Progress 07/01/20 to 06/30/24
Outputs
Target Audience:1. Undergraduate Students: Engaged through seminars conducted once every three years. 2. Graduate Students: Actively involved in the project, with three students graduating with MSc degrees on the project topic. Another non-thesis graduate student gained experience in data collection from the field experiments. 3. Faculty and Staff in the Department and on Campus: Students presented their work to faculty and peers through oral presentations within the department. At the campus level, students participated in oral presentations at college symposiums and poster presentations at various campus events. Demonstrated the ground rover and the mechanisms developed for crop height and density measurements. 4. Broader Academic and Professional Community: Graduate students and faculty presented their findings at international meetings in the US and Europe. Research findings were published in peer-reviewed journals and conference proceedings contributing to the scientific community. 5. Extension Agents and Farmers: Demonstrated the developed systems to extension agents and farmers. Delivered oral presentations and training sessions at field days and CAMM (Confined Animal Manure Managers Program) training by the PI and Co-PI to educate and train farmers and extension agents. Efforts: 1. Seminars for Undergraduate Students: Conducted seminars to provide insights and knowledge related to the project. 2. Active Involvement of Graduate Students: Engaged graduate students in hands-on project work, leading to MSc degrees and practical experience in data collection. 3. Presentations to Faculty and Staff: Facilitated oral presentations and poster sessions to share research findings within the department and at campus events. 4. International Conferences: Enabled graduate students and faculty to present their research at international conferences, broadening the reach of the project's findings. 5. Peer-Reviewed journal articles and Conference Papers: Published research findings in peer-reviewed journals and conferences to contribute to the scientific community and ensure the dissemination of knowledge. 6. Demonstrations and Training for Extension Agents and Farmers: Conducted demonstrations of the developed systems and provided training sessions to extension agents and farmers to ensure practical application of the project's outcomes. Changes/Problems:We encountered challenges with the Bermudagrass plots during thefirst couple of years of the project and faced delays in outreach and training activities due to COVID-19. However, a one-year project extension allowed us to collect more data on Alfalfa and Bermudagrass in pastures. This extension also provided opportunities to present and demonstrate UAV and ground-based forage biomass estimation techniques to producers and researchers in the field. Additionally, we used the extended year to present posters and deliver oral presentations at ASABE and campus-level conferences. We also submitted our research findings to peer-reviewed journals. What opportunities for training and professional development has the project provided?Three MSc students,one PhD student, and three undergraduate students were mentored by PI and Co-PIs. The mentorship covered aerial and ground-based measurements, data analysis, and interpretation of results. The studentsalso received hands-on training in operating UAVs, ground rovers, and various measurement systems including RGB, NDVI and depth cameras, development of compressed crop height and crop density estimation mechanisms including US-Ski, IMU-Ski, and IMU-roller. This training included practical sessions on equipment setup, planning for field data collection, troubleshooting and project management. Team members attended local, national, and international conferences to present their findings through poster and oral presentations, facilitating networking with other professionals in the field. They also had opportunities to demonstrate UAV and ground rover applications to producers and extension agents. Graduate students presented their work to freshmen, juniors, and seniors in Agricultural Mechanization and Business, andin Animal and Veterinary Sciences. Three graduate students gained valuable research experience and successfully submitted their thesis work, sharing their findings with faculty and staff in both programs. How have the results been disseminated to communities of interest?Research findings were published in peer-reviewed journals and presented at conferences, contributing to the scientific community and ensuring the dissemination of knowledge. The methods and research findings were presented at ASABE Annual International Meetings in 2021, 2022, 2023 and 2024 as oral and poster presentations, and 15th International Congress on Agricultural Mechanization and Energy in Agriculture (ANKAgEng 2023,Antalya-Turkiye). We participated in outreach activities which included field days at Clemson University Piedmont Research and Education Center where we demonstrated UAV and ground-based biomass estimation methods to approximately 40 producers and extension agents. We also presented and demonstrated the ground rover at the Clemson University Center for Agricultural Technology (CU-CAT) symposium three times where we showcased the ground-based biomass estimation methods and findings to students, faculty and agriculture industry representatives. We provided training on "Improving forage production and utilization using drones and rovers" to more than 40 producers at Hollywood Ruritan Building, Saluda, SC. This training was part of the Clemson University Confined Animal Manure Manager (CAMM) Recertification Trainings program. The participants gained credits towards their CAMM-certified growers' recertification. What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
For Objective 1 - Major Activities Completed / Experiments Conducted: We established research plots for Alfalfa, Tall Fescue, and Bermudagrass to conduct controlled experiments and collect data for crop surface and digital elevation models, as well as vegetation index maps. We acquired UAVs with RGB and NDVI cameras, set up concrete pavers, and established ground control points (GCPs) with RTK-GPS coordinates. Aerial images were captured every 10, 20, and 30 days, accompanied by ground-based measurements. After pre-harvest data collection, plots were manually harvested, and total wet biomass was recorded. Small biomass samples were sent to the lab for analysis. While Alfalfa and Tall Fescue performed well over the first two years, the Bermudagrass plots were unsuitable due to dry weather. Aerial images were processed using Agisoft Photoscan, allowing us to develop crop surface models and estimate crop heights. NDVI images were used to evaluate crop moisture and density. Data analysis with ESRI ArcGIS and zonal statistics facilitated correlation of estimated crop values with actual biomass, leading to prediction equations for Alfalfa and Tall Fescue. In the third and fourth years, we scaled measurement methods to larger pasture fields. For Objective 2 -Major Activities Completed / Experiments Conducted: We developed a ground rover for efficient data collection, designed with high clearance and an adjustable wheelbase. The four-wheel-drive rover was teleoperated and, in the project's third year, equipped with a solar panel to power its electronic components and sensors. Additionally, we created four ground-based crop-height measurement systems: 1)Ultrasound Sensor and Compression Ski (US-Ski): An ultrasound sensor was mounted on the rover, facing downward at a fixed height. An acrylic compression ski was attached below the sensor, allowing it to glide over the crop canopy. The sensor measured the distance from the ski to the ground, with the difference from its mounting height indicating crop canopy height. 2)IMU Sensor and Compression Ski (IMU-Ski): This system measured crop height by monitoring the orientation of an inertial measurement unit (IMU) sensor attached to a compression ski. As the ski slid over the canopy, changes in the connecting rod's angle translated into crop height measurements. 3)IMU Sensor and Roller (IMU-Roller): A modified version of the IMU-Ski, this system used IMUs mounted on connecting rods attached to a roller. The roller's design allowed it to glide over the crop canopy more easily, reducing resistance. An additional IMU was planned for enhanced data capture. 4)Depth Camera (DC): This system utilized stereovision with two monochromatic cameras to measure crop height. The disparity between the images provided the distance from the camera to the crop canopy, which was converted to height by subtracting the camera's mounting height. For Objective 3 -Major Activities Completed / Experiments Conducted: Weutilized aerial images collected via UAVs and processed orthomosaics to identify plot boundaries and reduce ground-based data. This allowed us to determine average crop height and density for each harvested plot. Data reduction was performed using ESRI ArcGIS software. Aerial NDVI measurements informed both wet and dry biomass estimation models, correlating crop density with actual biomass moisture content. Small samples collected during harvesting were analyzed for biomass moisture content and forage nutrients. NIR spectrometer measurements were conducted on dry biomass samples, and machine learning algorithms were applied for rapid assessment of key forage quality indicators.Data collected: The data we collected included high-resolution aerial images (RGB, NDVI) and ground measurements using sensors including RTK-GPS values, ultrasonic and IMU based crop height measurements, moisture analyses, major forage crop quality indicators using wet chemistry and NIR spectroscopic measurements. The processing of vegetation indices revealed varying levels of crop moisture and crop density. Wecreated prediction equations using estimated crop heights, crop densities and vegetation coverage values. Wealso collected weather data from a weather station set up near the research plots.Summary Statistics and discussion of results: Prediction models were developed using pre-harvest crop height (TCH), change in crop height (deltaH), canopy density (CD), maximum temperature (Tmax), and growing degree days (GDD) as independent variables. The US-Ski system showed significant correlations between TCH, deltaH, and wet biomass yield for alfalfa and tall fescue. For UAV-based structure-from-motion, deltaH was the best predictor for wet biomass. Additionally, Tmax and CD were used to predict dry matter fraction with satisfactory results. In pasture experiments, one-acre plots of Alfalfa and Bermudagrass were marked for crop height and vegetation coverage data collection. New systems (IMU-Ski, IMU-Roller, and DC) were developed alongside the UAV-SfM and US-Ski for crop height measurement. The data collection prioritized systems based on their impact on standing crops, starting with UAV-based measurements followed by UGV systems in order of weight. Vegetation coverage images were captured using an RGB camera during crop height data collection. UAV measurements were exported as DEM raster files, while UGV measurements were processed as point files using plot boundaries as masking layers. All systems demonstrated satisfactory results for alfalfa and bermudagrass, with UAV-SfM and IMU-Ski performing best for alfalfa, and IMU-Ski and DC excelling for bermudagrass. Overall, UAV-SfM, IMU-Ski, and DC systems outperformed others, with the lowest standard errors observed in pasture scale experiments.Key outcomes: We developed a ground rover and advanced measurement systems for crop data collection. We successfully used UAV aerial imaging to define plot boundaries and create crop surface and elevation models. Reliable wet and dry biomass estimation models were established for Alfalfa, Tall Fescue, and Bermudagrass. We demonstrated our systems to producers and students, scaled methodologies from small plots to larger fields, and shared findings at various conferences which contribute valuable insights to the agricultural research community.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2024
Citation:
Singh, J.; Koc, A.B.; Aguerre, M.J.; Chastain, J.P.; Shaik, S (2024). Estimating Bermudagrass Aboveground Biomass Using Stereovision and Vegetation Coverage. Remote Sens. 2024, 16, 2646. https://doi.org/10.3390/rs16142646
- Type:
Books
Status:
Published
Year Published:
2024
Citation:
Koc, A.B., Erwin, C., Aguerre, M.J., Chastain, J.P. (2024). Estimating Tall Fescue and Alfalfa Forage Biomass Using an Unmanned Ground Vehicle. In proceedings of 15th International Congress on Agricultural Mechanization and Energy in Agriculture. ANKAgEng 2023. Lecture Notes in Civil Engineering, vol 458. Springer, Cham. https://doi.org/10.1007/978-3-031-51579-8_32
- Type:
Journal Articles
Status:
Published
Year Published:
2024
Citation:
Singh, J.; Koc, A. B.; Aguerre, M. J.; Chastain, J. P. (2024). Estimation of aboveground biomass of Alfalfa using field robotics. Smart Agricultural Technology journal, Vol. 9 (2024). https://doi.org/10.1016/j.atech.2024.100597.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Singh, J.; Koc, A.B.; Aguerre, M.J. (2023). Aboveground Biomass Estimation of Tall Fescue using Aerial and Ground-based Systems. In Proceedings of the 2023 ASABE Annual International Meeting, Omaha, NE, USA. July 912, 2023.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2024
Citation:
Singh, J., Koc, A. B., Aguerre, M. J., Chastain, J. P., Shaik, S. (2024). Stereoscopic Morphometry in Forages: Predicting Pasture Quantity with Field Robotics. American Society of Agricultural and Biological Engineers Annual International Meeting at Anaheim, CA. July 28-31, 2024.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2024
Citation:
Shaik, S., Koc A. B., Singh J., Aguerre M. J., Chastain J. P. (2024). Aboveground Biomass Prediction of Bermudagrass: A Comparative Analysis of Machine Learning Models. 2024 AI in Agriculture and Natural Resources Conference at College Station, TX. April 15-17, 2024.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Singh, J., Koc, A. B. (2023). Aboveground biomass estimation of Tall Fescue using field robotics. 2023 CAFLS Graduate Research Symposium at Baruch Institute of Coastal Ecology & Forest Science, Georgetown, SC. October 16-17, 2023.
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2024
Citation:
Singh, J. (2024). Aboveground Biomass Estimation of Pastures Using Field Robotics, All Theses. 4347. https://open.clemson.edu/all_theses/4347
- Type:
Journal Articles
Status:
Under Review
Year Published:
2024
Citation:
Singh, J., Koc, A.B., Aguerre, M.J., Chastain, J.P., and Shaik, S. 2024. Biomass estimation of bermudagrass in pastures: a comparison of sensor-based field robotic systems.
- Type:
Other
Status:
Published
Year Published:
2024
Citation:
Singh, J., Koc, A. B., Aguerre, M. J., Chastain, J. P. (2024). Aboveground biomass estimation of alfalfa using field robotic systems. Clemson University- Center for Agricultural Technologies (CU-CAT) event at Watt Center, Clemon University, Clemson, SC. 17 September 2024.
- Type:
Other
Status:
Published
Year Published:
2024
Citation:
Singh, J., Koc, A. B., Aguerre, M. J., Chastain, J. P. (2024). Estimation of aboveground biomass in bermudagrass using field robotics. 2024 CAFLS Graduate Research Symposium at Clemson University-Piedmont REC, Pendleton, SC. August 19-20, 2024.
|
Progress 07/01/22 to 06/30/23
Outputs
Target Audience:1) Researchers in Agricultural Engineering, Agricultural Sciences, and Agribusiness 2) Forage and Livestock Producers, including Extension Agents 3) Graduate and Undergraduate Students, and Faculty Members These identified target audiences were engaged through various means, including publications, presentations, field days, and training sessions. For specific details about the efforts undertaken to reach these audiences, please refer to the corresponding sections in the report. Changes/Problems:A 12-month no-cost extension for the project was granted due to the following reasons: 1) Poor growth in bermudagrass plots during prolonged dry periods. 2) Covid-19 restrictions affecting planned extension activities for the first two years. This extension allows us more time to gather additional field data and evaluate UAV and UGV-based biomass estimation methods, especially for bermudagrass and other crops. It also facilitates data collection for tall fescue and alfalfa fields and enables us to participate in extension activities and conferences to share our research findings. What opportunities for training and professional development has the project provided?We provided training and professional development opportunities through the following activities. A field day where we demonstrated UAV and ground-based biomass estimation methods to approximately 40 producers and extension agents. An event organized by Clemson University Center for Agricultural Technology, where we showcased ground-based biomass estimation methods to students and faculty. Presentations of our results at international and national conferences, benefiting researchers, students, and producers. Graduation of one MSc student whose work was related to this project. Hands-on experience for one graduate and one undergraduate student in field data collection using UAV and ground rover. Ongoing project involvement for another graduate student working toward an MSc degree. How have the results been disseminated to communities of interest?We disseminated the research results using professional conferences, publications, field days and training Sessions. We have presented our research findings at both national and international conferences relevant to our field. These conferences provide a platform for disseminating our results to a wide audience of researchers, professionals, and stakeholders. Our project results have been published in peer-reviewed journals and academic publications. This dissemination method ensures that our findings are accessible to researchers and practitioners in our field. We have conducted training sessions aimed at sharing our findings with producers, professionals, and community members. We have also presented our research to students and faculty members to help train the next generation of professionals in our field. What do you plan to do during the next reporting period to accomplish the goals?We will continue data collection in 1 to 2-acre paddocks with alfalfa, tall fescue, and bermudagrass. We will expand our measurements using IMU sensors and depth cameras for crop height assessment. Additionally, we've acquired a field spectrometer to measure key biomass quality indicators, enhancing the depth of our data collection. We will present our findings at international and national conferences and present our findings to interested parties via extension trainings.
Impacts
What was accomplished under these goals?
The project aims to advance non-destructive sensing techniques for quantifying aboveground biomass in pasture and forage fields through the integration of UAV and ground-based sensing methods. The developmentsin biomass measurement have the potential toenhance the decision-making process for producers, enabling them to optimize their production activities more effectively. The accomplishments towards the project objectives included modifications in data collection, integration of vegetation coverage in biomass estimation, and rover enhancements to implement various measurement methods. Our achievements during this reporting period encompassed the following areas: data collection, and new methods for crop height measurement that included a new technique called the compression ski.We significantly expanded the scope of our data collection efforts. We broadened our focus to include approximately more plots for each of the three major crops - Alfalfa, Tall Fescue, and Bermudagrass. Moreover, we extended data collection to encompass expansive pasture fields, comprising 1-2 acre paddocks for Alfalfa and Tall Fescue. This expansion allowed for a more comprehensive analysis of crop growth patterns and variations. To ensure operational efficiency, we transitioned to mechanical harvesting methods for pasture fields, utilizing the Carter Harvester, in contrast to the hedge cutter used for plot-based sites. Importantly, we addressed the omission of Bermudagrass data collection from the previous year, which had been impeded by adverse weather conditions.In addition to our existing measurement methods, such as UAV-based Structure from Motion, we introduced innovative techniques to measure crop height more accurately using an Inertial Measurement Unit (IMU) sensor and compression ski: We developed a cutting-edge compression ski constructed from an aluminum sheet, equipped with an IMUsensor. This ski glides over the crop canopy behind the Unmanned Ground Vehicle (UGV), enabling precise measurement of variations in compressed crop height based on IMU sensor data.To further enhance accuracy, we integrated IMU sensors into a cylindrical roller (IMU Roller Method) attached behind the UGV using aluminum rods. This method facilitated measurement of changes in two dimensions asit traversed the crop canopy, yielding more realistic data on crop height variations. We also introduced a depth camera system equipped with two monochromatic cameras that utilized stereo vision for detecting and measuring crop height.In our quest to improve biomass prediction models, we introduced the analysis of vegetative cover using depth cameras (Vegetative Cover Analyses). These cameras, enhanced with RGB cameras, captured canopy images as the UGV moved through the fields. Leveraging machine learning tools and image segmentation, we extracted the percentage of vegetation in each image. To optimize our data collection capabilities, we implemented crucial enhancements to the rover used in our research. These enhancements included the installation of a solar panel, which powered electronics, ensuring extended operational capacity in the field. Additionally, we designed and attached mountings for various sensors, streamlining data collection by allowing the simultaneous use of multiple systems.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2023
Citation:
Koc, A.B., Macinnis, B.M. M.J. Aguerre, J.P. Chastain, A.P. Turner. 2023. Alfalfa biomass estimation using crop surface modeling and NDVI. Applied Engineering in Agriculture. 39(2), 251-264.
- Type:
Journal Articles
Status:
Under Review
Year Published:
2023
Citation:
Koc, A.B., Erwin, C., Aguerre, M., Chastain, J. 2023. Estimating Tall Fescue and Alfalfa Forage Biomass Using an Unmanned Ground Vehicle. Springer Scientific.
- Type:
Journal Articles
Status:
Awaiting Publication
Year Published:
2023
Citation:
Koc, A. B., Singh, J., Aguerre, M. J. (2023). Estimating forage biomass using unmanned ground and aerial vehicles. In Proceedings of International Grassland Congress May 14-19, 2023.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Koc, A.B., MacInnis, B.M. M.J.Aguerre, J.P. Chastain. 2022. Biomass Estimation of Alfalfa using UAV-based Crop Surface Models. 2022 ASABE Annual International Meeting - Houston TX.
- Type:
Conference Papers and Presentations
Status:
Submitted
Year Published:
2023
Citation:
Koc, A.B., Erwin, C., Aguerre, M., Chastain, J. 2023. Estimating Tall Fescue and Alfalfa Forage Biomass Using an Unmanned Ground Vehicle. 15th International Congress of Agricultural Mechanization and Energy in Agriculture (AnkAgEng'23 - Antalya-Turkiye, Oct. 29 - Nov. 2,2023).
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2022
Citation:
Erwin, Curtis, "Unmanned Ground Vehicle Proximal Sensing for Forage Biomass Production Estimations" (2022). All Theses. 3916. https://tigerprints.clemson.edu/all_theses/3916
- Type:
Other
Status:
Published
Year Published:
2022
Citation:
Koc, A.B. 2022. Pasture Biomass Estimation Using Remote Sensing Technology. Simpson Station Field Day, Piedmont Research and Education Center, Pendleton SC (Sep 15, 2022).
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Koc, A.B., Singh, J., Aguerre, M.J. 2023. Estimating forage biomass using crop surface modeling and vegetation indices. American Forage and Grassland Annual Conference, Jan 8-11, 2023, Winston-Salem, NC.
|
Progress 07/01/21 to 06/30/22
Outputs
Target Audience:- Formal classroom instruction of the UAV- and UGV-based aboveground biomass estimation to the Clemson University Agricultural Sciences, and Animal and Veterinary Sciences freshmen andsenior students, and graduate students. - Formal departmental seminarsrelated to the UAV- and UGV-based aboveground biomass estimationto the faculty, staff and graduate students at Agricultural Sciences, and Animal and Veterinary Sciences Department. - Poster presentationsabout theUAV- and UGV-based aboveground biomass estimationto the administrators,faculty, andstudents at Clemson University campus during the Clemson University Center for Agricultural Technologies kick-off event. - Demonstrationsofthe Pasture Biomass Estimation Using Remote Sensing Technology to the Clemson University Extension Associates at the Piedmont Research and Education Center Simpson Station Field Day. Changes/Problems:- Bermuda grass plots did not perform welldue to drought stress during the summer of 2022. - The field days in 2021 were cancelled due to Covid-19. What opportunities for training and professional development has the project provided?- One graduate student who worked on the project successfully defended his MSc thesis and graduated in May 2022. The student gained training on RTK-GPS setup and operation, planning aerial surveys, image acquisition and processing, and data analyses. Other technical skills the student gained was the Robot Operating System (ROS),Solidworks drawing, 3D printing and Python computer programming. The student also had the opportunity to present his work toundergraduate students andfaculty in the department and on campus. In addition, the student gained experience in presenting his work at international meetings, writing a thesis and draft manuscript for a journal. - Another graduate student who worked on the project gained experience onoperating an unmanned ground vehicle and developing a mechanism for the measurement of compressed biomass height, planning surveys for data collection, Solidworks drawing, 3D printing and data analyses.The student also had the opportunity to present his work at professional meetings and to undergraduate students in classroom settings and faculty in the department. In addition, the student gained experience in writing a thesis. - A PhD student in Animal and Veterinary Sciences gained experience inconducting the laboratory analyses of the biomass samples and validating the UAV- and UGV-based aboveground biomass estimation methods. - An undergraduate student assisted the graduate students in collecting data and samples, and data analyses. How have the results been disseminated to communities of interest?The results have been disseminated to researchers and industry personnel through scientific conferences, classroom, and laboratory teaching to undergraduate and graduate students. The field day activities scheduled to demonstrate the research results tothe producers were cancelled due to Covid-19 in 2021. What do you plan to do during the next reporting period to accomplish the goals?- Continue analyzing the collected data for tall fescue and other crops from the research plots. - Continue data collection from the established research plots and from larger pasture fields. Investigate various biomass height measurement mechanisms that would simplify the compressed biomass height measurements. Further improve the unmanned ground vehicleto integrate a solar panel to provide power to the sensors and computer on the unmanned ground vehicle. Modify the ground vehicle and make it suitable for automatic data collection from pre-defined paths. - Investigate the possibility of usingmultispectral/hyperspectral sensors for the estimation of various biomass quality parameters. - Participate in scientific conferences and field days to disseminate the results to a wider audience. - Demonstrate the UAV- and UGV-based aboveground biomass estimation methods to extension associates and pasture managers during the field day activities organized by Clemson University Research and Education Centers.
Impacts
What was accomplished under these goals?
Research plots were established by planting alfalfa, tall fescue, ryegrass and bermuda grass in September 2020. Data collection started in April 2021 and continued until October 2021. Periodic data collection from the research plots included aerial surveys using RGB and NDVI cameras, compressed biomass height estimationsusing the mechanisms mounted on an unmanned ground vehiclebefore and after harvesting the plots. Wet biomass yields for each plot were measured in the field. Samples were collected from each plot for moisture content determination and laboratory analyses.A total of 19 cuttings on different dates between April 15, 2021 and October 14, 2021 were accomplished. Majority of the data collection was on Alfalfa and Tall Fescue,as Bermuda grass plots suffered from heat and water stress in 2021 and earlysummer of 2022. The collected RGB and NDVI images were processed to develop crop surface models using the Structure from Motion (SfM) withAgisoft Metashape software. Processed images were then transferred to ArcGIS Pro for statistical analyses and yield measurements. Dry biomass prediction models for alfalfa were created utilizing the change in crop height determined from crop surface models to estimate the wet biomass yieldcombined with two options to estimate the dry matter fraction. The most generalizable method to estimate the dry matter fraction was a regression equation that predicted the dry matter fraction as a function of the average NDVI value for each plot which was used as a canopy density index in the prediction model. The biomass height measurements collected with the compressed biomass height estimationmechanism mounted on the unmanned ground vehiclewere processed using MS Excel and ArcGIS Pro. Yield maps for alfalfa plots were developed. Alfalfa biomass estimation models have been developed based on the compressed height measurements.
Publications
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2022
Citation:
Brendan MacInnis. 2022. Pasture Biomass Estimation Using Crop Surface Modeling and Vegetation Indices. Clemson University Graduate School.
- Type:
Theses/Dissertations
Status:
Submitted
Year Published:
2022
Citation:
Curtis Erwin. 2022. Human-Robot System Proximal Sensing for Biomass Production Estimations. Clemson University Graduate School.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Brendan MacInnis, A.B.Koc, M. Aguerre, J. Chastain, A. Turner. 2022. Biomass Estimation of Alfalfa Using UAV-Based Crop Surface Models. Presented at the 2022 ASABE Annual International Meeting, Houston TX. July 17-20, 2022.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Curtis Erwin, A.B.Koc, M. Aguerre, and J. Chastain. 2022. Aboveground biomass estimation using an unmanned ground vehicle. Presented at the 2022 ASABE Annual International Meeting, Houston TX. July 17-20, 2022.
- Type:
Other
Status:
Published
Year Published:
2022
Citation:
A.B. Koc. 2022. Aboveground biomass estimation using and unmanned ground vehicle. Presented at the Clemson University Center for Agricultural Technology (CU-CAT) Kick-off Event on August 31, 2022. Clemson SC.
- Type:
Other
Status:
Published
Year Published:
2022
Citation:
A. B. Koc. 2022. Biomass Estimation of Alfalfa Using UAV-Based Crop Surface Models. Presented at the Clemson University Center for Agricultural Technology (CU-CAT) Kick-off Event on August 31, 2022. Clemson SC.
- Type:
Journal Articles
Status:
Submitted
Year Published:
2022
Citation:
Ali Bulent Koc, John P. Chastain, Brendan M. MacInnis, Matias J. Aguerre, Aaron P. Turner. 2022. Alfalfa Biomass Estimation Using Crop Surface Modeling and Vegetation Indices.
|
Progress 07/01/20 to 06/30/21
Outputs
Target Audience:Theultrasonic compressed biomass height estimation mechanism using a rover and the biomass estimation from aerial images produced some preliminary results. We are still collecting data to develop models that would be used for biomass estimation in pastures. Upon completion of data collection and model development, we will disseminate the resultsand demonstrate the UAVs and rover-based biomass estimation toolsto cattle producers at the field days. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Three graduate students (two master'slevel and one doctorallevel) have been directly working on this project. The two master'sstudents are working on aerial and ground-based biomass measurements. The doctoralstudent has been working on the sample collection and analyses. In addition to these graduate students, two undergraduate students have been trained on vegetation coverage development. How have the results been disseminated to communities of interest?The preliminary results of the project were presented to graduate students and undergraduate students majoring in Agricultural Sciences, and Animal and Veterinary Sciences at Clemson University. A poster presentation about ultrasonic compressed biomass height measuring mechanism was delivered at the 5th CIGR International Conference 2021 on May 11-14, 2021 (Virtual). Another poster presentation about the preliminary results of the use of UAV-based biomass estimation will be delivered at the ASABE 2021 Annual International Meeting on July 12-16, 2021. What do you plan to do during the next reporting period to accomplish the goals?For objective 1: We will continue collecting aerial images over the research plots. The vegetation season for this year's crops will end around October 2021. We will start the second-year experiments in October by planting ryegrass and plating other annual grasses in the spring of 2022. We will process the aerial images and combine all the first-year and second-year data to develop a biomass estimation model based on the aerial images. For objective 2: We will continue collecting biomass compressed height measurements data. We will integrate a camera on the ground rover and record the vegetation coverage data along with biomass compressed height to estimate biomass amount.After completing the first-year data collection, we will use nearby grazing fields at the Piedmont REC to evaluate the accuracy of the biomass estimations from aerial images and ground measurements. We are currently designing a new, more versatile ground rover. The current rover that has been used for data collection does not have enough width to cover each plot in one pass and the tires of the rover compress the plants which makes itdifficult to collect data over compressed grass in a plot on the same day. Therefore, we will complete the new rover design and equip it with a depth camera and a lidar sensor and install the ultrasonic biomass compressed height sensingmechanism. We will use the new rover for biomass compressed height measurements for the second year data collection. For objective 3: We have designed a prototype biomass collection mechanism that can be mounted on the rover that cuts and collects grass samples. The collected samples will be placed in a sample holder on which we will install an NIR sensor to evaluate the quality of the sample.We will fabricate the sample collection mechanism and integrate the NIR sensor for infield biomass quality evaluations. We are aiming to use the sample collection and NIR quality evaluation system during the second-year tests. We are planning to submit two manuscripts for publication and present the results of our findings at the ASABE annual international meeting in 2022.
Impacts
What was accomplished under these goals?
Accomplishment - We established the research plots for the first-year experiments. For data collection, model development and system accuracy tests, we establishedresearch plotsat theClemson University Piedmont Research and Education Center (REC). Four forage crops on which the developed systems to be tested are Ryegrass, Alfalfa, Bermudagrass and tall fescue. In addition to these crops, we also included oats in the experimental design. Among these crops, ryegrass and oats are cool season annual grasses. We planted annual ryegrass and oats, and perennial alfalfa and tall fescue on October 1, 2020. Split plot in randomized complete block experimental design was used to establish the research plots. Each crop in experimental design is replicated four times. For testing the performances of developed aerial and ground-based systems, each plot was further split into three subplots for measurements at 10-, 20-, and 30-day intervals. We completed the first-year measurements on the Ryegrass and Oats plots. We have started conducting periodic measurements on tall fescue and alfalfa and currently collecting data at 10-, 20-, and 30-day intervals. Accomplishment for Objective 1 - We acquired aerial RGB and NDVI images with the UAV and developed crop height models from digital elevation and crop surface models. A UAV with an onboard RGB camera and a MAPIR-NDVI camera were used to collect aerial images over the research plots. The NDVI camera was mounted on the UAV. Aerial images of the plots with the UAV have been taken at 10-, 20- and 30-day intervals. Ground control points in the research field were set using 1-ft2 tiles. A multiband RTK-GNSS receiver was used tocollect GPS coordinates of the ground control points. UAV aerial images over the plots have been taken periodically. Crop Surface Models (CSM) and Digital Elevation Models (DEM) of the research plots using the Multiview 3D reconstruction with structure from motion (SfM) algorithm have been developed. After each measurement interval, Crop Height Model (CHM) was generated by subtracting the DEM from the CSM. In addition to the CHMs, we have also acquired NDVI images and generated NDVI maps of the research plots for each measurement interval. Accomplishment for Objective 2 - We built a ground rover and developed a biomass compressed height sensing mechanism. A four-wheel-drive ground rover was built. Fixed plates with attached load cells were mounted to the rover to measure biomass compressed height. Initial testing of the sensing plates measured the force deflecting the plants when the rover operated in a straight line. However, the sensing plates lost their functionality while the rover was turning. Therefore, we developed a more reliable biomass compressed height sensing mechanism based on ultrasonic time of flight measurements. Using ultrasonic sensors directly over biomass does not provide reliable biomass height measurements due to the irregular and non-perpendicular canopy of the plants. Instead, we came up with a mechanism that uses an acrylic plate mounted underneath the rover. The size of the acrylic plate was selected such that its applied pressure is less than 8 kg/m2. This pressure limit has been reported in the literature for designing manually operated falling plate meters. Each side of the plate was equipped with side mount ball bearing slides positioned vertically to let the plate move in vertical direction freely. The plate slightly compresses the grass but still floats over the grass as the height and density of the grass varies. An ultrasonic sensor was mounted near the top of the rover frame faced down and centered above the acrylic plate. Instead of sensing the plant height directly using ultrasonic distance sensor, we measure the distance between the ultrasonic sensor and the acrylic plate as it is pulled by the rover. During calibration before each test, the plate is lowered all the way to bare ground. When the plate is on the ground, the ultrasonic sensor measures the maximum distance between the sensor and plate. As the plate moves in vertical direction, depending on the height of the grass, the distance between the sensor and the plate gets smaller than the initial position. Subtracting the measured distance from the maximum distance (distance measured when the plate is on the ground) gives the compressed height of the plants. We installed the RTK-GNSS receiver on the rover during tests. An Arduino Mega microcontroller and a computer were used to record the compressed height measurements and corresponding GPS locations at one-foot intervals. The initial testing of the biomass compressed height measuring mechanism over different grasses provided promising results, therefore we decided to use this mechanism to measure the grass height in research plots. We have also developed a computer program that calculates the vegetation coverage from images. We will mount a camera on the rover to capture images as the rover moves over the crop. The captured images will be processed in real time to estimate the vegetation coverage. The vegetation coverage information and weather data will be integrated with ultrasonic biomass compressed height measurements to estimate the biomass amount. Accomplishment related to Objectives 1 and 2 - Data collection. The following order of field operations and measurements was used when collecting data over the research plots. Acquire RGB and NDVI images with the UAV, Measure crop height in plots using commercially available rising plate meter manually, Measure crop height using the ground rover and the ultrasonic compressed biomass height measuring mechanism, Collect biomass samples from each plot with clipping, Measure biomass samples for moisture content, wet and dry weights, and analyze samples for nutrients in the lab, Process images and develop crop height models and NDVI maps from the UAV images, Conduct data analyses and enter measured and calculated values on an Excel worksheet.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Erwin, C., Koc, A. B., Aguerre, Macinnis, B., Pena, M., Koparan, C. 2021. Non-destructive forage biomass estimation with ground robots. 5th CIGR International Conference, Qu�bec City, Canada, May 11-14, 2021, Virtual.
- Type:
Conference Papers and Presentations
Status:
Submitted
Year Published:
2021
Citation:
MacInnis, B., Koc, A.B., Erwin, C., Aguerre, M. 2021. Pasture Biomass Estimation Using Crop Surface Modeling and Vegetation Indices. ASABE 2021 Annual International Meeting, Virtual and On-Demand. July 12-16, 2021.
|
|