Progress 10/01/20 to 06/30/21
Outputs
Target Audience:1. The target audience included high school agriculture teachers, engineers, scientists, college professors, and undergraduate students. 2. Technology was demonstrated and shared with farmers at the Black Farmers Conference. Approximately 20 farmers, students and staff attended. 3. Meetings were held with approximately 8-10 agriculture industry entrepreneurs, farmers and investors to discuss the efficacy of directed energy as an integrated pest management strategy. 4. Undergraduate student researchers received significant training in a variety of computer software skills including: python computer language, basic coding style and theory;Raspberry Pi computer and display; remote computing skills and sensor integration for atmospheric readings and ultra-sonic sensor integration;basic code style including trouble shooting, structure and integrating class files, effective sorting algorithms and programming style and data normalization; andsoftware code upgrades including safety features, redesign of turning algorithm and diagnosing communications error between code and hardware. 5. Research team gave 4 talks to visiting high school students, high school teachers and volunteers concerning agriculture robotics and alternative methods of killing weeds including directed energy. Each group had approximately 15-20 students and teachers. 6. Workshop to 10 agriculture high school teachers on weed seed viability and killing weed seeds with directed energy as an integrated pest management strategy. 7. Virtual presentation to engineers on direct laser image recognition for self-driving vehicles as part of the Frontiers in Optics+Laser Science Conference and Exhibition. Presentation was targeted to agriculture technology experts and engineers. 8. Integrated Computer Vision and Machine Learning topics in three upper division computer science courses targeted to undergraduate students. Changes/Problems:One research faculty hired as the Principle Investigator and Robotics Engineer started and left 7 months later. A new faculty member with a robotics background has been assigned to the project. Academic faculty make up the bulk of the researchers for this project, so fewer FTEs are devoted to this project than is needed. Requests for new hires, including a mechanical engineer, electrical engineer and robotics technician are moving through the system. These additions will help expand the capability of this plan of work at CSU. What opportunities for training and professional development has the project provided?Training and professional development opportunities were limited due to the pandemic restrictions. Researchers had 4 presentations based on research from summer, 2021, one workshop to high school agriculture teachers and one farmer conference to share research information. How have the results been disseminated to communities of interest?Research was disseminated to communities of interest through workshops with high school agriculture teachers and farmers, meetings with entrepreneurs, professional conferences (engineers and scientists), high school student (FFA) and undergraduate student researchers. What do you plan to do during the next reporting period to accomplish the goals?Efforts will be continued in the next five-year cycle to develop advanced agricultural technologies in following areas: ?Robotics and AI: Targeted to small-scale farmers, low-cost robotics platforms for multiple precision agriculture solutions will continue to be developed and disseminated. Focus will be on automating complex, repetitive and labor-intensive agricultural operations for sustainability. UAVs: UAVs or drones software will be developed to enhance sustainable farming solutions including scouting and crop monitoring for weed pressure, seed emergence, field conditions, and detecting irrigation system defects. We plan to use drone technology to develop solutions that benefit small farms including evaluation of efficacy of dry and liquid product application using UAVs for small-scale operations. Internet of Things (IoT) and Precision Irrigation: We plan to develop low-cost robotic platform for sensing soil properties. Most crop agriculture is rain-fed in Ohio and other parts of Midwestern USA, but irrigation plays a critical role in specialty crop production. Demand is growing for automated irrigation systems that integrate irrigation requirements, soil sensing, rainfall data, and IoT can enhance productivity and sustainability for small-scale operations.
Impacts
What was accomplished under these goals?
Goal - Expand field testing and the range of plants/applications for current directed-energy, mechanized weed control systems already being tested at CSU.An image acquisition robot that was built from scratch and dubbed AGI-1 can now traverse an agricultural field and automatically take and store images using its two facing-down cameras. Databases were developed of commodity and specialty crops to train visualization software to identify crop from both broad leaf and grass weeds. Databases include: 5-year Soybean database; 3-year Sweet Corn database; and 1-year Hemp database. A study of 1000 broad leaf and grass weeds in agricultural fields has demonstrated that directed energy is up to 84% effective in killing and damaging weeds with under a 10x10 cm squared leaf spread. A combination of IR and LED blue lights was validated to kill non-shattering, broad leaf weed seeds including morning glory, pigweed and water hemp. Implications are that weed seeds entering chaff during harvest will not contribute to the soil seed bank and subsequent year infestation of herbicide-resistant weeds. Goal - Develop novel, field-ready, scalable, integrated pest management strategies for agriculture, forestry, and other land management needs.Robotic software was upgraded by optimizing functionality including: added an auto resizing feature to graphic user interface; identified and fixed alignment issues with video capture program GUI; fixed 'broken' buttons in the software to switch between cameras in preview mode; improved picture recording program to make use of multi-thread to eliminate desynchronization between cameras; resigned the picture recording program to make it user friendly and to be used with any computer (assuming proper software is installed); improved the base framework's readability, efficiency, and overall structure of the program to make future modifications much easier; adjusted frame rate calculation to ensure it runs exactly at the desired frame rate; and added adaptive folder detection and image synch functionality to field robot. The field robotic platform was upgraded including: rewired circuitry; recalibrated sensors; reprogramed and upgraded the controller; replaced tires and patched battery cage.Magnetic sensors were aligned.A custom camera housing was designed for the field robotic platform for 3D crop scenes and developed a new process for taking pictures that will decrease the number of cameras required saving up to $900 in equipment.This will make the field robot cheaper to run and maintain for farmers. Testing and experimenting continued with robot arms for the robotic platform to deliver directed energy to weeds in field.Robotic arm was maintained and repaired. Computer software and controls for the CSU Greenhouse were upgraded including rewiring, replacing faulty sensors, weather station and control modules.Changes and upgrades allow greenhouse research on seed germination and vegetative soybean stress indicators to continue. Goal - Develop, test, and apply business models that take advantage of these benefits to small/specialty farms (existing and new farmers, producers).A validation study was completed for an industry produced Weed Seed Destroyer prototype to determine parameters to kill broad leaf and grass weed seeds in chaff to develop a new option for non-chemical and sustainable integrated pest management to control weed infestations, especially of herbicide resistant weeds. Researchers have met with entrepreneurs and organic farmers to describe the results and potential benefits of minimizing viable weed seeds in the soil seed bank. Goal - Adapt the underlying technology platform to new applications, informed by outcomes and potential new uses identified from CSU's existing (capacity building) research.Broad spectrum directed energy (UV A, B, visible, near-IR and IR) on vegetative soybean plants in the field (V2/V3 stages) has been shown to enhance soybean yield in the field by up to 20%. Directed energy treatment does not affect quality of the bean, just the quantity. Research is ongoing to identify stress markers and the biochemical mechanisms for yield enhancement. A combination of IR and LED blue light (440 nm) have been shown to kill weed seeds to potentially reduce viable weed seeds in the soil seed bank. Research has demonstrated that low levels of IR+LED blue light also may enhance seed germination rates and total germination for soybeans. Experiments are ongoing for soybean and hemp seeds.
Publications
- Type:
Journal Articles
Status:
Under Review
Year Published:
2021
Citation:
Hongbo Zhang, Cadance Lowell, Deng Cao and Augustus Morris. Optical Weed Control: A Critical Review. Computers and Electronics in Agriculture, Elsevier, 2021 (under review).
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
" Hongbo Zhang, Wenjing Zhou, Jie Lin, Isreal Williamson, Wakeel Idewu, and Deng Cao. Robust Direct Laser Image Recognition of Vehicle Parts Among Self-driving. In Frontiers in Optics + Laser Science (FiO LS) conference and exhibition, 2021
- Type:
Other
Status:
Other
Year Published:
2021
Citation:
Lowell, C., Henry, J. 2021. Will My Seed Grow: Viability Testing Standards. Agricultural Education Summer Conference. Wooster, OH. Workshop, June 15-17.
- Type:
Other
Status:
Other
Year Published:
2021
Citation:
Lowell, C., Cao, D. 2021. Organic Weed Control. Black Farming Conference, September 10-11, Wilberforce, OH. Invited Talk.
|
Progress 07/21/16 to 06/30/21
Outputs
Target Audience:
Nothing Reported
Changes/Problems:The Agriculture Technology research program was initiated with this 5-year plan of work at Central State University. It has incorporated undergraduate students, 3 academic faculty, research faculty and one programming technician. A second engineering technician was hired and left after 2 years.One mechatronics research faculty was hired in 2020 and left in 2021. A second precision agriculture research faculty was transferred from another research group to Agriculture Technology in 2021. Moving forward and to expand the research, new hires in engineering and programming are in the works to expand this vital plan of work at Central State University. What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Under this project, efforts will be continued next five-year cycle to develop advanced agricultural technologies in following areas: Robotics and AI: Data driven, and AI-based digital agriculture can provide radical solutions to challenges in agricultural production systems not just by automating data acquisition and processing, but also by automating complex, repetitive and labor-intensive agricultural operations like weed management, pest management, disease management, crop scouting, fruit picking, and harvest automation. Under this project, more work is being done to develop artificial intelligence-based algorithms and low-cost robotics platform suitable for multiple precision agriculture solutions for small -scale growers. These technologies will be evaluated in lab and field conditions. UAVs: UAVs or drones are steadily becoming a mainstream technology on a farm. Drones offer an effective approach to efficient agricultural management that helps farmers, and service providers to quickly gain insights into their crops or assets. These provide solutions like scouting and crop monitoring for weed pressure, seed emergence, field conditions, and detecting irrigation system defects. We plan to use drone technology to develop solutions that benefit small farms including evaluation of efficacy of dry and liquid product application using UAVs for small-scale operations. We also plan to develop variable rate application and generating as-applied maps for product application technology using UAVs. IoT and Precision Irrigation: many of the row crop agriculture is rain-fed in Ohio and other parts of Midwestern USA, but irrigation plays a critical role in specialty crop production. Irrigation is considered to play important for row crop production in more areas in Midwest in near future under climate change scenarios. More data and information are required to measure evapotranspiration and irrigation water requirements for specialty crops of Ohio. The demand for automated irrigation systems that integrate irrigation requirements, soil sensing, rainfall data, and IoT technology is on the rise in general, and it can certainly benefit the small-scale operations as well. We plan to develop low-cost robotic platform for sensing soil properties. Smart irrigation controllers will be developed for specialty crops. In case of row crops, we plan to use historical weather, precipitation, soil, and yield data to develop crop yield functions under climate change scenarios. AI algorithms will be developed to study crop yield spatial and temporal variability and its dependencies on other factors. The objective of this study is to find areas that might benefit from investing in irrigation setup in near future.
Impacts
What was accomplished under these goals?
A robotic lab was established with equipment including tools, a robotic arm, drones and computers to develop autonomous pest management strategies for small-scale farmers. Field equipment was purchased to initiate non-chemical, integrated pest management strategy field trials in corn and soybeans. Research faculty and technicians were hired with experience in autonomous systems and computer visualization. These faculty and staff joined existing academic faculty in computer science, manufacturing engineering and plant science to build the agriculture technology program at Central State University from the ground up. In addition to the Evans-Allen funds, this research has generated 2 research Capacity Building Grants ($600,000 total), and one validation study ($14,000) during the 5 years of this plan of work. Goal - Expand field testing and the range of plants/applications for current directed-energy, mechanized weed control systems already being tested at CSU. Three picture databases used for robotic visualization training were produced including: 5-year Soybean database; 3-year Sweet Corn database; and 1-year Hemp database. A prototype weed seed destroyer developed by an industry partner was validated to kill up to 90% weed seeds in chaff as an integrated pest management strategy to keep herbicide resistant non-shattering broad leaf weed seeds out of the field soil seed bank. A model system using morning glory (Ipomoea tricolor L.) was established for testing all parameters of the this weed seed destroyer prototype to maximize weed seed control. Morning glory seeds are easy to purchase, germinate quickly and respond to changes in testing parameters. A commercially available hand-held "WeedErase" device was used to deliver 0.05W/cm2 mid-IR (2-5 um) and 1.54 W/cm2 of 440 nm light per weed of directed energy. Using a visual rating system after 3 days, 51% of the weeds were killed, 32% were damaged, and 16% showed little to no damage. With 83% monocot and dicot weeds controlled using DE, prototype agricultural autonomous delivery systems are being developed for field use. Small scale testing in the CSU greenhouse and newly developed Precision Agriculture Laboratory, show that directed energy kills dandelion, crabgrass and ragweed. Testing also has been initiated on the riparian corridor invasive species, Japanese knotweed. In joint project with the natural products research group here at CSU to study if American species of knotweed contain resveratrol, a heart healthy compound. This compound currently is imported from Asia. Goal - Develop novel, field-ready, scalable, integrated pest management strategies for agriculture, forestry, and other land management needs.One agricultural robot was built and field tested -- Agi-1, a payload and image acquisition robot.Agi-1 is a radio controlled, battery powered, robotic platform that can deploy a payload of up to 300 pounds, as well as, deployment of optical sensors for data gathering. Two directed energy devices were developed and manufactured by business partner GlobalNeighbor, Inc. that were purchased and are being adapted for use at CSU. A field ready, directed energy device was field tested on soybeans beginning spring 2017. A second directed energy device was purchased to deliver up to 180X sun power to kill weeds up to 20 feet. This device is being adapted for use to sanitize lumber and kill insect and bacteria pests, and to systemically kill weeds. The device was delivered in 2021 with hopes of testing in 2022. A console based program named, "FieldAnalytics.jar" was developed that utilizes OpenCV libraries to calculate the amount of green in a user defined quadrilateral or group of quadrilaterals within an image. This software is being utilized to calculate the amount of green plant life in CSU research fields. This data is then used to determine the viability of various farming techniques. A Multi-Threaded user-friendly software called "Multi-Cam Capture or MCC" was designed for the Agi-1 platform that captures the feed from multiple cameras at precisely the same time. Recorded data is then used in training algorithms for object detection. Software was designed to detect plant life, track its location in real time in relation to a directed energy field delivery system, the DA2. light payload, and send commands to operate the payload. Has been updated numerus times to keep it functional with updated version of OpenCV. Goal - Develop, test, and apply business models that take advantage of these benefits to small/specialty farms (existing and new farmers, producers).The majority of this plan of work was to establish agriculture technology and precision agriculture at Central State University. Faculty had worked with entrepreneurs at Global Neighbor, Inc., a small, woman owned local business developing new solutions for pest management. Faculty have reached out to The Entrepreneur Center, Dayton, OH funded through Jobs Ohio, international groups (Ireland farmers) interested in the technology and worked with faculty at the University of Dayton student design teams and an NGO to determine if directed energy weed control would work in African nations such as Tanzania. Goal - Adapt the underlying technology platform to new applications, informed by outcomes and potential new uses identified from CSU's existing (capacity building) research.Directed energy (DE) with different combinations of wavelengths and intensity have been shown both in the greenhouse and field that it can kill weeds both in a field and at a distance, kill weed seeds, sterilize bark and wood pallets, potentially help control Japanese knotweed, and enhance seed germination and soybean yield - all without the use of chemicals. In the form of broad spectrum light (UV A/B to IR), DE has been shown to field kill weeds. A combination of IR and LED blue light (440 nm) have been shown to kill weed seeds to potentially reduce viable weed seeds in the soil seed bank. It also has been shown to surface sterilize wood (bark and wood pallets), decreasing bacterial and fungal colonies. Combinations of UV A/B and IR also have been shown to field enhance soybean yield up to 20% when vegetative plants are treated (V2/V3 stages). Research is ongoing to identify the biochemical mechanism for yield enhancement.
Publications
- Type:
Conference Papers and Presentations
Status:
Under Review
Year Published:
2022
Citation:
Cao, D., Jolly, J. Cloud Computing Based Plant Classifiers and Their Real- Life Research Applications, American Society for Engineering Education Annual Conference, 2022 (Abstract accepted, paper under review)
- Type:
Other
Status:
Accepted
Year Published:
2022
Citation:
Jolly, J., Cao, D., Oral Presentation. Field Sweet Corn Classification Using Artificial Intelligence and Cloud Computing, Association of 1890 Research Directors (ARD) symposium, 2022 (Abstract submitted in 2021 and accepted)
- Type:
Other
Status:
Accepted
Year Published:
2022
Citation:
Jones, D., Robinson, A., Lowell, C. 2022. Poster Presentation. Directed Energy Applications for Weed Control, Association of 1890 Research Directors (ARD) symposium, April 2-6, 2022 (Abstract submitted in 2021 and accepted).
- Type:
Other
Status:
Accepted
Year Published:
2022
Citation:
" Robinson, A., Lowell, C. 2022. Poster Presentation. Model System to Evaluate Directed Energy Weed Seed Control at Harvest, Association of 1890 Research Directors (ARD) symposium, April 2-6, 2022 (Abstract submitted in 2021 and accepted).
|
Progress 10/01/19 to 09/30/20
Outputs
Target Audience:The target audience included the public, engineers, scientists, collegeprofessors, graduate students and ungraduate students. The following information was shared: Database collection and plant classifersat the virtual annual conference American Society for Engineering Education (ASEE). Significance of directed energykilling weed seedsincorn and wheat chaff was sharedat the in-person annual conference of the Weed Science Society of America. Video produced by Ohio Country and aired on cable TV on non-chemical means of killing weeds and the importance of cover crops in integrated pest management strategies- https://www.youtube.com/watch?v=BAE6XdB1rWw Graduate students at the University of Cincinnati and University of Dayton who participated in the design of automated field prototypes to deliver directed energy to kill weeds. Undergraduate students who participated in processing pictures for databases; and in the design of experiments and testing of directed energy in killing weeds, weed seeds,and enhancing growth and yield in soybeans. Changes/Problems:When the University shut down March 19, all research with students went online and virtual. Students were not available to help insummer research plots.Delivery for atractorthat was purchased for spring planting 2020 has been delayed until spring 2021 at the earliest. All meetings went virtual or were cancelled. This atmosphere made it very difficult to collect data and to disseminate research findings. The research team alsohas been hampered by not have a dedicated researchlab, an engineer technician,or a dedicated, 100%research faculty. We hope to rectify this in the next year. What opportunities for training and professional development has the project provided?Training and professional development has targeted the general public, engineers, scientists, collegeprofessors,six graduate students, six ungraduate students and one high school student. Database collection and plant classiferswere presented at the virtual annual conference American Society for Engineering Education (ASEE). The significance of directed energykilling weed seedsincorn and wheat chaff was sharedat the in-person annual conference of the Weed Science Society of America. A video produced by Ohio Country and aired on cable TV on non-chemical means of killing weeds and the importance of cover crops in integrated pest management strategies- https://www.youtube.com/watch?v=BAE6XdB1rWw Six graduate students at the University of Cincinnati (1)and University of Dayton (5)who participated in the design of automated field prototypes to deliver directed energy to kill weeds. Directed energy delivery systems and programing was part of their class projects or summer internship. Sixundergraduate student participated in processing pictures for databases; and in the design of experiments and testing of directed energy in killing weeds, weed seeds,and enhancing growth and yield in soybeans. Each undergraduate student has a specific project related to the plan of work. One high school student is learning to optimize the movement of a robotic carthrough a field using rasberry Pi. He has learned how to structure and integrate class files; basic coding style make and use sorting algorithms effectively; basic optimization practices;data normalization; and problem solving skills. How have the results been disseminated to communities of interest?Results have been disseminated to individuals at the scientist level through scientific conference papers and presentations both through the American Society for Engineering Education (ASEE) and the Weed Science Society of America (WSSA). Information on directed energy as an organic method of killing weeds was disseminated through Ohio Cable TV to the general public. What do you plan to do during the next reporting period to accomplish the goals?The team will continue to work on automated agriculture technology including robots to provide delivery systems to organically kill weeds in agricultural and natural settings. There currently are 2 field robots available and several pull behind devices. A new mechatronics engineer was hired (starting in January 2021) to continue to develop and refine the use of a small, relatively inexpensive robot to kill weeds using directed energy in the field. Databases for corn and soybean are being used to train a robotic arm to distinguish weed from crop both between and within rows and kill the weeds with above 90% accuracy. The robot and robotic arm will be field evaluated for accuracy in 2021. New tough laptop computers have been purchased to better supply computer power in the field.Traditional laptop computers cannot handle field temperatures and overheat. We also plan to weatherproff the circuitry on all robotic parts. Designs have been implemented with engineer graduate students at the University of Dayton to develop a push behind directed energy device to work in conjunction with cover crops as an integrated pest management strategy for small farmers. There will be a push to build this device and validate its effectiveness by 2022. A large directed energy device, the LS, wasdonated to Central State University by Edwards Airforce Base for additional validation testing. Designed to kill weeds at a distance of 10-20 feetusing directed energy, the device will be field testedon killing weeds in powerline cut throughs. Validation of directed energy as a means to kill weeds will contnue. Directed energy of different wavelength combinations are being tested to demonstrate capability in killing weed seeds, insects and bacteria on plants in the Central State University Greenhouse and in research field plots. Directed energy also has demonstrated the ability to increase growth and possibly yield of C3 crops such as soybean. While growth has been shown to increase in directed energy treated soybean plants, the yield and seed composition has yet to be documented. A continued push will be to demonstrate that directed energy can kill non-shattering weed seeds in chaff and decrease viable weed seeds in the seed bank.
Impacts
What was accomplished under these goals?
Research Goal 1: Develop novel, field-ready, scalable, integrated pest management strategies for agriculture, forestry, and other land management needs. The team is developing automated organic weed control system using directed energy and deep learning. The directed energy device is called the distributed array (DA), which is developed by a local small company, Global Neighbor Inc. We also use transfer learning to fine-tune a number of pre-trained neural networks and develop customized neural networks to distinguish various weeds from crops. The system is implemented using two languages: MATLAB and Python (with Tensorflow library), and is tested on three platforms: PC, Android Device and Raspberry PI. In order to train deep learning recognition of crop plants (versus weeds) for automated systems, three databases of sweet corn, soybean and hempwere developed. The soybean database was started in 2017 and in 2019-2020, 33,000 new soybean images were added and this database was manually cleaned by deleting redundant images. The cleaned databaase was downsized to approximately 50,000 images and the original database has been archived. For the hemp database, an additional 14,000 images were added. The Central State University-made robot designated AGI1 had a flat tire during field operation. During repairs,a number of software and hardware upgrades are made to AGI1 to improve its functionality. The software was upgrated by optimizing fuctionalility by: 1) adding an auto resizing feature to GUI interface 2) identifying and fixing alignment issues with video capture program GUI; 3) fixing 'broken' buttons in the software to switch between cameras in preview mode; and 4) imaging sync functionality to AGI1. The robot was rebuilt including a circuitry rewire, sensor recalibration, controller reprogramming and upgrading, tire replacement and battery cage patching. Industry partner, Global Neighbor, Inc. developed a new LED directed energy light system based on predominantly blue and infrared wavelengths rather than broad spectrum light wavelengths.The energy consumption is reduced by 70% and the operational temperature of the new LED light is much lower. The light is small and can be armed by tempered glass, which means improved safety and toughness. However, the empirical lethality seems to be lower. The new recipe works better on seedlings than on grown plants. A new object-detection-based study of weed identificationwas initialized. In this approach we trained YOLO (you only look once)to detect one or more custom objects in the scene using Google's free online GPU resource. This study can lead to a more precise weed localization when they are sparsely distributed in the field. Research Goal 2: Expand field testing and the range of plants/applications for current directed-energy, mechanized weed control systems already being tested at CSU. The Central State University greenhouse control room was rewired and software was upgraded so to make the software/server control suite operational to monitor and control climate controls in each grow room. The temperature controlled growth chamber also was fixed so that it may be used for future research. In 2019 on the Central State Experimental Farm, directed energy was shown to stimulate soybean growth in the field. In 2020 on a farm in Wellington, Ohio, this experiment was repeated. Field grown soybeans were designated toa random plot design were treated with directed energy for 2, 5 and 10 sec duration under both wet and dry field conditions; and at the VC/V1 stage versus the V2/V3 stage. Growth was both stimulated and plants reached more advanced growth stages faster compared to a control when plants at the V2/V3 stage were treated for 10 sec with directed energy in dry soil. Directed energy also wasevaluatedfor weed seed control at harvest. Chaff was collected during harvest from a waterhemp infested field in Lac qui Parle County, MN. Chaff was treated with 4 W/cm2 of directed energy from a metal halide light source (340-2000 nm) for 15 or 30 sec using either a quartz (transmitting UVA/UVB) or borosilicate glass (blocking UVA/UVB) plate. After treatment, seeds were separated from the chaff and germinated at 90 degrees F in the dark with 100 seeds per dish, 3 dishes per treatment and control for a total of 1500 seeds. Control germination rates averaged 35%. There was no difference in germination rates (p<0.05) among treatments up to 15 days. Total number of seeds germinated ranged from 26% (glass, 15 sec) down to 4% (glass, 30 sec). Seeds from each treatment and control were planted in seed mix in pots with 50 seeds per pot, 4 pots per treatment. No seeds germinated from the glass treatment at 30 sec, while other treatments germinated similarly to the petri dish experiments. Directed energy from high intensity discharge lamps with a broad light spectrum does work to kill waterhemp seeds in chaff and prevents viable seeds from entering the seed bank. Experiments are underway to test this directed energy treatment on other non-shattering broad leaf weed seeds collected after harvest from agricultural fields in Greene County, Ohio including marestail, evening primrose, burdock, and poison hemlock. Research Goal 5: Adapt the underlying technology platform to new applications, informed by outcomes and potential new uses identified from CSU's existing (capacity building) research. We are expanding the potential use of directed energy as an organic integrated pest management strategy in residential areas to agricultural fields and natural areas. Directed energy, broad spectrum light that includes UVA/UVB and IR has been shown effective against recently germinated broad-leaf and grass weeds. It also has been shown effective to kill weed seeds from non-shattering weeds such as waterhemp in chaff post-harvest. On an entirely new possibility, directed energy with increased IR has been shown to increase the growth and potentially the yield of field grown soybeans. Delivery systems are being developed to better fieldtest these new applications.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2020
Citation:
Deng Cao, Cadance Lowell, Craig M. Schluttenhofer, Augustus Morris, Austin R Erdman and Torry Johnson, Undergraduate Research: Deep Learning Based Plant Classifiers and Their Real-Life Research Applications. In American Society for Engineering Education (ASEE) Annual Conference, 2020. The original paper can be found here:
https://drive.google.com/file/d/10z9AXaSgzIDKMQzbyxHWiaY5eG90UOLS/view?usp=sharing
- Type:
Other
Status:
Published
Year Published:
2020
Citation:
Common Waterhemp (Amaranthus tubercualtus): Directed Energy Manangement of Weed Seed Bank in Corn. Cadance A. Lowell*1, Marcus Nagle1, Deng Cao1, Jon Jackson2; 1Central state university, Wilberforce, OH, 2Global Neighbor, Inc., Centerville, OH (235)Presented at the 2020 Annual WSSA Conference, Maui, HI.
|
Progress 10/01/18 to 09/30/19
Outputs
Target Audience:This agricultural technology integrated pest management research will impact the following target audiences: Individual and Institutional Researchers (university or industry): to benefit from access to affordable technical assistance for idea generation and modeling, proto-type development, independent product assessment, continuing research recommendations, and specialized facilities. Industry Partners: to benefit from accelerating product integration and market access while reducing risks and costs. Growers (farmers, community gardeners, land managers, others): to benefit from an affordable, proven non-chemical technology for weed mitigation in agriculture applications. CSU Faculty and Research Faculty: improve partnerships with CSU faculty, CSU research faculty, industry experts and business partners; create better opportunities for new researchers and scholarly publications; and increase the relevance of hands-on learning for their students. CSU Undergraduate Students: to increase the relevance of hands-on learning and to improve access to future employment or graduate programs through mentoring, professional internships, conferences, and publications. Prospective CSU Undergraduate Students: to increase attraction to CSU programs, with an emphasis on new courses of study being developed as part of the 1890 Land Grant University system. Changes/Problems:Dedicated research science faculty in the areas of agricultural technology and robotics, GIS and agricultural engineering positions are being advertised and these new research scientists will be added to this project. Progress may move at a faster pace with additional expertise. What opportunities for training and professional development has the project provided?Six undergraduate minority student research through summer internships in agriculture related computer science visualization, weed science and manufacturing engineering. The research team visited Kory Krofft, employee at Trimble Navigation Inc., for autonomous driving and advanced navigation technology. The research team also worked closely with engineers at Global Neighbor, Inc. for additional training and learning about directed energy systems. How have the results been disseminated to communities of interest?Finding integrated pest management strategies to control weeds in agricultural settings is key to maximize profit. This project has generated information for Ohio farmers, K-12 instructors and students, and the general public regarding non-chemical methods for weed control including directed energy, cover crop crimping and no-till weed seed suppression. Knowledge has been transmitted through participation in professional engineering conference (poster), outreach to Ohio educators and farmers, outreach to high school students on campus visits and extension summer research programs. What do you plan to do during the next reporting period to accomplish the goals?The future work includes the following goals: The current directed energy recipe takes up to 20 seconds irradiation time to ensure 100% empirical lethality, which is time consuming comparing to other organic weeding methods. In order to address this issue, a new LED based recipe is under development and the goal is to reduce the treatment time to 5 seconds and reduce the energy consumption by 70%. The new recipe will also be a safer solution because the LED will generate much less heat. Currently, the towing robot and the directed energy distributed array (DA)can be remotely controlled, with limited automated functionalities, including taking synchronous images based on their driving speed and running the neural networks. However, the weed classification and weed treatment are not happening simultaneously. The tractor or robot can wiggle when driving, which can cause mismatching between expected weed locations and actual weed locations. Although this can be rectified by mounting sonar sensor, accelerometer or gyroscope to track deviations and perform course correction, a simpler solution is to mount the cameras and reflectors close to each other, which requires redesigning the structure of the DA. The directed energy sources are being refined to minimize energy usage. With each refinement and wavelength change, the effectiveness of the light source on killing weeds needs to be tested and compared to other methods of organic weed control such as flaming and mechanical tilling. Weed seed control experiments using directed energy is a new push that will be continued to demonstrate that directed energy may be used in this novel method.
Impacts
What was accomplished under these goals?
Overall Summary of Accomplishments: A comprehensive soybean database that contains over 100,000 images is available. New databases of sweet corn and hemp were initiated Neural networks have been tested on a new dataset of 3,000 images of soybean seedlings using 12 new neural networks. Weed identification application is being tested for accuracy. 6 undergraduate students in agricultural technology research. A robotic system platform built for testing has been refined toward a prototype autonomous weed suppression system. 1 publication was submitted and 5 presentations were made by researchers and undergraduate researchers on technology weed control. Goal I Accomplishments: The goal is to develop an automated organic weed control system using directed energy and deep learning targeted for small farms in conjunction with Global Neighbor Inc. Design focused on moving from a hitch platform to combining all components to a robotic platform. Directed energy arrays and a camera detection system were incorporated to a remote controlled four-wheel drive robotic platform. The new platform allows field applications to occur remotely. We used transfer learning to fine-tune a number of pre-trained neural networks and develop customized neural networks to distinguish various weeds from crops in anticipation of moving an automated weed control device into the field. The system is implemented using two languages: MATLAB and Python (with Tensorflow library), and is tested on three platforms: PC, Android Device and Raspberry PI. The original pull-behind device to kill weeds using directed energy was adapted to an automated towing robot (Agi 1) was customized and tested for field use. It successfully identified crop rows. A second pull-behind device was developed for Edwards Air Force Base. The donation of this second device for field testing has been initiated. Various solutions for total autonomous field driving of the robot Agi 1 are being tested. A virtual environment to safely test-drive protocols, as well as ultra-sonic safety measures/obstacle avoidance are in process. Small robotic cars were purchased to safely test multiple self-driving protocols rather than use the larger and heavier field robot Agi 1 prototype. The RGB camera system for an automated robot weed killing system using directed energy was developed and improved camera interface with the following changes: Resigned the picture recording program to make it user friendly and to be used with any computer (assuming proper software is installed). Improved picture recording program to make use of multi thread to eliminate desynchronization between cameras. Improved the base framework's readability, efficiency, and overall structure of the program to make future modifications much easier. Adjusted frame rate calculation to ensure it runs exactly at the desired frame rate. developed a new process for taking pictures that will decrease the number of cameras required saving up to $900 in equipment New neural networks have been customized and tested on a new dataset that contains 3,000 images selected and cropped from the 2017-2019 database. We tested 12 new neural networks including alexnet, densenet201, resnet101, nasnetmobile, squeezenet, mobilenetv2, aception, googlenet, resnet18, inceptionresnetv2, inceptionv3, resnet50 and shufflenet. The system was evaluated on a highly diverse dataset that consists of images taken from a real soybean field. In the dataset, 9 weed species against soybean seedlings in five stages were tested and various conditions, including weather, humidity, year, shadow and dirt, were considered. The results showed that up to 98% weed classification accuracy, which is comparable to the current state-of-the-art plant classification accuracy. Goal II Accomplishments: Plots of soybean and sweet corn (approximately 0.25 acres each), 36 inch rows, were planted in April, June and July, 2019 to provide: 1) a continual source of crop seedlings for database development to train neural networks to distinguish soybean and corn seedlings from common weeds in Ohio; and 2) a ready source of newly germinated weeds to test efficacy of directed energy weed suppression. A comprehensive database that contains over 100,000 soybean images across 2017 - 2019 is now available. A database of sweet corn seedlings was initiated in 2018-2019. A database of hemp seedlings was initiated in 2018-2019. A weed identification application developed in 2017-2018 available on iOS and Android Platform continued to be tested for accuracy. Two different directed energy devices were tested on common field weeds. The first device delivered a maximum of 4 w/m2 visible/IR and 932 mw/cm2 UVA. The second prototype device delivered 0.66 w/m2 of 440 nm visible and 0.04w/m2 squared IR. The first device showed almost 100% weeds killed with no re-growth, while the other showed 56% weed kill without re-growth. With both tests of 100 weeds each, there was no account for weed age and maturity. The younger the weed, the better the control. This suggests that directed energy is best used as a pre-emergent and prior to canopy close. Goal III Accomplishments: Cereal rye was planted fall 2019 to be evaluated spring 2020 for its ability to suppress weeds in combination with crimping and directed energy replacing glyphosate treatment. The cover crop for 2019 was insufficient for testing due to weather conditions. Goal IV Accomplishments: Current energy sources are mounted gas generators and batteries. Energy consumption is still a major source of concern for deploying into a field. Researchers are working to design lighter autonomous robot platforms still rugged enough to withstand field use. Prototype directed energy systems using LED light sources and specific wavelengths has dramatically decreased energy use. Once a prototype has been successfully deployed, cost estimates of production will follow. Energy requirements are being balanced with cost, speed and effectiveness as the design moves forward to field testing. Goal V Accomplishments: Directed is a systemic weed control that will control vegetation and kill young weeds. The original use of directed energy was to develop weed killing capacity for the Air Force to clear bomb sites, roadways and the like of brush. This technology currently is being added to an autonomous robot to kill weeds in a crop field as part of an integrated pest management study. Tests of this prototype should begin in 2020. Additional uses have shown that directed energy, in conjunction with repeated cutting with or without black plastic mulch, may control knotweed species such as Japanese knotweed in greenhouse studies. Directed energy has been adapted in a new way to control weeds. In fields with non- or low-shattering weeds found in Ohio such as waterhemp, amaranth species, poison hemlock, burdock, evening primrose, and giant goldenrod, directed energy may be used to prevent weed seed germination in the soil seed bank. Initial trials on chaff from fields infested with waterhemp show up to 100% prevention of germination after a 30 sec dose of directed energy high in infrared light as tested on 1500 seeds. Experiments are ongoing on other broadleaf weed seeds collected at the end of the growing season 2019. Delivery systems are being designed to hit weed seeds in no-till fields to suppress weed seed growth the following season and minimize viable weed seeds in the seedbank. Another, novel control method for noxious knotweed species is to use the weed for human use. Knotweed species in Greene County, OH identified as Bavarian knotweed, have been shown to contain resveratrol, a heart healthy supplement for humans. The current source of resveratrol is Southeast Asia. Studies have been initiated to extract resveratrol from local sources of knotweed as a possible alternative natural source for resveratrol.
Publications
- Type:
Journal Articles
Status:
Submitted
Year Published:
2020
Citation:
Deng Cao, Austin Erdman, Cadance Lowell, Augustus Morris. An Organic Weed Control Prototype with Directed Energy and Depp Learning. Submitted to PLoS One Journal.
- Type:
Other
Status:
Accepted
Year Published:
2020
Citation:
Deng Cao, Cadance Lowell, Undergraduate Research: Deep Learning Based Plant Classifiers and Their Real- Life Research Applications, abstract accepted in ASEE annual conference 2020
- Type:
Other
Status:
Accepted
Year Published:
2019
Citation:
Deng Cao and Cadance Lowell. Building your Agricultural Network Using Neural Networks, Oral presentation, Outreach Education and Technical Assistance Farmers Conference (O.E.T.A.), April 6, 2019.
- Type:
Other
Status:
Accepted
Year Published:
2019
Citation:
Jeffrey Taylor and Dr. Deng Cao. A Real Time Weed Classifier Using Convolutional Neural Networks, Poster Presentation, Association of 1890 Research Directors (ARD) Symposium, Jacksonville, FL, March 30 to April 3, 2019.
- Type:
Other
Status:
Accepted
Year Published:
2019
Citation:
Deng Cao, Cadance Lowell, Central State University 1890 Land-Grant Research & Extension Stakeholder Event, Ohio Statehouse Atrium, Columbus, OH, March 19, 2019.
- Type:
Other
Status:
Accepted
Year Published:
2019
Citation:
Akil Cornish, Cadance Lowell, Marcus Nagle. Using Directed Energy for Natural Control of Japanese Knotweed. Poster Presentation, Association of 1890 Research Directors (ARD) Symposium, Jacksonville, FL, March 30 to April 3, 2019.
|
Progress 10/01/17 to 09/30/18
Outputs
Target Audience:This agricultural technology integrated pest management research will impact the following audience: Individual and Institutional Researchers (university or industry): to benefit from access to affordable technical assistance for idea generation and modeling, proto-type development, independent product assessment, continuing research recommendations, and specialized facilities. Industry Partners: to benefit from accelerating product integration and market access while reducing risks and costs. Growers (farmers, community gardeners, land managers, others): to benefit from an affordable, proven non-chemical technology for weed mitigation in agriculture applications. CSU Faculty and Research Faculty: improve partnerships with CSU faculty, CSU research faculty, industry experts and business partners; create better opportunities for new researchers and scholarly publications; and increase the relevance of hands-on learning for their students. CSU Undergraduate Students: to increase the relevance of hands-on learning and to improve access to future employment or graduate programs through mentoring, professional internships, conferences, and publications. Prospective CSU Undergraduate Students: to increase attraction to CSU programs, with an emphasis on new courses of study being developed as part of the 1890 Land Grant University system. Changes/Problems: While there are dedicated technicians and summer researchers available for this project, this research project needs dedicated research faculty in the areas of agricultural technology and robotics, computer visualization and integrated pest management with emphasis on agroecology sustainability and conservation. Progress would go faster with this expertise and FTE allocation. Investing in the infrastructure necessary to prepare, plant and treat research plots has somewhat slowed progress in planting corn and soybean research plots. There currently is no machinery available on campus to harvest small plots (1 acre or less). Yields have been estimated and crops have been plowed under. Partnering with outside industry has helped provide additional expertise for field work. Protocols for the use of animals in testing naturally extracted resveratrol have been developed and are pending campus approval. Extremely wet weather hampered efforts to plant soybean research plots. Beans were not planted until late June. More mature weeds were harder to control, even with glyphosate. The rain-caused time delay made it more difficult to collect meaningful data to compare to last year. What opportunities for training and professional development has the project provided?Undergraduate students participated in this research and develop skills in agriculture technology and weed science. Each student had a specific research project. In addition, each student worked on the overall goals of the research team. Research projects included: Computer smart vision in weed identification. Undergraduate student helped developed and trained software to identify soybean and corn plants. Global Informational System identification of a noxious weed in Greene County. Undergraduate student tested the possibility of using GIS techniques to monitor growth of Japanese Knotweed in three locations in Greene County. Student also tested various techniques to control weeds as part of an integrated pest management strategy. Verification of weed control. Undergraduate student designed and implemented experiments to determine the effect of directed energy on root respiration in dandelions. Hydroponic systems design. Undergraduate student researched and designed small scale hydroponic systems for the campus greenhouse for future systems and production experimentation. Natural Products in Weeds. Undergraduate student initiated research for the extraction of resveratrol from Japanese Knotweed and subsequent testing to determine heart health potential in rats. How have the results been disseminated to communities of interest?Results have been disseminated to educators through a summer workshop and to professionals at a professional ASEE conference. What do you plan to do during the next reporting period to accomplish the goals?The future work includes the following objectives: Improve the power efficiency of the directed energy recipe: The current directed energy recipe takes up to 20 second irradiation time to ensure 100% empirical lethality, which is time consuming comparing to other organic weeding methods. In order to address this issue, a new LED based recipe is under development and the goal is to reduce the treatment time to 5 seconds and reduce the energy consumption by 70%. The new recipe will also be a safer solution because the LED will generate much less heat. Making towing robot fully autonomous: Our final goal is that the weeding can be done regularly while the owner is on vacation. For a small and slow-moving vehicle that only works in the crop field, this goal can be achieved by equipping the robot with 2D simultaneous localization and mapping (SLAM) system and object detection-based accident avoidance system. Hardware implementation is projected to be completed by January 2019. A form of the classification system should be integrated on the new platform before the end of Spring semester 2019. Other methods of non-chemical weed control are being researched for efficiency and viability. It is hoped the existing robotic platform can be quickly re-engineered to accommodate other modalities as the research develops.
Impacts
What was accomplished under these goals?
Impact Statement : There at least 112 weed species in Ohio, including 21 on the noxious weed list. Some of these weeds are herbicide resistant and all weeds reduce crop productivity. This project has generated information for Ohio farmers, K-12 instructors and students, and the general public regarding non-chemical methods for weed control including directed energy, cover crops and cover crop crimping. Knowledge has been transmitted through participation in professional engineering conference, outreach to Ohio Agricultural Educators, outreach to high school students on campus visits and extension summer research programs. 5 undergraduate minority studentss were trained in research through summer internships in computer science visualization, weed science and mechanical engineering. Major goals of the project: (I) Develop integrated pest management protocols to use directed energy and cover crops for weed management in soybeans(II)Computer visualization for field deployed robotics, (III) Investigate crimping cover crops for weed control (IV)nvestigate feasibility of using directed energy to kill Japanese knotweed in natural settings. (V) Design and build small-scale hydroponic systems for consumers to evaluate cost and ease of use. 1) Major activities completed - Goal I : Ten-foot soybean test plots in a random block design were established to measure effectiveness of directed light energy to control weeds. Round-up ready beans were planted in 7.5" rows and light energy was compared to mechanical till and glyphosate application. All treatments were applied twice. Effectiveness was measured by percent weed ground coverage and estimated yield. Cereal rye was tested as a cover crop in conjunction with soybeans. Cereal rye was planted fall 2017 in conjunction with cereal rye tests in the greenhouse to demonstrate that a combination of crimping and directed energy may replace chemical spray to kill the cover crop at the time of planting. While crimping technology is common in areas around the country, it is not frequently used in Ohio. A crimper for small test plots was p modified and fit to an IPO. Goal II : To start taking pictures of corn plants for computer visualization to train software to identify corn plants, students planted 8 varieties of sweet corn. Over 3000 pictures were taken for software development. A side-by-side sweet corn variety comparison was initiated with measurements of height, yield, incidence of corn fungus and nutrient content. Goal III :Cereal rye was planted fall 2017 and evaluated spring 2018 for its ability to suppress weeds in combination with directed energy. The crimped cover crop was drilled with round-up ready soybeans. A preliminary side by side comparison of crimping + a new light system designed by a local industry partner, crimping alone and crimping + glyphosate to evaluate both suppression of the cover crop and weeds showed that crimping does knock down the cover crop without herbicides. Goal IV : A greenhouse study on Japanese knotweed was initiated and completed. Preliminary results suggest that directed energy and black plastic mulch can suppress growth of this noxious weed. Goal V : The manufacturing capability for the production of different hydroponic systems were researched, designed and built in small-scale to maximize promising designs for hydroponic systems in the CSU Greenhouse. The purpose of this was to design and create scaled hydroponic systems that are cheap to make, scalable, and have potential to produce a lot of food. 2) Data collected- Goal I: Directed energy as a weed control method for soybeans was compared to mechanical till and glyphosate control. To verify results, an algorithm was developed to used drone pictures taken of the field at approximately 100 ft that measure the percent "green" or plant coverage per plot area. Results were an average of a minimum of 12 trials. Goal II: A control system is being developed for field-deployed weed control for corn and soybean crops using deep learning based image classification and object detection algorithms. A transfer learning algorithm is being developed to distinguish various weeds from crops. The algorithm will be implemented using two languages: MATLAB and Python (with Tensorflow library), and it will be tested on three platforms: PC, Android Device and Raspberry PI. The algorithm utilized pre-trained convolutional neural networks such as GoogleNet (Inception), Inception-v3, and MobileNet. Goal III :Cereal rye was planted on a 0.25 acre plot to evaluate crimping and directed energy as a weed control method. Preliminary information were collected on cover crop seeding density and weed suppression. Goal IV : Greenhouse grown plants were separated into one trial of 3 pots each and either not treated (control), treated with 30 sec or 60 sec of light and treated with light and covered with black plastic. Goal V : Four small scale hydroponic systems were designed and built and tested for the ability to run water without leaking. 3)Summary statistics and discussion of results -Goal I : The major objective was to integrate agriculture technology into non-chemical weed suppression. The secondary objective was to integrate directed energy arrays and a camera detection system to a remotely controlled four-wheel drive robotic platform. The new platform would allow field applications to occur remotely and be a step closer toward the development of an autonomous weed control system. Design focused on moving from a hitch platform to combining all components to a robotic platform. Energy requirements are being balanced with cost, speed and effectiveness as the design moves forward to field testing. Goal II : A comprehensive database that contains over 20,000 images across 2017 - 2018 is available. A weed identification app is now available on iOS and Android Platform (for test purpose only). A cost-effective weed control system tmainly designed for small organic farms has been developed and is in testing for soybeans. Computer vision is used for weed-crop classification and directed energy for organic weed control. The system was evaluated in-field with up to 98% accuracy, which is comparable to the current state-of-the-art image classification accuracy in general. Goal III :Preliminary data suggested that the crimped cover crop does appear to act as a physical barrier to decrease weeds. Crimping plus directed energy instead of herbicide cover crop suppression may be a good tool for farmers as a sustainable weed control practice. The experiment will be redesigned to insure that cover crop densities are increased to improve weed suppression. Goal IV : Data on Japanese Knotweed suppression in riparian corridors is ongoing. Biomass reductions were seen when the plants were cut, treated with light and covered with black plastic. A qualitative GoogleEarth study of three different locations of Japanese Knotweed in Greene County was initiated to determine when it was introduced to the county and to determine best practices on identifying incidences of the invasive weed in Greene County, OH. Goal V : Five hydroponic systems were designed and built using inexpensive materials. All systems are in the CSU greenhouse to be evaluated for ease of use and to fully automate at least one of the systems. 4) Key outcomes or other accomplishments realized A comprehensive database that contains over 20,000 images across 2017 - 2018 is available. A weed identification app is now available on iOS and Android Platform (for test purpose only). Methodology and algorithm to determine percent plant coverage of a field plot using aerial drone photographs. Mentored 5 undergraduate students in agricultural technology research. Five different, small-scale hydroponic systems were designed and built for green house monitoring and testing. A robotic system platform was designed and built for testing to a prototype autonomous weed suppression system for soybean field use.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2018
Citation:
Deng Cao, Cadance Lowell and Augustus Morris. Undergraduate Research: Introducing Deep Learning Based Image Classification and Object Detection to Undergraduate Students. In American Society for Engineering Education (ASEE) Annual Conference, 2018.
- Type:
Journal Articles
Status:
Submitted
Year Published:
2018
Citation:
Deng Cao, Austin Erdman, Jeffery Taylor, Cadance Lowell, Augustus Morris. An Organic Weed Control Prototype Based on Automated Weed Classification using Convolutional Neuron Networks. To be submitted to PLoS One Journal.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2018
Citation:
Lowell C, Henry J. 2018 Jobs + Billions with a Capital B! What is the Bioeconomy. Professional Development Ohio Agricultural Educators Annual Conference. June 14-17. Columbus, OH.
|
Progress 10/01/16 to 09/30/17
Outputs
Target Audience:The target audience included organic farmers and gardeners, scientists and undergraduate students. The directed energy technology is non-ionizing and therefore, organic. Directed energy is an integrated pest management strategy. Any farmer, conventional to organic, interested in methods to reduce herbicides is the target of this technology. Machinery built to house directed energy wth selective computer visualization to kill weeds focused on farmers in organic farm fields such as soybeans. Information was shared in a variety of venues that included organic farmers, Ohio state legislators, and fellow scientists. Impacted target audiences are listed below. Individual and Institutional Researchers (university or industry): to benefit from access to affordable technical assistance for idea generation and modeling, proto-type development, independent product assessment, continuing research recommendations, and specialized facilities. Industry Partners: to benefit from accelerating product integration and market access while reducing risks and costs. Growers (farmers, community gardeners, land managers, others): to benefit from an affordable, proven non-chemical technology for weed mitigation in agriculture applications. CSU Faculty and Research Faculty: to benefit from improved partnerships with CSU faculty, CSU research faculty, industry experts and business partners; better opportunities for new researchers and scholarly publications; and increased relevance of hands-on learning for students. CSU Undergraduate Students: to increase the relevance of hands-on learning and to improve access to future employment or graduate programs through mentoring, professional internships, conferences, and publications. Prospective CSU Undergraduate Students: to increase attraction to CSU programs, with an emphasis on new courses of study being developed as part of the 1890 Land Grant University system. Efforts The majority of directed energy machinery and computer visualization science-based knowledge was disseminated through both formal and informal workshops and demonstrations. Smaller versions of the equipment were demonstrated during outreach informational talks to farmers, gardeners, the general public and statehouse legislators. Four CSU undergraduate students had the opportunity to work with directed energy delivery equipment in field and greenhouse trials. Directed energy as an integrated pest management solution also was shared with prospective undergraduate students through high school visits and campus tours. Changes/Problems:We are having some problems with experimental design on measuring decreases in bacterial and fungal counts that will be addressed with the use of QtPCR and developing this capability on the CSU campus. Also while the purchase of tractors, and other farm equipment is ongoing, the lack of this equipment initially slowed down some field trials. A lack of a technician with the appropriate spray and operator licenses has been solved with the recent hiring of a farm technician. What opportunities for training and professional development has the project provided?Undergraduate students received training in basic laboratory and field research technique and help implement research projects in directed energy. All involved faculty, students and technicians attended CULTIVATE 17, a trade show, education series and networking event in Columbus, OH that brings together all interested in horticulture in Ohio to observe and interact with new products, ideas and innovation trends. All involved faculty, students and technicians attend Farm Science Review, London, OH that helped this audience learn about the latest trends in agricultural production in Ohio and the mid-west. How have the results been disseminated to communities of interest?The results have been disseminated through: 1) high school classroom demonstrations; 2) high school visitations at CSU; 3) a video developed to showcase all the ongoing NIFA sponsored research at CSU which is posted on the CSU website; 4) Ohio State Legislature Land Grant Day at the Ohio State House, Columbus, OH; 5) professional presentations at the American Society for Plant Biologists, and ARD Research Symposium, Atlanta, GA; and 3) mentoring to undergraduate students. What do you plan to do during the next reporting period to accomplish the goals?Future plans include presentations at community gatherings and professional conferences. Field trials are underway to test directed energy on suppressing cereal rye cover crop prior to drilling in the crop. Directed energy should work along with crimping to replace the need for herbicides and GMO crop seed. We plan to repeat field test plot trials using directed energy to suppress weeds in a corn field and develop weed and corn seedling recognition algorithms similar to those developed for soybean crops. We hope to continue and conclude greenhouse and field trials on using directed energy to kill Japanese Knotweed, an invasive plant species in Greene and Montgomery Counties, OH. We will complete preliminary trials to sterilize plants of bacterial and fungal infections. We plan to continue to develop automation technology to deliver directed energy and other weed control to crop fields without a tractor or operator.
Impacts
What was accomplished under these goals?
Major activities completed Computer Visualization - The computer scientist team is developing an automated organic weed control system using deep learning based image recognition and object detection algorithms. In particular, a transfer learning algorithm was developed to distinguish various weeds from crops. The algorithm was implemented using two languages: MATLAB and Python (with Tensorflow library), and it was tested on three platforms: PC, Android Device and Raspberry PI. The algorithm utilized pre-trained convolutional neural networks such as GoogleNet (Inception), Inception-v3, and MobileNet. We currently achieved above 98% accuracy in real life test. Details in this work are to be published. Robotics - As an outgrowth of a capacity building grant, robot parts and a drone were purchased for this Agriculture Technology project. The drone currently is being used for: 1) aerial monitoring of weed densities; and 2) promotional videos to showcase research to different audiences. The robot parts were assembled to pull up to 500 lbs. The goal is to automate a directed energy weed system that can go through a field, identify and kill weeds without an operator. Organic Weed Control -- Since the directed energy is non-ionizing radiation and non-chemical, it is organic. We established a 0.7 acre soybean test plots to compare directed energy with herbicide (glyphosate), mechanical till and no treatment. Directed energy was comparable with mechanical till and did not damage the soybean plants as did the herbicide treatment. A series of experiments also were established to gather preliminary data on: 1) weed control of Japanese Knotweed, an invasive species in local wetlands; 2) killing leaf bacteria without killing the host plant; 3) sterilizing wood; 4) evaluation of root damage following a directed energy treatment in dandelion; 5) killing a cover crop during crimping and drilling in the crop. These are ongoing projects. 2) Specific objectives met Specific objective met for Agriculture Technology are: Completed one-year soybean trial using directed energy to control weeds. Initiated multiple preliminary trials to kill bacteria in living and dead plant systems, controlling cover crops and researching any root damage to a weed following directed energy treatment. Making progress on integrating robotics and computer visualization to complete a working, automated machine to kill weeds in a field. Significant results achieved include the following: A transfer learning algorithm was developed and tested to distinguish various weeds from crops. It has up to 98% accuracy in real life test. Separate soybean and weed image databases were built with more than 30,000 raw images. Six different varieties of soybeans were used for the soybean database. Demonstrated that directed energy from a tractor pulled device has similar weed killing capacity to mechanical till compared to herbicide and no treatment controls. ) Key outcomes or other accomplishments realized Engaged six students in undergraduate student activities related to killing weeds with directed energy and engineering design. Presented 2 research posters and 1 invited talk at scientific conferences. Developed one manuscript
Publications
|
Progress 07/21/16 to 09/30/16
Outputs
Target Audience:This precision agriculture research has impacted the following audiences: Individual and Institutional Researchers (university or industry): benefit from access to affordable technical assistance for idea generation and modeling, proto-type development, independent product assessment, continuing research recommendations, and specialized facilities. Industry Partners: benefit from accelerating product integration and market access while reducing risks and costs. Growers (farmers, community gardeners, land managers, others): benefit from an affordable, proven non-chemical technology for weed mitigation in agriculture applications. CSU Faculty and Research Faculty: improves partnerships with CSU faculty, CSU research faculty, industry experts and business partners; create better opportunities for new researchers and scholarly publications; and increase the relevance of hands-on learning for their students. CSU Undergraduate Students: increases the relevance of hands-on learning and improves access to future employment or graduate programs through mentoring, professional internships, conferences, and publications. Prospective CSU Undergraduate Students: increases attraction to CSU programs, with an emphasis on new courses of study being developed as part of the 1890 Land Grant University system. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?The results of the progress made in this project have been disseminated to various audiences regional and National as outlined below: Lowell, C., Jackson P, Jackson J. 2016. CENTS 2016. Tradeshow of the Ohio Nursery & Landscape Association. Columbus, OH. January 11-13. Directed energy research was shared with public and growers. Lowell C, Erdman A, Jackson J. 2016. Open CV software iterative training for weed image recognition in residential and agricultural settings. Poster presentation to Weed Science Society of America Annual Conference, San Juan Puerto Rico, February 9, 2016 . Dr. Lowell shared research with other scientists at this national event. Hayes F, Lowell C, Jackson J. 2016. Directed energy ragweed control. Poster presentation to Weed Science Society of America Annual Conference, San Juan Puerto Rico, February 9, 2016. Undergraduate student shared research with other scientists at this national event. Lowell, C. 2016. Emerging Markets and Technology. 2016 Minority, Women, and Small Farmers Conference. Central State University Cooperative Extension Program, Wilberforce, OH. April 8-9. Dr. Lowell shared research with small farmers. Hayes F, Lowell C, Jackson J. 2016. Non-chemical weed control of ragweed. Ohio Academy of Science Annual Conference, Athens, OH, April 16. (Poster Presentation. Undergraduate student shared research with other scientists and students in Ohio. Gill-Pillow M, Lowell C, Jackson J. 2016. Directed energy dandelion control. Ohio Academy of Science Annual Conference, Athens, OH, April 16. (Poster Presentation. Undergraduate student shared research with other scientists and students in Ohio. Hudson T, Lowell C, Jackson J. 2016. Control of crabgrass by directed energy. Ohio Academy of Science Annual Conference, Athens, OH, April 16. (Poster Presentation). Undergraduate student shared research with other scientists and students in Ohio. What do you plan to do during the next reporting period to accomplish the goals?General: The tractor pulled directed energy distributed array (DA) has been field tested once on underbrush in a second growth forest. Final data will occur with evaluation of weed regrowth in spring 2017. Spectrophotometric changes in peroxidated lipids in ragweed leaves suggest membrane damage leading to plant necrosis. Additional testing will include measurement of chlorophyll loss, and changes in plant respiration will commence as soon as the purchased LI-6800 is delivered. Tests using the directed energy Distributor Array platform to agriculture and conservation efforts, and the Long Stand Off device for log and pallet sanitization have shown early promise. Specific Activities: The completion of search for the mecahtronics/ robotocs application engineer position for the grant and engaging him/her to lead the research. Complete hiring of the technician needed in the project Purchase of additional equipment needed as guided by the new research scientist Complete installation of the directed energy (DE)/LI 6800 unit in the tractor. Complete the experimental design for field tests. Prepare the field for planting in Spring 2017. Conduct field tests and assess the results. Disseminate the results in appropriate forum.
Impacts
What was accomplished under these goals?
Objective I - Develop novel field-ready, scalable, integrated pest management strategies for agriculture, forestry, and other land management needs. Two directed energy devices were developed and manufactured by business partner Global Neighbor, Inc. that were purchased and are being adapted for use at CSU. A field ready, directed energy device will be field tested on soybeans beginning spring 2017. A second directed energy device was purchased to deliver up to 180X sun power to kill weeds up to 50 feet. This device is being adapted for use to sanitize lumber and kill insect and bacteria pests. Objective II -- Expand field testing and range of plants/applications for current directed energy mechanized weed control systems already being tested at CSU. Two undergraduate students have been employed to work on the research. Job advertisements have been posted to hire a mechatronics engineer and two technicians, one in mechatronics and one in computer vision, to help expand the research into precision agriculture at CSU. Soybean field test plots have been established at the CSU farm and field testing will commence spring 2017. A tractor, plow, 2 roto-tillers, sprayer and seeder were purchased as well as UV protective gear and laboratory equipment and sensors for testing purposes. Small scale testing in the CSU greenhouse and newly developed Precision Agriculture Laboratory, show that directed energy is effectively kills dandelion, crab grass and ragweed. Testing also has been initiated on the riparian corridor invasive species, Japanese Knotweed. Objective III -- Demonstrate the economic benefit of these non-chemical alternatives versus traditional chemical weed control. New technology has been demonstrated to stakeholders at three small farmer conferences and to general public at one demonstration. A Business-University partnership has developed as a result of this research. A model has been initiated to develop pathways for transition of precision agriculture technology between the university and small, emerging businesses. Objective IV -- Develop, test, and apply business models that take advantage of these benefits to small/specialty farms (existing and new farmers, producers). The new, precision agriculture technology has been shared with stakeholders. Business models have been initiated with one small, woman owned company, Global Neighbor, Inc. Objective V -- Adapt the underlying platform to new applications, informed by research directions and potential new uses identified from CSU's existing (capacity building) research. Research has expanded from primarily residential to agriculture, forestry and natural areas to provide non-chemical options to kill weeds. General Accomplishments. One research faculty mechatronic engineer has been advertised and CSU with interviews to begin in January 2017.One computer vision technician has been approved to hire with an offer forthcoming.One mechatronics engineer technician has been advertised with interviews to begin in January 2017.Two undergraduate students have been hired. Permanent capacity and capability at CSU for future research, education, and extension: The CSU greenhouse now hosts the Precision Agriculture and Mechatronics Laboratory. CSU has also updated technology and enhanced earlier capacity supporting smaller scale, single-lamp (hand held) student-supported projects under faculty supervision. Purchased were a tractor, plow, sprayer, 2 rototillers, commercial field directed energy device (DA), LI-6800, and large capacity directed energy device, the LS-8 that can kill weeds up to 50 feet, small support sensors and safety equipment, and small laboratory equipment.
Publications
- Type:
Other
Status:
Published
Year Published:
2016
Citation:
Lowell, C., Jackson P, Jackson J. 2016. CENTS 2016. Tradeshow of the Ohio Nursery & Landscape Association.
Columbus, OH. January 11-13.
- Type:
Other
Status:
Published
Year Published:
2016
Citation:
Lowell C, Erdman A, Jackson J. 2016. Open CV software iterative training for weed image recognition in residential and agricultural settings. Poster presentation to Weed Science Society of America Annual Conference, February 9, 2016.
- Type:
Other
Status:
Published
Year Published:
2016
Citation:
Hayes F, Lowell C, Jackson J. 2016. Directed energy ragweed control. Poster presentation to Weed Science Society of America Annual Conference, February 9, 2016.
- Type:
Other
Status:
Published
Year Published:
2016
Citation:
Hayes F, Lowell C, Jackson J. 2016. Non-chemical weed control of ragweed. Ohio Academy of Science Annual Conference, Athens, OH, April 16.
- Type:
Other
Status:
Published
Year Published:
2016
Citation:
Gill-Pillow M, Lowell C, Jackson J. 2016. Directed energy dandelion control. Ohio Academy of Science Annual Conference, Athens, OH, April 16.
- Type:
Other
Status:
Other
Year Published:
2016
Citation:
Hudson T, Lowell C, Jackson J. 2016. Control of crabgrass by directed energy. Ohio Academy of Science Annual Conference, Athens, OH, April 16
- Type:
Other
Status:
Published
Year Published:
2016
Citation:
Salchak P., Lowell C. 2016. Sustainable directed energy technology for small farmers and landowners. Propolis Day, WholeFoods Market, Centerville, OH. June 18
|
|