabilities of robots? This has happened in the past with the mechanization. As explicated in the problem, machines do one task, and they do it very well, thus crops were converted to one type. Robots, with their complete automation and artificial intelligence, can do many things very well. This means robots can handle diversity and, at the same time, improve efficiency and productivity.

We begin our design of the robot with the design of the field. Instead of planting one species of plant over thousands of acres, our field will include multiple plant species, a polyculture. This biodiversity, in itself, will greatly reduce the need for pesticides and fertilizers. Historically, pests were controlled temporally by rotating crops, which disrupted the life cycle of the pests [3]. This was replaced with the monoculture and its chemical control. In this design, pests are controlled spatially and temporally by blocking their spread throughout the field with different species and varying the composition of the field over time. Hence, pesticides will no longer play the primary role in pest control. The biodiversity of a natural community results in a closed nutrient cycle. By mimicking the nutrient cycling of a diverse natural community, the use of fertilizer will be reduced. A minimal amount of fertilizer will have to be used to compensate for the removal of nutrients through harvesting. This fertilizing will be done by the robots and described later.

In addition to minimal amounts of fertilizing, the field will be equipped to irrigate in an intelligent way. The three major current methods of irrigation are surface irrigation, sprinkler irrigation, and microirrigation [6]. Which method is used will depend on the local context. However, irrigation poses the problem of "when" it is necessary to irrigate and "how much" to irrigate. The information to answer such questions is obtained in the following part of the solution.

Saving the World By Design: Solution
Part 2: The Sensor Network Successful robot autonomy on a farm requires retrieval of field data, processing the data, and correct condition-based actions. As described in the research of current robotic technology, just sensing what's out there is an enormous hurdle to robots. With diverse fields, large amounts and, not to mention, different types of data must be considered in management. The first problem would be getting the data from the field to the robots. There are a few approaches to doing this. Robots could be equipped with a multitude of sensors and collect data directly from the field as they pass through the field. However, sensors consume processing power and electrical power, both of which are limited in a robot. Not only would having a robot cover acres of land, while collecting data every few feet, result in slow, inefficient use power and time but would require overcoming significant technical hurdles. A more efficient option would be to embed the field with cheap, small, wireless sensors to create a wireless sensing network. In this design, wireless remote sensors, known as motes, are spread across the field. These motes, designed for environmental conditions, contain sensors for environmental conditions such as temperature, humidity, and soil conditions. By transferring some of the burden of data collection from the robot to the wireless sensors, robots have more power available for their own operations. The motes obtain their power from small amounts of thin, photovoltaic film. To conserve power, motes spend most of the time in sleep mode. Master robots, described later, can collect motes from the field for storage or maintenance of motes. The sensor network will collect data on air conditions, light conditions, and soil conditions. Instead of looking for nutrient deficiencies symptoms in plants, the sensors will detect if the nutrient levels are too low in the soil for the particular plant species.

As a robot, specifically a hunterbot described later, passes by a mote, the robot sends a packet to the mote and receives the mote's sensor data. The robot stores the data temporarily and passes to a base station for storage and processing upon return to the base station. Based on the data originally provided by the sensor network of motes, robot management of the base station decides what treatment to apply to the field. For the minimal amount of management required for the polyculture field, robot management will decide to irrigate, fertilizer locally, weed, plant or harvest.

After the Sensor Network
Step 1 - Taking an Inventory After the sensor network is established, several robots will be sent into the field for mapping and initial inventory. These robots, known as hunter robots will be equipped with the tools necessary to take DNA samples from every plant on the field. For navigation in the field, the robots use a virtual map combined with range finding sensors. At the beginning, the virtual map contains nothing but the general plot areas of the farm. The hunter robots seek out the plot areas and map out navigable paths in each plot area by using the drunken sailor approach. To start the mapping process, a robot will circle around a plot area until it finds an area not blocked by plants. It will then head down that path until it gets too close to an obstacle, and then it will shift its direction until it reaches another obstacle, and so on. As the robot does this, it records the distances and directions traveled. Upon return to the base station, the robot overlays its paths on the base station's virtual map for sharing. As more and more robots report their paths, the virtual map assigns greater confidence values locations of paths. In this way, the virtual map "learns" where everything is, and the robots can rely more on the shared virtual map and less on their own sensors. With the map established, the robots determine and record the plants' locations. After collecting the samples, the robots will return to the base station where the samples will be subjected to numerous tests that can identify the plant. The laboratory will contain a revolutionary device-a recent invention that can perform "instant" DNA tests [8]. This invention will utilize silicon to shorten DNA amplification from several hours to just a few minutes. Using this new device, the lab will be able to quickly compare the DNA structure of a plant to a known structure, thus establishing and verifying the identity and location of all plants on the field. Because each type of plant has a unique DNA pattern, being able to identify and test plant DNA will be efficient and effective. The tests will include but will not be limited to "instant" DNA tests, mass tests, texture tests, and color tests. After performing the DNA test on each plant, the lab will note the type and location of each plant. After the initial inventory, the hunter robot, while making its rounds about the farm fields, will collect samples from new plants. The plant sample will be taken with a circular projection which will be placed into contact with the plant (see Figure). Finally, instead of wheels, as is most commonplace today, both the hunter and master robots will use treads to navigate through the fields. This mechanism will help the robots avoid soil erosion while also being able to move quickly throughout the field.

After the Sensor Network
Step 2 - Eliminating Weeds and Harmful Plants After the DNA tests and inventory round is completed, the location of the weeds will be sent by the network hub to another robot-the master robot. The master robot will be the principal farming tool on the farm-it collects crops and weeds, spreads fertilizer, and can plant seeds. The first task of the master robot, once inventory has been taken, will be to travel about the farm, going to the locations specified by the network hub, and picking the weeds. To pick plants, the master robot utilizes a crane like device. This device can not only increase and decrease in height, but it also has various extension which are used to pick plants or weeds. For example, a giant pair of scissors can extend from the edge of the crane and can be used to cut and harvest vines. After picking the weeds, the robot can place them in the large compartment located on the back of the robot for disposal.

After the Sensor Network
Step 3 - Eliminating Insects While eliminating monoculture and changing farming techniques will eliminate most insects on a farm and will do a great deal to decrease the need for pesticides, a few insects and pests will still exist on the farm. Accordingly, the robots should have a more active method of dealing with these pests. In order to further decrease the harm of insects and the need for pesticides, both the hunter robot and the master robot are equipped with a biomimicry box that will be used to attract and capture pests. The biomimicry box will contain a variety of pheromones that attract the most common pests and insects . Once inside the box, the insects will be unable to escape (See figures). Thus, while performing their normal tasks, the robots will also be fighting pests and insects, saving the crops and, perhaps more importantly, protecting the environment from the use of potentially dangerous pesticides.

In addition to mitigating harmful insects, the robots will also be able to assist in fending off plant disease. Once again, it is important to realize that by ending the practice of monoculture, most disease problems will disappear. But because the potential for disease still exists, the robots must be prepared. In order to fight diseases, the base lab will be able to perform the "TIGER", or Triangulation Identification for Genetic Evaluation of Risks, process. This process, as developed by the Agricultural Research Service, will inform farmers which plants are infected [9]. Further, TIGER, in tandem with the DNA samples, will allow the master robot to collect infected plants before disease can spread to other crops.

After the Sensor Network
Step 4 - Planting Seeds and Working with Fertilizer At certain times along the way, it will be necessary for a farmer to plant new crops and to spread fertilizer to encourage plant growth. The master robot will be responsible for both of these tasks. At the bottom of the master robot will be a slot that can be used to distribute fertilizer, soil, or seeds. For fertilizers and soil, the object will be placed in the storage compartment, and a fan-like device will be used to spread the fertilizer/soil along a given radius, allowing the robot to spread fertilizer in selected areas, at selected times, and under selected conditions, so as not to over-fertilize (see figure). Once again, however, the need for robotic fertilization will be miniscule-eliminating monoculture will greatly decrease the need for fertilizer. For seeds, the master robot works with the sensor network to drop seeds in the necessary location. Once the seeds are dropped, the robot travels around again, covering the seeds in soil and watering them.

After the Sensor Network
Step 5 - Harvesting the Plants As explained above, the master robot has several extensions to its crane that are used to harvest plants. Because the plants will not be monocultures, the farmer can decide to collect all crops together, or can have the sensor network tell the robot to go only to the locations with a certain type of crop. By using the crane extensions in conjunction with the large storage compartment, many crops can be collected before the storage maximum is reached.

Powering the Robot
Our robots and the base station are powered with renewable energy to promote sustainability. Depending on the geographic region, the base station is equipped with photovoltaic solar cells or wind turbines. Using solar power, the robots recharge their batteries during the day at the base station, and activate at night. If necessary, the robots can activate during the day on a partial charge. Using wind power, the base station is equipped with its own batteries to store energy from variable winds, and the robots recharge when necessary.

After the Robot: Looking to the Future For centuries agriculture has played an important role in human society. It has provided food and nutrients for billions, it has created jobs for millions of farmers, and it has woven its way into our culture. Now, as we venture into the 21st century with our many technological advances and new philosophies, agriculture is becoming a practice of the past—primitive and antiquated. In order for agriculture to remain a practical and important part of our society and culture, it needs to adjust with to the technological advancements of the world. To a certain extent, these adjustments have already begun. Genetically modified plants are increasing in popularity, and automated farming techniques are becoming more prevalent. Our proposal takes these practices one step further—creating a fully automated farm. Our proposal uses cutting edge technology from many fields of science—Biology, Robotics, Chemistry, Computer Science and Physics—in tandem with age old agricultural practices (monoculture and irrigation), in order to create the most efficient and technologically advanced farm in the world. This is just the first step in a long journey of technological discovery. The model put forth here paves the way for a future full of technological advancements and discovery—the possibilities are endless!

Bibliography [1] http://www.globalchange.umich.edu/globalchange2/current/lectures/human_pop/human_pop.html

[2] http://www.equalearth.org/agriculturepollution.htm

[3] http://www.cnr.berkeley.edu/~agroeco3/modern_agriculture.html

[4] http://www.historyplace.com/worldhistory/famine/begins.htm

[5] http://www.sciencemag.org/cgi/content/abstract/314/5805/1598?rss=1

[6] http://www.fao.org/docrep/W4367E/w4367e0e.htm

[7] http://en.wikipedia.org/wiki/Arable_land

[8] http://www.msnbc.msn.com/id/16498079/site/newsweek/

[9] http://www.sciencedaily.com/releases/2007/01/070102140731.htm

[10] http://pmep.cce.cornell.edu/facts-slides-self/facts/mod-ag-grw85.html

[11] http://www.aces.uiuc.edu/news/stories/news3772.html

[12] http://dsc.discovery.com/news/2006/11/02/roboticweeder_tec.html?category=earth&guid=20061102134500

[13] http://www.calstate.edu/newsline/2007/n20070131fre1.shtml