Progress 04/01/24 to 03/31/25

Outputs
Target Audience:We have hosted three demonstrations at I-FARMS. Our first demonstration was hosted at our satellite site at Tuskegee University, where we demonstrated the use of drones, sensor robots and cover crop robots. Due to rain, we had about 10-12 farmers participate in the demonstration. Our second demonstration was conducted on August 15th in collaboration with Regenerative Agriculture. At the demonstration, we highlighted case uses of drones (targeted spraying), robotic dogs (livestock herding), benefits of cover crop seeding, and watershed demonstration. This was a very successful event with about 60 participants, including farmers and the public. Our third demonstration occurred in December. This demonstration was done in coordination with SAFER AG, and we invited researchers and companies to demonstrate their autonomous equipment on campus. We had 14 exhibitors (including 3 from the I-FARM thrusts). We had about 100 participants coming from diverse backgrounds, including industry, academia and farmers. In our first six months we participated in four field days /farm visits with a total of 137 participants. In the second part of the year, we participated in 8 field days/farm visits. These events targeted farmers, students and ag professionals and had approximately 1,100 participants. During these events we showcased the cover crop robot, the robot dog (for animal control), and various drones. We have focused our efforts on developing the I-FARM University training series this year. We have developed a 5 webinar series called the "I-FARM University fall series". This webinar series is a 30 minutes monthly program on latest research regarding autonomous agriculture. We have completed three webinars so far with emphasis on connectivity, robotics, and cost of weeding and we are currently on target to complete our last two webinars (animals, and small holder farmers) this fall. We have had a total of 273 participants (average 55) with an average score of 8.30 out of ten. Each of these webinars are recorded and placed on youtube with an average viewership of about 350 views per webinar. We used these webinars to highlight research and news in our robotics, weeding, small holder, connectivity and animal thrusts. We have also hosted a lecture on AI in agriculture in the Crop Management Conference hosted by extension. We have completed the development of 1 virtual reality video focused on showcasing the robotic dog and how it interacts with grazing cows. Changes/Problems:We are thankful for the No Cost Extension for one more year to complete the project objectives. What opportunities for training and professional development has the project provided?? We had the IFARM learning series: https://i-farm.illinois.edu/i-farm-university-learning-series/ 5 presentations over 5 months Had several outreach activities to the target audience as described earlier in our target audience question. Launched Industry Partnership Program that can facilitate the use of the test bed for future agtechnology Conducted a workshop in IROS 2025 in Abu Dhabi with over 120 attendees from around the world. How have the results been disseminated to communities of interest?We have hosted three demonstrations at I-FARMS. Our first demonstration was hosted at our satellite site in Tuskegee University, where we demonstrated the use of drones, sensor robots and cover crop robots. Due to rain, we had about 10-12 farmers participate in the demonstration. Our second demonstration was conducted on August 15th in collaboration with Regenerative agriculture. At the demonstration we highlighted case uses of drones (targeted spraying), robotic dogs (livestock herding), benefits of cover crop seeding, and watershed demonstration. This was a very successful event with about 60 participants including farmers and public. Our third demonstration occurred in December. This demonstration was done in coordination with SAFER AG, and we invited researchers and companies to demonstrate their autonomous equipment on campus. We had 14 exhibitors (including 3 from the I-FARM thrusts). We had about 100 participants coming from diverse backgrounds including industry, academia and farmers. In our first six months we participated in four field days /farm visits with a total of 137 participants. In the second part of the year, we participated in 8 field days/farm visits. These set of events targeted farmers, students and ag professionals and had approximately 1,100 participants. During these events we showcased the cover crop robot, the robot dog (for animal control), and various drones. We have focused our efforts on developing the I-FARM University training series this year. We have developed a 5 webinar series called the "I-FARM University fall series". This webinar series is a 30 minutes monthly program on latest research regarding autonomous agriculture. We have completed three webinars so far with emphasis on connectivity, robotics, and cost of weeding and we are currently on target to complete our last two webinars (animals, and small holder farmers) this fall. We have had a total of 273 participants (average 55) with an average score of 8.30 out of ten. Each of these webinars are recorded and placed on youtube with an average viewership of about 350 views per webinar. We used these webinars to highlight research and news in our robotics, weeding, small holder, connectivity and animal thrusts. We have also hosted a lecture on AI in agriculture in the Crop Management Conference hosted by extension. We have completed the development of 1 virtual reality video focused on showcasing the robotic dog and how it interacts with grazing cows. What do you plan to do during the next reporting period to accomplish the goals?Thrust 1: The final goal is to monitor crop growth environment and quantify crop productivity accurately in real time. To do this, we are continuously improving CropEYE's functionality and stability by adding more cameras and sensors in the past year. Figure 4 shows the design of the old version and current version of the CropEYE system. The old version only measures canopy structure parameters (LAI, MTA) with several cameras. However, weather information and leaf biochemical traits are also important for crop productivity monitoring. Here are some improvements we are making (Figure 5). First, several weather sensors are added to an extra component "Env", which will collect local weather data, including air temperature, humidity, air pressure, wind speed and solar radiation. They are mostly essential variables for crop modeling. Second, spectral sensors in the Nadir view can collect multispectral reflectance of the canopy. Then we can predict leaf traits from multispectral reflectance of the canopy by establishing empirical relationship between spectral data and leaf traits. Lastly, a LTE module is being developed and tested. The current version can only access to the Internet via WiFi, but most crop fields don't have WiFi available. It's important to add the LTE module to extend the coverage of the system and enhance its applicability. In the coming growing season, we expect to collect the complete dataset mentioned above, and quantify crop canopy photosynthesis with 3D models Thrust 2: Continue our cover cropping and data collection using autonomous robots for the next year. Thrust 3: · Integrating sensors into a Farm-of-the-future app: We are working on creating a new version of the MyFarm app with a better interface and end-to-end integration of sensor data. · Generative AI integration: We plan to improve the CropWizard integration to incorporate historical sensor data into its knowledge base. Agronomic Goals: We will align data collection pipeline with agronomic goals over the next crop cycle, under advisement of Prof. Sunoj Shahjahan Thrust 4: In the no-cost extension year, the project will focus on real-world validation and optimization of developed models at the FoF site. The BCS and weight estimation models will be tested under on-farm conditions to assess their robustness in diverse management environments. The grazing behavior and forage monitoring studies will be expanded, incorporating additional environmental and management factors to refine dry matter intake prediction models. The locomotion monitoring system will be further developed to identify early signs of lameness in confined cattle. Additionally, efforts will be made to integrate data from different sources, including RGB-D cameras, drones, and accelerometers, into a unified framework for continuous livestock monitoring. The team will also prioritize outreach through peer-reviewed publications and conference presentations Thrust 5: During the no-cost extension period, we will focus on finalizing data analyses, publishing key findings, and advancing machine learning models. Specific objectives include: Enhancing machine learning models with expanded datasets for improved predictive accuracy. Finalizing technical training modules for students and stakeholders. Preparing manuscripts and conference presentations based on research findings. Expand our free crop health diagnostic and experiential training on using autonomous crop phenotyping tools to additional farmers through collaboration with extension officials. Thrust 6: Complete two manuscripts for nitrogen management and phosphorus and potassium management, respectively, and send them to journals for publication Extension: On-site Demonstrations: Similar to last year, we are currently working with our thrust partners in hosting three-four on-site demonstrations on autonomous technologies, two-three at our site in UIUC and one in Tuskagee University. We plan to host a field day in spring on animal grazing and second in August (when cover planting starts). Some of the demonstrations we are hoping to conduct include: Operating multiple ground robots remotely from an office; Livestock grazing demonstration, and demonstrate internet-of-things technology. Depending on the status, we hope to showcase App as well. I-FARM University: Our goal for next year us to record 2-3 training presentations and make them available for the general public to review on our I-FARM University Website. We also want to revamp our website to be able to better host the resources and training programs we have organized over the last three years. We also aim to complete development of a second virtual reality module that will be used in our demonstration programs. Other Activities: We aim to continue outreach to regional conference, conduct farm visits, and generate news articles as we conducted in the first year. In addition, we plan to host a space at the Farm Progress Show in Decatur, IL 2025.

Impacts
What was accomplished under these goals? Canopy & Productivity Monitoring - Collected Leaf Area Index (LAI) data using our CropEYEs camera network during the 2024 growing season (July-September). Developed an algorithm to calculate LAI from field photos using gap fraction theory. - Installed a weather station to measure local conditions including temperature, humidity, and solar radiation. This data supports photosynthesis modeling and GPP product generation, with a related publication in preparation. Soil Moisture Sensing - Deployed eight soil moisture stations during the 2024 season (July-October): - Six stations monitoring at 5, 15, and 30 cm depths - Two additional stations monitoring at 50 cm depth - Implemented rigorous quality control with paired sensors at each depth, resulting in higher quality data compared to previous years. Agricultural Robotics - Successfully demonstrated autonomous cover crop robots across 80 acres in just 4 days - Featured robotics technology at IROS 2024 conference workshop in Abu Dhabi Network & Computing Infrastructure - Established a multi-modal connectivity testbed integrating CBRS, LoRa, and Starlink networks - Enhanced TerraSentia robots with custom multi-camera arrays for advanced crop monitoring - Published research in Remote Sensing journal and submitted to USENIX NSDI 2025 - Integrated sensor data with CropWizard's generative AI for farmer decision support Livestock Monitoring - Developed computer vision systems for: - Automated body condition scoring - Cattle facial recognition (>95% accuracy) - Lameness detection (99% precision) - Conducted field demonstrations of real-time monitoring technologies Outreach & Education - Hosted community events honoring Dr. George Washington Carver's legacy - Engaged students through: - High school STEM programs - Drone technology workshops - Hands-on agricultural robotics demonstrations - Provided training for local farmers on autonomous crop monitoring tools Nutrient Management - Analyzed nitrogen application strategies across multiple optimization scenarios - Conducted farmer survey on phosphorus/potassium management: - 12.8% response rate from Midwest farmers - Preliminary results show farmers' willingness to accept compensation for reduced application rates

Publications

• Type: Conference Papers and Presentations Status: Accepted Year Published: 2025 Citation: Xavier, D. B.; Benicio, Rahman, M.; Lopes, L. B.; Condotta, I. C. F. S. Computer vision model as a tool to automate body condition score classification of dairy cows.
• Type: Conference Papers and Presentations Status: Published Year Published: 2024 Citation: Sivakumar, A. N., M. V. Gasparino, M. McGuire, V. A. H. Higuti, M. U. Akcal, and G. Chowdhary. "Demonstrating CropFollow++: Robust Under-Canopy Navigation with Keypoints." Proceedings of Robotics: Science and Systems, Delft, Netherlands (2024).
• Type: Conference Papers and Presentations Status: Accepted Year Published: 2024 Citation: Benicio, L. M., Condotta, I. C., Lopes, L. B., Xavier, D. B., Vendrusculo, L. G., & Lima, I. B. (2024). Individual Facial Identification of Beef and Dairy Cattle Based on Computer Vision. ASABE Annual International Virtual Meeting
• Type: Conference Papers and Presentations Status: Submitted Year Published: 2025 Citation: Sie, Emerson, Bill Tao, Aganze Mihigo, Parithimaal Karmehan, Max Zhang, Arun N. Sivakumar, Girish Chowdhary, and Deepak Vasisht. "BYON: Bring Your Own Networks for Digital Agriculture Applications.
• Type: Other Status: Other Year Published: 2024 Citation: Xinlei Wang, Kaiyu Guan, Sheng Wang, et al. The role of vertical profiles of soybean leaf traits in estimating canopy photosynthesis. AGU24 Xavier, D. B., Benicio, L. M., Condotta, I. C. F. S. (2024). Computer Vision Model as Tool to Automate Key Point Detection from Top-view images on Dairy Cows. [Poster]. CDA Conference 2025. Benicio, L. M., (December 2024). Applications of Precision Management Tools in Beef Cattle Management. Presented at Safety for Emerging Robotics and Autonomous Agriculture (SAFER AG) Workshop, Urbana, IL. Benicio, L. M., Condotta, I. C. F. S., Adbdala, F., Xavier, D. B. (October 2024). Individual Identification of Cattle using Computer Vision Models. Poster session presented at AIFARMS Year 4 Annual Conference and Reviewers, St. Louis, Missouri. Xavier, B. D., Benicio, L. M., Condotta, I. C. F. S. (October 2024). AI-Driven Body Condition Score Classification of Dairy Cows using depth image processing. Poster session presented at AIFARMS Year 4 Annual Conference and Reviewers, St. Louis, Missouri. Italo B. G. Lima, Isabella C. F. S. Condotta, Wei-Chen C, Diego B. Xavier, Luciano B. Lopes, Priscila J. R. Cruz. (2024). Navigating grazing preferences: Using geopositioning to understand the influence of temperature, time of day, and weather conditions on animal grazing behavior. [Poster]. CDA Conference 2024. Benicio, L. M., Adbdala, F., Xavier, D. B, Cardoso, F., Condotta, I. C. F. S. (March 2025). Individual Identification of Cattle Using Computer Vision Models. Poster session presented at CDA Annual Conference  Center for Digital Agriculture, Urbana, Illinois. Empowering Small-Scale and Minority Farmers: Expanding Access to Autonomous Crop Phenotyping Technologies Through Targeted Outreach during the Professional Agricultural Worker Conference, Montgomery, AL Menglin Liu, Madhu Khanna, Shady S. Atallah. Policy Instruments to Promote the Adoption of Sustainable Nitrogen Management Practices. 2024 Heartland Workshop in Environmental and Resource Economics, October 26-27, Urbana, IL. Menglin Liu, Madhu Khanna, Shady S. Atallah. Policy Instruments to Promote the Adoption of Sustainable Nitrogen Management Practices. 2024 joint FACS (Food, Agriculture, and Commodities) and pERE (program in Environmental and Resource Economics) seminar, November 20, Urbana, IL. Menglin Liu, Madhu Khanna, Shady S. Atallah. Policy Instruments to Promote the Adoption of Sustainable Nitrogen Management Practices. 2024 SEA (Southern Economic Association) 94th Annual Meeting, November 23-25, Washington D.C. Menglin Liu, Madhu Khanna, Shady S. Atallah. Policy Instruments to Promote the Adoption of Sustainable Nitrogen Management Practices. 2025 AERE (Association of Environmental and Resource Economists) Summer Conference, May 28-30, Santa Ana Pueblo, NM.

Progress 04/01/23 to 03/31/24

Outputs
Target Audience:Engagement with Industry: Members of the I-FARM team have had one advisory board meeting with personnel from Cargill, EarthSense,Verizon, AIGEN, Grain Weevil and Corteva. In addition, we are working towards launching a Industry Partnership Program (IPP) for continuing the testbed beyond the project duration. EarthSense is closely engaged with Under-canopy Robotics for Sustainable Farm Management thrust. We were successful in obtaining a $5M climate smart grant in collaboration with Earthsense and our partners Tuskegee University to scale the robotic cover cropping to 10k acres in three years. The technology has come out of thorough testing from the IFARM testbed. Engagement with Farmers: We conducted farm visits at local community colleges and nearby farms, engaging with a total of 210 attendees. These visits offered participants hands-on experience and firsthand knowledge of the latest agricultural practices and technologies. In addition, we have 5 farmers in our advisory board who attended our annual board meeting and provided feedback to demonstrate different use cases instead of our existing technology demos to have a better impact. Changes/Problems:We are pleased to announce the extension of the testbed from 80 acres to 115 acres. This expansion aims to facilitate better lateral tiling and ensure a sufficient number of plots to accommodate additional industry and university collaborators, thereby enhancing the sustainability of the testbed beyond the grant period. Furthermore, we are transforming the testbed into an integrated crop-livestock research and demonstration testbed. This strategic move will enable us to comprehensively study the impact of various management practices, fostering a deeper understanding of agricultural systems and promoting sustainable practices in crop and livestock management. We are excited about these developments and look forward to the opportunities they will bring for collaborative research and innovation What opportunities for training and professional development has the project provided?Students and post-docs: 14 students (graduate and undergraduate) and 2 post-doctoral scientists and 1 research technician work on I-FARM projects. The students and post-docs are spread across all the I-FARM research thrust with several of them working between agriculture and computer science disciplines. Extension: The extension efforts for our testbed have been multifaceted and impactful, encompassing a range of demonstrations, farm visits, trade shows, and conferences. Over the past year, we have hosted three demonstrations at I-FARMS, each focusing on different aspects of agricultural technology and innovation. These demonstrations attracted a total of 37 participants and provided valuable insights into the practical applications of cover crop robotics, drones, and other advanced technologies in agriculture. In addition to these demonstrations, we conducted farm visits at local community colleges and nearby farms, engaging with a total of 210 attendees. These visits offered participants hands-on experience and firsthand knowledge of the latest agricultural practices and technologies. At prominent trade shows such as the Farm Progress Show and the National Association of County Ag. Agents Annual Meeting Trade Show, we showcased our research and technologies to a wide audience of industry professionals and stakeholders. These events provided us with invaluable opportunities to network, disseminate information, and raise awareness about our research initiatives. Furthermore, our participation in extension training workshops and conferences has allowed us to reach out to extension professionals and agricultural experts, sharing our expertise and insights on topics such as cover cropping, drone technology, and artificial intelligence in agriculture. Looking ahead, we are continuing to plan and participate in conferences and events to further disseminate our research findings and engage with the broader agricultural community. Our goal is to continue driving innovation and sustainability in agriculture through collaborative research, education, and outreach efforts. Outreach: In addition to our outreach efforts at Tuskegee University in Alabama, we are actively engaged in expanding our outreach through I-FARM University. Over the past year, we have made significant progress in developing resources and materials to support agricultural education and outreach initiatives. One notable achievement is the completion of the last factsheet on Internet of Things (IoT), which has been printed and distributed at various events, including the Iowa trade show and Farm Progress Show. These factsheets serve as valuable resources for farmers and interested attendees, providing them with essential information on cutting-edge agricultural technologies. Additionally, the factsheets have been published on our website, ensuring accessibility to a broader audience. To enhance our online presence and educational resources, we have launched a new website for I-FARM University (i-farm.illinois.edu). This platform serves as a centralized hub for accessing educational materials, videos, and virtual reality (VR) content related to agriculture and technology. Speaking of VR content, we have developed our first VR video, which will be launched alongside VR headsets in the fourth quarter of this year. This immersive experience will provide users with a firsthand glimpse into the future of farming and technology. Furthermore, we have updated our demonstration toolkit with new tents and other necessary items to conduct engaging and informative demonstrations. These demonstrations play a crucial role in showcasing our research and technologies to a wide audience, fostering awareness and understanding of modern agricultural practices. Looking ahead, we are in the process of developing two training sessions, which will be recorded in the fourth quarter (February-April). These training sessions will provide valuable insights and practical knowledge to agricultural professionals and stakeholders. In addition to our digital and on-campus initiatives, we are collaborating with the Children's Discovery Museum in Normal/Bloomington, IL, to develop a new agricultural exhibit. We have invited them to campus to demonstrate the farm of the future and will be meeting with them in Bloomington to help plan their new exhibit. This collaboration aims to inspire and educate children and families about the importance of agriculture and technology in shaping our future. Overall, our outreach efforts through I-FARM University are aimed at fostering education, innovation, and community engagement in agriculture, ensuring a sustainable and prosperous future for the industry. How have the results been disseminated to communities of interest?Engagement with Research Community: Naveen Uppalapati and Girish Chowdhary organized a workshop on Agricultural Robotics for a Sustainable Future in IROS 2023 held in Detroit. The workshop had close to 200 attendees from around the world. In addition, work on autonomy improvement and connectivity are under review in different conferences. Engagement with Industry: Members of the I-FARM team have had one advisory board meeting with personnel from Cargill, EarthSense,Verizon, AIGEN, Grain Weevil and Corteva. In addition, we are working towards launching a Industry Partnership Program (IPP) for continuing the testbed beyond the project duration. EarthSense is closely engaged with Under-canopy Robotics for Sustainable Farm Management thrust. We were successful in obtaining a $5M climate smart grant in collaboration with Earthsense and our partners Tuskegee University to scale the robotic cover cropping to 10k acres in three years. The technology has come out of thorough testing from the IFARM testbed. Engagement with Farmers: We conducted farm visits at local community colleges and nearby farms, engaging with a total of 210 attendees. These visits offered participants hands-on experience and firsthand knowledge of the latest agricultural practices and technologies. In addition, we have 5 farmers in our advisory board who attended our annual board meeting and provided feedback to demonstrate different use cases instead of our existing technology demos to have a better impact. What do you plan to do during the next reporting period to accomplish the goals?Thrust 1 Plans: - Continue ground, airborne, and satellite data collection. - Generate spatial maps of crop traits using airborne and satellite remote sensing data. - Quantify differences in cover crop planting dates via robot vs. traditional methods. - Use isotopic tracer to quantify nitrogen immobilization and leaching. - Quantify cover crop root biomass to explain differences in nitrate leaching reductions. - Collaborate with Animal Science Dept. on grazing impact studies. - Participate in outreach and educational activities. Thrust 2 Plans: - Scale up robotic cover cropping with drones and ground robots. - Implement high-throughput biomass estimation using cover crop height measuring robot. - Collaborate with other thrusts to achieve milestones. - Engage other robotic industries for mechanical weeding and spot spraying. - Develop multi-robot autonomy using CBRS to reduce operation costs. Thrust 3 Plans: - Integrate end-to-end robot+networking platform for farm data collection. - Incorporate CropWizard into the MyFarm app. - Create a new mobile connectivity module for flexible farm data collection. Thrust 4 Plans: - Correlate collected information with image data for model validation. - Evaluate drone-generated images for animal and cover crop monitoring. - Retrofit animals with microphones, GPS, and accelerometers for data collection. - Enhance activity detection algorithm for dairy cows. - Improve dry matter intake prediction models. Thrust 5 Plans: - Implement cutting-edge technology for crop health assessment using AI and advanced algorithms. - Provide free crop diagnostic services for local and international agricultural communities. Thrust 6 Plans: - Nitrogen (N) Management: Examine green insurance impact on MRTN and VRT adoption, conduct cost-effectiveness assessment, present findings in conferences, and write a paper. - Phosphorous (P) and Potassium (K) Management: Collect survey data, analyze, and write a paper. Extension Plans: - On-site Demonstrations: Host on-site demonstrations showcasing autonomous technologies, internet-of-things, and livestock demonstrations. - I-FARM University: Develop new factsheets, record training presentations, create virtual reality modules, and host training events. - Other Activities: Continue outreach to regional conferences, conduct farm visits, and generate news articles

Impacts
What was accomplished under these goals? Accomplishments in Integrating Remote-sensing and IoT with Farm Planning and Management (Thrust 1) Canopy Monitoring: Continuous LAI data collection via CropEYE cameras for agroecosystem modeling. Soil Moisture Monitoring: Aclima sensors for precise soil moisture tracking, with regular data checks. Research Focus: Integrated cover cropping and grazing experiments, root biomass analysis, and automated robot impact evaluation. Experimental Designs: Collaboration led to two experimental setups for studying grazing integration and cover crop impact. Root Biomass Analysis: Labor-intensive process aimed at understanding soil carbon response. Robot Experimentation: Assessing robot impact on cover crop growth, with potential for increased biomass and reduced nitrate loss. Accomplishments in Under-canopy robotics for sustainable farm management (Thrust 2) Robotic cover crop planting showed twice the biomass compared to tractor seeding. Evaluation of three robots showed an average autonomy of 767 meters between interventions, with a best-case scenario of 3571 meters. Developed a cover crop height measuring robot for efficient biomass estimation. Utilized drones and robots for seeding Oats/turnips in soybean fields. Collaborated with various thrusts to study nitrate loss reductions, grazing models, and remote robot control. Cover cropped 60 acres of soybean with Oats/turnips mix using drones and 55 acres with under-canopy wheeled robots. Videos showcasing drone and robotic cover cropping were produced for extension and outreach activities Accomplishments in Rural Internet Connectivity and Edge Computing (Thrust 3) Current Deployment Status: We have deployed a cutting-edge connectivity testbed on the Farm of the Future. This testbed includes private wireless (CBRS-Citizens Broadband Radio Service), LoRa, and satellite connectivity by Starlink. Key Results: We performed the first set of extensive measurements over the last growing season on the performance and reliability of CBRS and satellite connectivity. A key emphasis in our project was to study the impact of crops on CBRS signals. Some insights from the deployment: Range: The CBRS connection enabled connectivity over a range of 3.5 Km in our deployment. Upload and download speeds: We observed a max download speed of 100 Mbps and upload speed of 25 Mbps from the robot. Effect of crops: Crops significantly reduce the range and capacity of CBRS operation. We experience an attenuation of up to 30 dB (1000 times lower power) at the same location. This reduces the range to 0.5-0.7 Km and correspondingly limits the data rates. Satellite connectivity: Starlink achieved a download capacity of 200 Mbps and uploads in the range of 50-60 Mbps, with temporal variation due to weather. Starlink terminals are heavy (4.2 Kg) and cannot be carried by robots. Starlink connectivity does not work under the canopy due to high-frequency signals being completely blocked by crops. Activity: Specifically, we highlight some key activities over the last year below: Robot Integration: We integrated new high-resolution cameras on the robots to collect extensive per-plant data throughput the season, in preparation for the next season. Machine Learning Models: We trained several new machine learning models for pest, weed, and disease detection using data curated from extensive online sources MyFarm app: We created a prototype MyFarm application to integrate satellite imagery and sensor data from the farm. Accomplishments in Integrating Animals in the Farm of the Future (Thrust 4) Acquired over 4,000 images of beef and dairy cattle using surveillance and depth cameras, capturing various data including behavior, body metrics, and health indicators. Ongoing annotation process for images and videos to enhance data analysis. Weather-related challenges affected data collection during cover crop seeding, prompting plans for a repeat experiment in spring. Accomplishments in Accessibility of Technology to Small-Holder and Minority Farming (Thrust 5) Conducted outreach and experiential learning initiatives featuring autonomous farming robots tailored for small-scale and minority farmers, students, and stakeholders. Experiential learning sessions conducted for Tuskegee students and faculty members, including demonstrations of TerraSentia crop phenotyping rover and agricultural robotics. Involvement in AGRI-TREK Summer Program, providing instruction in autonomous tools for agricultural functions and conducting lytic peptide assays. Experiential learning activities for high school students, local farmers, and visiting collaborators, including demonstrations of autonomous farming robots, multispectral drones, and crop diagnostics. Accomplishments in Socioeconomics of technology adoption (Thrust 6) Nitrogen (N) Management Used the general quadratic model to estimate the yield response to N function and N leaching response to N function at the soil level based on a simulated dateset Assumed three different corn and N price ratios to compare the nitrogen application rate, yield, profitability, and N leaching between yield goal approach, MRTN (maximum return to nitrogen), and VRT (variable rate technology) derive the critical values of risk aversion that result in farmers being indifferent between two nitrogen management practices and determine the risk aversion distribution assess the impact of three policy instruments on the adoption rates of MRTN and VRT Phosphorous (P) and Potassium (K) Management Drafted a choice experiment (CE) survey to investigate farmer preferences for potential changes resulting from adopting maximum return to P (MRTP) and/or maximum return to K (MRTK) recommended rates and willingness to accept (WTA) payment to adopt these practices Conducted a focus group with six farmers to test the survey draft and discussed the survey language and content? Notable results Nitrogen (N) Management The expected profit gains varied from 8 to 13 $/acre after transitioning from yield goal to either MRTN or VRT. MRTN exhibited the lowest variance in profits, while VRT demonstrated the highest variability. In the absence of policy instruments, farmers would choose MRTN over yield goal approach if the utility maximization assumption hold true. Price premium increased both the expected profits and their variance. This altered the critical values of risk aversion from negative to positive for MRTN and reduced the critical values for VRT when compared to yield goal, which was not effective to promote adoption. The lump-sum payment and the tax exhibited similar effects by augmenting differences in expected profits between practice pairs without altering profit variance differences. Accomplishments in Extension and Research Integration Efforts Conducted three demonstrations at I-FARMS showcasing cover crop robotics, drones, and cover cropping techniques. Hosted farm visits at local community colleges and nearby farms, engaging approximately 210 attendees. Participated in Farm Progress Show, National Association of County Ag. Agents Annual Meeting Trade Show, and extension training workshops, reaching over 565 attendees. Demonstrated cover crop, drone, and AI technology at various conferences/events in Illinois and Iowa, engaging about 260 attendees. Completed development of extension factsheets and launched a new website for I-FARM University. Developed VR video and updated demonstration toolkit for future use. Collaborating with the Children's Discovery Museum in Normal/Bloomington, IL to develop a new agricultural exhibit, showcasing the farm of the future.

Publications

Progress 04/01/22 to 03/31/23

Outputs
Target Audience:The target audience for this project are industry, farmers, researches in agriculture, digital agriculture, and related disciplines, and the general public. Research in I-FARM has been disseminated to various audiences from local and national media to presenting at several conferences and workshops. Through these engagements, I-FARM has trained farmers, industry, and academic scientists with demonstrations of the technologies I-FARM is developing for a more sustainable future. I-FARM has targeted the public audience of our activities through news stories by local media. Additionally, we have targeted the national audience for our activities by being featured on the Big Ten Network for their segment on B1G Impact Research. We are collaborating with several companies and farmers with an interest in our research to demonstrate and adapt our approaches to meet their needs. Changes/Problems:Nothing to report What opportunities for training and professional development has the project provided?Students and post-docs: Twelvestudents and post-doctoral scientists work on I-FARM projects. The students and post-docs are spread across all the I-FARM research thrust with several of them working between agriculture and computer science disciplines. Extension: Extension successfully conducted a combined technology demonstration with presentations on the future of technology at UIUC on November 8, 2022 with 100 attendees participating from the farming community, industry, government and academia. In addition, offsite demonstrations were conducted at a University property in Ewing, Illinois, with 72 participants. Extension along with the Accessibility of Technology to Small-Holder and Minority Farming thrust, are hosting a demonstration at Tuskegee University at the annual Farmer's Conference. Outreach: This year there have been several outreach efforts with I-FARM Extension in farm visits, visits to local community colleges, and trade show participation. These interactions have helped in making connections with farmers and industry to develop techniques for adoption. How have the results been disseminated to communities of interest?Engagement with Research Community: Research has been disseminated to academic communities of interest through conference presentations in a broad range of disciplines. For example, Bin Peng from Thrust 1 presented at the fall American Geophysical Union (AGU) conference. Engagement with Industry: Members of the I-FARM team have had active discussions and meetings with personnel from AGCO, Cargill, EarthSense, Indigo Ag, and Verizon. In addition to active discussions, I-FARM has developed an industry advisory board with personnel from EarthSense, Verizon, Ingio Ag, John Deere, Microsoft, Grain Weevil, Corteva, Cargill, Farmer Business Network Inc, AGCO, and AIGEN. This board will serve as a working group for the project to provide industry insights for technology development. EarthSense is closely engaged with Under-Canopy Robotics for Sustainable Farm Management thrust. We are collaborating with the company for improving the vision based autonomy for the row following. Furthermore, the company provided multiple robots and logistics services for planting the cover crops. Engagement with Farmers: Members of the I-FARM team are having active discussions with local and national farmers on socioeconomic technology adoption. To guide this discussion, a Farmer Advisory Board was developed for input from farmers on the technology and ways to adopt it. The board encompasses a variety of small and large farm holders from local and national. Engagement with the Media: I-FARM has engaged with the media with two news stories from the Champaign-UrbanaNews Gazette and FarmWeekNow. Additionally, the Big Ten Network's B1G Impact Research developed a thirty-second research highlight that appears on the BigTen Network during commercial breaks. What do you plan to do during the next reporting period to accomplish the goals?I-FARM has made significant progress on our goals in the first year. We will continue to strive for more connections across thrust in Year Two. In this season, we will focus on the broad plans for Year Two, spanning the overall project. Thrust 1 will continue the ground, airborne and satellite data collection in Year Two. From this data, researchers will be able to generate spatial maps of crop traits using airborne and satellite remote sensing data. Additionally, drainage tiling will be added to the test-bed to allow for additional experiences with soil sensor monitoring to be done. The tiling facility will facilitate studying several replications of cover crop, no-till and animal integration. Thrust 2 plans to deploy one person controlling multiple robots using the network connectivity setup on the farm. The network connectivity will also enable a remote operator to correct the robot from an unexpected incident instead of following the robot as done currently. We plan to work on improving autonomy speed. The robots themselves are capable of reaching a maximum speed of fivemph, and improving autonomy speed will help us achieve planting efficiency. Thrust 3's goal is to evaluate the deployed connectivity solutions in the context of different agricultural applications and overcome the shortcomings using edge computing techniques. To evaluate the deployed connectivity, they will enable new types of connectivity, e.g., direct-to-satellite IoT connectivity. Once the evaluation has been completed, the next step is to get feedback on and improve the MyFarm app with new features such as mobile phone compatibility, better visualizations, and multi-modal data integration. Thrust 4 will implement outdoor sensor evaluation. For this evaluation, commercially available wearable sensors for cattle were put together. These sensors will be purchased and tested for applicability in both indoor and outdoor settings. UAV images will be tested for outdoor applications, as well as images from stationary cameras. Points of evaluation will include efficiency, autonomy (memory and battery), and usability. Thrust 5 will continue to work closely with Extension on the efforts for technology adoption for minority farmers. For this adoption to happen, thrust 5 and Extension will continue to host demonstrations with farmers to teach them about the technologies and how they can be implemented into their farming practices. Thrust 6 plans to disseminate the survey that was produced in Year Oneto elicit farmers' willingness to adopt different nitrogen management practices without any incentives or with different types of incentives when they have different risk preferences. Extension will continue to host on-site Demonstrations. We are currently working with our thrust partners in hosting three on-site demonstrations on autonomous technologies, two at UIUC and one at Tuskegee University. This will be our inaugural demonstration where a majority of our thrusts will do live demonstrations. I-FARM University: With the completion of our factsheets, we aim to launch the I-FARM University portal that will provide farmers, Extension and others with information on autonomous systems and needed infrastructure. We will also aim to record twotraining presentations and include pre and post tests that can be available for the general public to take. We also aim to start developing one or twovirtual reality modules that will be used in our demonstration programs.

Impacts
What was accomplished under these goals? Accomplishments in Integrating Remote-Sensing and IoT with Farm Planning and Management (Thrust 1) Set up eightsoil moisture stations (each with tworeplicates, 16 total vertical profiles). We set up both conventional and digital camera systems: conventional (onedownward + eightupward), and digital (onedownward + twoupward). Collected airborne hyperspectral imaging and UAV imageries during the growing season of 2022. Used a new method developed by our group to monitor leaf area index (LAI) and green leaf area index (GAI) New monitoring method included downward viewing cameras mounted above crop canopy for LAI/GAI measurements when the crop is short, and upward viewing cameras mounted below crop canopy for LAI/GAI measurements when the crop is tall. Computer vision methods were used to calculate the view zenith angle of individual pixels, segment images into vegetation and background (soil or sky), quantify gap fractions as a function of the view zenith angle, and estimate green fractions of the crop. Accomplishments in Under-Canopy Robotics for Sustainable Farm Management (Thrust 2) Redesigned a cost-effective cover cropping robot: In close collaboration with our industry partner, we redesigned a cost-effective cover cropping robot that uses cameras instead of two LiDARs, making it less expensive while maintaining its performance. Improved the multi-camera vision-based row-following algorithm: We worked on improving the multi-camera vision-based row-following algorithm, which is crucial for the robot's accurate and efficient planting. Validated the algorithms on the fields: We validated the algorithms on the fields, and the results were satisfactory. Successfully planted fortyacres of the digital farm using robots: We planted 38 acres with Cereal Rye and twoacres with Oats from early August to mid-September and had emergence before the harvest in late October. Reduced interventions: Initially, we had an average of fifteeninterventions per 1,140 feet long row, but with each planting date, the number of interventions reduced, and we had less than 5fiveinterventions on the last planting date. Seeded at 35-40 pounds per acre seeding rate. Accomplishments in Rural Internet Connectivity and Edge Computing (Thrust 3) Divided connectivity infrastructure along two axes: (a) Connectivity to the farm, i.e., internet connectivity solutions, and (b) Connectivity within the farm to extend the internet connectivity for on-farm devices Fiber Connectivity: These endpoints enable easy access to high-speed fiber connectivity for all the data needs of the farm. The fiber endpoints can be used to support local Wi-Fi, CBRS, and other connectivity solutions. Satellite Connectivity: Broadband satellite networks are an emerging connectivity solution, targeted toward rural areas. Additionally, we have acquired and tested a Starlink terminal for satellite connectivity on the Farm of the Future. Our work has helped quantify the variability in satellite networks for on-farm applications. CBRS: Citizen Band Radio Services (CBRS) is a new technology aimed at allowing public entities to operate cellular-like networks that offer high bandwidth connectivity. Trails still start in Year Two. This trial, in partnership with Celona, will help us evaluate the use of CBRS for precision agriculture. LoRa: LoRa enables long-range connectivity for low bandwidth data transmissions (few Kbps). We have set up a LoRa gateway on the farm of the future for IoT (Internet of Things) and sensor workloads. CBRS supports high-bandwidth data transfer and LoRa supports low-power long-range data transfer at low bandwidths. Together, they will enable a variety of connected farm applications (smart robotic cover crop planting, animal monitoring, soil monitoring, etc.) on the farm of the future. Accomplishments in Integrating Animals in the Farm of the Future (Thrust 4) Image dataset of beef and dairy cattle (over 2,000 animals) was acquired. Both surveillance and depth cameras were used for data collection. Animal information on behavior, Body Condition Score (BCS), Body Weight (BW), Docility Score, Locomotion Score, carcass information, age, sire line, pregnancy status, and feed intake was acquired according to relevance for each round of acquisition. Images and videos were partially annotated and will continue to be annotated over the next few months. This information will be correlated with animal information in the next steps of this project for the validation of image models for the prediction of animal behavior, ID, BW, BCS, carcass quality, feed intake, locomotion score, and docility score. Different camera sensors were evaluated for their performance. Stereo vision cameras (Intel RealSense) outperformed Time of Flight (ToF) cameras (Microsoft Kinect v2) for applicability. Surveillance cameras seem to be a good solution for 2D image model development for indoor applications. Accomplishments in Accessibility of Technology to Small-Holder and Minority Farming (Thrust 5) Integrated the multispectral drone and two TerraSentia field rovers to enhance the university's research and Extension infrastructure. Using drone imaging software, we have mapped crop production locations and are learning about mission planning for the TerraSentia plant phenotyping rovers. Accomplishments in Socioeconomics of Technology Adoption (Thrust 6) Examining the impact of different policy instruments or incentives on the adoption of Maximum Return to Nitrogen (MRTN) calculator and Variable Rate Technology (VRT). Developed a survey to disseminate to farmers in Year Twoon the adoption of MRTN and VRT. Accomplishments in Extension and Research Integration Efforts Successfully conducted a combined technological demonstration with presentations on the future of the technology at UIUC on Nov 8th. About 100 attendees participated coming from the farming community, industry, government and academia. Demonstration at a University property in Ewing, Illinoiswith about 72 attendees We have conducted sevenfarm visits based at local community colleges with about 185 attendees. Lastly, we participated in nineregional trade shows and conferences based in Illinois and received about 475 attendees. Regarding our work on I-FARM University, we have completed the development of threefactsheets Connectivity Solutions, Under-Canopy Robotics Systems, and Integrating Animals into Future Farming practices.

Publications

• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: B Peng, Y Yang, K Guan, M Pan and M Cosh. 2022. A Multi-scale Soil Moisture and Temperature Observation Network over Cropland in Central Illinois. AGU Fall Meeting, December 11-15, 2022. Chicago, Illinois.