Technical Summary: 

Year 1 Achievements

An ATRV-Jr. robot was purchased and outfitted with orientation sensors and a DGPS for autonomous guidance (See Figure 1).  The autonomous guided vehicle (AGV) is operational and is capable of GPS-based autonomous point-to-point guidance within a geo-referenced perimeter as per Year 1 objectives.  Obstacle avoidance routines are nearing completion (May 2002).  The ATRV-Jr. robot is ready for sensor package implementation (Year 2 objectives).    

Figure 2.  Computer generated search pattern with obstacles.

Year 2 Technical Summary

With the hardware and software programming necessary for the ATRV-Jr. to navigate autonomously within a geo-defined field complete, the next step is to develop sensor technologies for detection of fire ant mounds.  After reviewing various technologies that might be applicable, the research team has concluded that a combination of thermal infrared and sonar sensors boom mounted (similar to a spray boom) to the ATRV-Jr. would provide the best methodology for infield real-time fire ant mound detection. Green (8) reported that fire ant mound temperatures were generally higher than the surrounding soil.  This trend was consistent from April until August.  Preliminary testing, by the authors, revealed a 10 ^oF difference in fire ant mounds and surrounding vegetation (~70 ^oF ambient temperature) using a thermal infrared sensor.  Sonar (ultrasonic transducers) will be used to analyze feedback signals from the ground surface.  Using signal processing techniques, algorithms will be developed to discriminate fire ant mounds.  The combination of thermal infrared and sonar sensor technologies should provide the necessary delineation. 

Once fire ant mounds are located, the ATRV-Jr. will dispense the appropriate pesticide.  By systematically searching the entire field, pesticide will be applied only to the target.  This methodology will reduce environmental impact and improve economics by reducing the quantity of pesticide used to effective eradication. 

Objectives: 

1.      Year 1 - Successfully modify existing industrial robotic technology to accept and follow GPS-acquired position information as a primary control parameter for guidance.  Status: Complete.

2.      Year 1 - Test a modified autonomous guided vehicle (AGV) to determine its safety and usefulness as a platform from which to perform routine and repetitive agricultural tasks.  Status: Complete.

3.      Year 2 - Design and integrate sensor array package  - Ant Colony Sensor (ACS).

4.      Year 2 - Investigate the usefulness of the AGV/ACS integration. 

Procedure: 

Remote Sensing

Fire ant mound detection will be performed as a multi-step process.  First, a near infrared remote sensing approach will be used to determine the most likely locations for the ant mounds.  Ant mounds will appear dark in the infrared image and vegetation will appear light.  Other areas, such as bare soil, will also appear dark and will be differentiated using other methods.  Based on the near infrared data, the ATRV-Jr. robot will be dispatched to field locations with a higher probability for fire ant mounds for thermal infrared sensor testing. 

A remote sensing image will be taken using a single band digital Charge Coupled Device (CCD) camera with near infrared filters.  The camera will be supported by a tall camera pole to obtain high resolution images of the field to be tested.  The images will be geo-rectified using Erdas Imaging.  Once the image is rectified, the test algorithm will look for dark areas within the light vegetation.  A location file of the suspected sights will be loaded to the robot for further analysis.     

Ant Colony Sensor (ACS) Array Development, Construction, and Implementation

A sensor boom will be attached to the ATRV-Jr. robot platform.  The sensor boom will consist of multiple thermal infrared sensors to ensure full coverage of the ground.  The width of the boom will be input into the search grid generation software to insure that all portions of the test area are covered by the ATRV-Jr. during a systematic field survey.   

The sensor array will search for peaks in temperature on the ground surface.  Due to the thermal warming of the sun, ant mounds will typically have a higher temperature than the surrounding vegetation and soil.  Seasonal effects do influence the temperature differential.  This limitation would not be a prohibitive for implementing a fire ant eradication program using autonomous robotics, since field operations will typically be done in weather conditions that favor temperature differentials in the fire ant mounds.    

To provide another degree of freedom to the thermal infrared sensors, sonar sensors may be employed.  Due to the hollow nature of the mound, reflected sound waves will appear different than solid ground.  Using signal processing routines, a sound signature for fire ants will be used to truth the thermal data.     

Once an ant mound is detected, pesticides are distributed on the mound on a site specific basis.  Standard granular fertilizer applicators will be modified and fitted to the ATRV-Jr. for pesticide application.   

Effectiveness of the detection of mounds and of chemical application will be based on the correct identification of ant mounds and the distance the insecticide is discharged from the fire-ant mound detected.   

Justification: 

Automation technologies have greatly increased the productivity of manufacturing processes.  The two most commonly employed guidance systems for industrial robots today are wire guidance and laser guidance.  These two methods along with advanced guidance systems based on such technologies as radio telemetry and radar have been successfully employed for autonomous rovers and robots within structured environments such as laboratories, warehouses and manufacturing facilities in industry, worldwide.  Use of these technologies, however, is typically infeasible on the scale required for agricultural operations.  For this reason, more robust technologies are sought for use in the field and many potential solutions have been investigated (please refer to Section  F of the original 2001 ASTA proposal for a discussion of these).  GPS seems to be a good option given that GPS data is accurate enough to give acceptable position information for many applications.  This is the basis for the guidance system produced during Year 1 of this project and has proven to be valid as a means of guidance.  The ATRV-Jr. robot, developed during Year 1, provides a platform that can be used to provide a “proof of concept” for many agricultural operations.  As guidance and sensor technologies are proven and implemented, the system can be scaled up or down to achieve “real-world” agricultural/horticultural operations.  

Other Uses for Autonomous Guided Vehicles (AGVs)

Automation technologies have greatly increased the productivity of manufacturing processes.  The two most commonly employed guidance systems for industrial robots today are wire guidance and laser guidance.  These two methods along with advanced guidance systems based on such technologies as radio telemetry and radar have been successfully employed for autonomous rovers and robots within structured environments such as laboratories, warehouses and manufacturing facilities in industry, worldwide.  Use of these technologies, however, is typically infeasible on the scale required for agricultural operations.  For this reason, more robust technologies are sought for use in the field and many potential solutions have been investigated (please refer to Section  F of the original 2001 ASTA proposal for a discussion of these).  GPS seems to be a good option given that GPS data is accurate enough to give acceptable position information.  This is the basis for the guidance system produced during Year 1 of this project and has proven to be valid as a means of guidance.  The ATRV-Jr. robot, developed during Year 1, provides a platform that can be used to provide a “proof-of-concept” for many agricultural/horticultural operations.  As guidance and sensor technologies are proven and implemented, the system can be scaled up or down to achieve agricultural/horticultural operations using autonomous guided vehicles.  

Potential commercial applications for an all-terrain autonomous guided robot are extensive.  Conceivably, most of the repetitive field operations that now require human operators could be automated with accurate geospatial-based guidance and sensing capabilities.  Here are just a few examples: 

1.      Fire-ant control (as described in the original project proposal).

2.      Grid-based soil/vegetation sampling.

3.      Pesticide application (agricultural, horticultural, green industries, etc.)

4.      Precision weed control on large regions such as pastures, airport runways, theme parks, parking lots, etc.

5.      Automated monitoring of turf (height and condition) on golf courses and sports fields.   

Pesticide Application. 

The implications of the use of robotic technology in the area of worker safety alone are enormous.  Use of automated pesticide application, for example, would prevent worker exposure to spray drift.  A noteworthy consideration in this line of thought is that many substances thought to be safe at one point in time have later been found to have toxic effects, such as thalidomide (treatment for pregnancy discomfort which caused birth defects) and dichlorobromopropane (DCDB, a nematocide banned by the USDA in 1976 which caused male sterility).  Many chemicals themselves are non-toxic but become toxic upon being metabolized in the human body.  Because compounds affect different species in different ways—even different individuals within a species have varying susceptibility to toxicity—long-term effects are difficult to predict in humans based on lab tests on animals.  Clearly, it behooves employers and workers to minimize their exposure to pesticides.  Robotic technologies could reduce workers exposure to potentially harmful substances.

Worker Safety – Repetitive Tasks.

Another worker safety issue specifically addressed by automated task performance is the reduction of injuries in workers; especially cumulative trauma disorders (CTDs) in which long term exposure to ergonomically hazardous tasks causes the gradual development of a musculoskeletal or neurological injury.  Long work cycles in the presence of such stressors as awkward postures, extreme temperatures, and noise/vibration lead to these conditions, which cost not only in terms of medical expense but also in terms of productivity and product quality.  Well-designed robotic work increases precision in the performance of routine and repetitive tasks as well as isolating human workers from potentially harmful substances and decreasing the incidence of injury. 

Green Industries (Golf Courses, Sports Facilities, Theme Parks, etc.).

Golf courses and other green industries utilize large quantities of pesticides and fertilizers.  Site specific application of pesticides using autonomous robotics would provide a tremendous cost savings in terms of labor, and more importantly, this technology will reduce the environmental impact associated with usage of these products.  Since pesticide application is accomplished with the autonomous robot, human interface with potentially harmful chemicals is greatly reduced.   

Fire Ants. 

Fire ants cost Americans billions of dollars per year.  The cost of fire ant control in Texas alone is a billion dollars annually.  It is a multi-million dollar problem in Mississippi, with no county in the state immune.  Environmentally friendly site-specific eradication of fire ants is a priority for the USDA ARS.  In Mississippi alone, the federal request is 5.2 dollars for research in the area of fire ants (about 1.3 million to the ARS research group in Stoneville, MS).  With rising labor costs and shrinking profit margins, it will become increasingly important to automate as many repetitive operations as possible to shore up farm income.  

Self-Guided Soil Sampler: 

AVG travels to predetermined waypoints in a field and uses a low power-consumption coring tool to collect soil cores and returns multiple samples to the lab.  Goal:  Collection of soil samples with minimal labor costs. 

Autonomous Landscape Mapper: 

AGV traverses a region with known boundaries, collecting precise elevation data for input to GIS applications.  Goal: Very accurate (1/10^th m topographic resolution) scanning of target regions.

Robotic Plant Monitor for Horticultural Applications: 

AVG, equipped with colorimetric sensors, moves in and around plants (small trees, vines, or ornamentals) in a nursery, vineyard, or orchard setting to analyze the condition of plants and fruits/vegetables.  Goal: Reduction in judgment requirement for agricultural workers, quicker response to disease, and minimization of labor requirements. 

Literature Review: 

However, many advances have been made in the area of automation in agriculture just in the past decade.  Of course GPS/GIS technologies have been on the forefront of precision agriculture for a number of years.  Advances in colorimetric techniques have led to new uses for satellite imagery in crop stress detection (1,3,4).  Faster and more capable computing power, a limiting factor in some instances such as image processing for machine vision, is now available.  Machine vision has been used for crop condition analysis (5), for plant recognition (6,9) and as a guidance parameter (14,15,16,18).   

Advances in sensor development make it possible for robotic manipulators to physically handle plant material without damaging it (4,5,6,10,11).  Off-the-shelf guidance systems based on laser or telemetry techniques are also available to make autonomous roving machinery a reality.  These technologies are converging to produce real opportunities for advances in agricultural automation.  Some examples of existing high-tech equipment include robotic tomato, grape, and cucumber harvesters, as well as geranium-cutting and chrysanthemum-cutting transplanters (7,9,10). 

The limiting factor in many cases of agricultural automation is the lack of a suitable platform from which to operate such high-tech alternatives to human labor as machine vision and advanced manipulation techniques.  This machinery must be moved into the proximity of the product, using a guided vehicle, to be effective.  In an enclosed environment, such as factory floors, warehouses and greenhouses, nominal guidance technologies such as radio telemetry, laser, and electromagnetic wire, are employed with great success.  Use of these methods, however, does not extrapolate well into an outdoor situation.  For this reason, such methods as geomagnetic direction sensing, fiber-optic gyroscope, machine vision, wireless networking, and GPS have been investigated as possible guidance parameters for use in the field (13,15,16,20,21,22). 

Reid relates that GPS has been used with success as parameter vehicle guidance (17,18), but cautions that the most accurate tracking can be accomplished by fusion of a sensor types.  Noguchi, et al, reported success in guiding a tractor using this approach in 1998 (13).  It is important to note two important caveats with regard to standing work on guidance of this type:

1.     So far, research has been done with large vehicles—tractors with conventional power-trains and steering mechanisms.  It might be that smaller, nimbler machinery can utilize GPS position information as a sole navigation signal.

2.     Research published heretofore has proceeded during the era of Selective Availability (SA), an artificial introduction of error into the raw GPS signal by the Department of Defense designed to thwart its use by hostile parties.  Now that the DOD has disabled SA over the continental U.S., raw signals can be used to provide more accurate positional information that may be useful for guidance technologies.

Summary:

An ATRV-Jr. robot was purchased and outfitted with orientation sensors and a DGPS for autonomous guidance.  The ATRV-Jr. is operational and capable of GPS-based autonomous point-to-point guidance within a geo-referenced perimeter as per Year 1 objectives.  Obstacle avoidance routines are nearing completion (May 2002).   

The implementation of an ant colony sensor (ACS) array and pesticide distribution system will allow the ATRV-Jr. to apply pesticides in a site-specific manner.  More importantly, however, the success of this project will open the door for utilization of robotics in many other areas of agriculture. 

The entrance of MSU into the research area of autonomous robotic technology, as it applies to Precision Agriculture, opens the door to other research funding opportunities (NASA, DOD, etc.) and demonstrates our commitment to cutting-edge technological development.  The technology for the SCEU concept is real; it is only a question of who will do it.   

An all-terrain robotic platform has been developed during Year 1 of this project.  The robot can systematically survey fields at a search interval defined by the operator.  The goal of building an autonomous robotic platform capable of applying pesticides on a site-specific basis can be achieved by installing a fire ant colony sensor array package and a pesticide dispensing apparatus on the ATRV-Jr. 

References: 

Ahmad, I.S., J.F. Reid, N. Noguchi, and A.C. Hansen.  1999. Nitrogen sensing for precision agriculture using chlorophyll maps.  ASAE Paper 993035. St. Joseph, MI. 

Benson, E., Stombaugh, T., Noguchi, N., Will, J., and J.F. Reid. 1998. An evaluation of a geomagnetic direction sensor for vehicle guidance in precision agriculture applications. ASAE Paper 983203. St. Joseph, MI.  

Blackmer, T.M., Schepers, J.S., Meyer, G.E. 1995. Remote Sensing to Detect Nitrogen Deficiency in Corn. 

Fujiura, T., Yamashita, J., Kondo, N.  1992.  Agricultural Robots (1): Vision Sensing System.   ASAE Paper 923517. St. Joseph, MI. 

Kondo, N., Monta, M., Ymashita, J., Fujiura, T.  1992. Agricultural Robots (2); Manipulators and Fruits Harvesting Hands. ASAE Paper 923518. St. Joseph, MI. 

Monta, M., Kondo, N., Shibano, Y., Mohri, K., Ymashita, J., Fujiura, T. 1992.  Agricultural Robots (3): Grape Berry Thinning Hand. ASAE Paper 923519. St. Joseph, MI. 

Gopala Pillai, S., Tian, L, and Beal, J. 1998. Detection of nitrogen stress in corn using aerial imaging.  ASAE Paper 983030.  St. Joseph, MI. 

Green, P. 1998.  Effect of Imported Fire Ants on Mississippi Soils.  Masters Thesis, Department of Plant and Soil Science, Mississippi State, MS. 

Iida, K., Suguri, M., Umeda M., Matsui, T.  2000.  Estimation of Nitrogen Content Using Machine Vision in  Paddy Field.  ASAE Paper 003021.   St. Joseph, MI. 

Kondo N., Monta, M., Fujiura, T.  1996. Fruit Harvesting Robots in Japan. Advanced Space Research, 18, 1 / 2, 181-184. 

Kondo N., Shibano, Y., Mohri, K., Fujiura, T., Monta, M.  1992.  Request to cultivation method from tomato harvesting robot.  Acta Horticulturae, 319, 567-572. 

Kondo N., Ogawa, Y., Monta, M., Fujiura, Shiban, Y.  1996b.  Visual Sensing Algorithm for Chrysanthemum cutting sticking robot system.  Acta Horticulturae, 440, 383-388. 

Noguchi, N., J.F. Reid, J.Will, and E. Benson. 1998. Vehicle automation system based on multi-sensor integration.  ASAE Paper 983111. St. Joseph, MI.  

Noguchi, N., Reid, J.F., Q. Zhang. and L.F. Tian. 1998. Vision intelligence for precision farming using fuzzy logic optimized genetic algorithm and artificial neural network. ASAE Paper 983034. St. Joseph, MI 

Pinto, F., J.F. Reid,  Q. Zhang, and N. Noguchi. 1999.  Guidance parameter determination using an artificial neural network classifier.  ASAE Paper 993004. St. Joseph, MI.  

Pinto, F. and J.F. Reid. 1998. Heading angle and offset determination using principle components analysis. ASAE Paper 983113. St. Joseph, MI. 

Reid, J.F. 1998. A status report on Autonomous Guidance of Agricultural Vehicles in the US: New Frontiers in the 21st Century. RSJ Meeting, Sapporo, Japan. UILU-ENG-98-7026. 

Reid, J.F. 1998. Precision Guidance of Agricultural Vehicles. JSME Meeting, Sapporo, Japan. UILU-ENG-7031. 

Simonton, W.  1992.  Automation in the Greenhouse: Challenges, Opportunities, and a Robotics Case Study.  Hort Technology 2(2), 231-235. 

Will, J., T. Stombaugh, E. Benson, N. Noguchi, and J.F. Reid. 1998. Development of a flexible platform for automatic agricultural guidance research. ASAE Paper 983202.  St. Joseph, MI. 

Will, J.D., D.D. Moore, E.N. Viall, J.F. Reid and Q. Zhang. 1999.  Wireless networking for control and automation of off-road equipment.  ASAE Paper 993183. St. Joseph, MI. 

Zhang, Q., J.F. Reid, and Noboru Noguchi. 1999. Agricultural vehicle navigation using multiple guidance sensors.  Proceedings of the International Conference on Field and Service Robotics. August 29-31.  Pittsburg, PA. UILU-ENG-99-7013. 

Current Research: 

1.      Phase II GPS Wastewater Tracking System (GWTS) for Commercial Swine Production Facilities; Sponsor: USDA/CREES Special Grant (Burcham, Thomasson, Lee).  In progress.  An investigation of the use of GPS position information to effect variable rate application of animal waste lagoon water onto a field.  GPS information is transmitted from a mobile station mounted on traveling gun of a hard hose reel by radio modem to a remote base station.  The position information is then used by an in-house program to make decisions about speed control and a control voltage is output to motor controller.  This project is complete.  We are preparing a journal article to be submitted the Applied Engineering in Agriculture. 

2.      Robotics in Agriculture Research project (Lee, Thomasson).  Ongoing investigation and literature review on the use of robotics in horticultural applications.  This has been conducted with the collaboration of the Howell family at Rocky Creek Nursery, Inc. (Lucedale, MS), Dr. Hagen Schempf of the Robotics Institute, Carnegie Mellon University, and the Department of Plant and Soil Sciences, Mississippi State University. 

(PI) Timothy N. Burcham, Assoc. Prof, Ag. & Bio. Eng. Dept.

(Co-PI)  Paul Lee, Research Scientist, Ag. & Bio. Eng. Dept.

Back to Top

Back to Menu