Progress 06/15/22 to 06/14/23
Outputs
Target Audience:During this reporting period, our project efforts have reached a variety of target audiences across different platforms and activities: Commercial Breeders, Technology Specialists and AgTech Companies: In March, 2023, we shared information about the FieldDock project with commercial breeders and technology specialists with a presentation at the Precision Breeding Academy at Bayer Crop Science's Global Breeding Organization. As an invited panelist at the InfoAg Conference in July 2022, PI Shakoor and project team members also reached a target audience of diverse agtech and precision agriculture companies. Scientific Faculty and Graduate Students: PI Shakoor connected with scientific faculty and graduate students as an invited speaker at multiple conferences and seminar series, including the UC Berkeley Plant Gene Expression Center Seminar Series, the ASA - CSSA - SSSA Annual Meeting, the University of Missouri Interdisciplinary Plant Group Seminar Series, and the CROPS Conference at the HudsonAlpha Institute for Biotechnology. PI Shakoor also presented and discussed the FieldDock project at the Taylor Geospatial Institute Research Day at Saint Louis University in April 2023. K-12 and General Public: In May 2023, the FieldDock team hosted a demonstration booth at the Danforth Center PlantTech Jam, a large community outreach event focused on robotics, engineering, and plant science. The FieldDock team showcased project activities and developed interactive displays with field drones and environmental sensors. We engaged with over 500 attendees of all ages, providing them with the opportunity for hands-on science and engineering experiences. Overall, our project's outreach efforts have successfully reached diverse sectors of society, ranging from students at different education levels, to professionals in academia and industry, and the broader public. Changes/Problems:In the project's initial two years, progress was significantly hindered by global supply chain disruptions due to COVID-19. Several vital FieldDock hardware system components were either unavailable or challenging to source for many months. However, we've since found alternative parts for the initial FieldDock prototype. COVID-19-related on-site work restrictions also significantly delayed hiring key project personnel in 2020 and 2021. The granted No-Cost Extension (NCE) has proved crucial, allowing us to recruit two early-career engineers to assist in completing project tasks. Despite the early Covid-related project delays, we're making steady progress and remain confident in achieving our project goals. What opportunities for training and professional development has the project provided?Through participation in several conferences and seminars in the field, this project has provided training and professional development for four postdoctoral researchers, one graduate student, one early career engineering research scientist, and one masters level engineering research associate. Additional training material for members of the project team has included online coursework for drone pilot certifications and career development workshops. How have the results been disseminated to communities of interest?During the reporting period, the results of the FieldDock project have been disseminated fairly extensively to various communities of interest through a multi-faceted approach, to not only inform about the project's progress but also to inspire interest in science and technology. Scientific and Academic Communities: In addition to publications, the project's findings were disseminated to scientific faculty, graduate students, and other academics through presentations and talks at various events. These included the UC Berkeley Plant Gene Expression Center Seminar Series, the ASA - CSSA - SSSA Annual Meeting, the University of Missouri Interdisciplinary Plant Group Seminar Series, and the Taylor Geospatial Institute Research Day, among others. Industry and Technology Specialists: The FieldDock project's relevance to AgTech and precision agriculture companies was highlighted at the InfoAg Conference and the Precision Breeding Academy at Bayer Crop Science's Global Breeding Organization. These activities provided an opportunity to discuss the project's latest developments with industry professionals and gain valuable feedback. Public Outreach and Education: The most significant outreach effort undertaken was the organization of a demonstration booth at the Danforth Center PlantTech Jam in May, 2023. This event, attended by over 500 participants of all ages, provided an interactive platform to explain the project's objectives and methodologies, demonstrate the use of drones and sensors, and inspire interest in science and technology. What do you plan to do during the next reporting period to accomplish the goals?The final phase of the FieldDock project will necessitate a comprehensive end-to-end systems integration, demonstrating interoperability between front-end/back-end software, user-defined parameters entered into the GUI, and reliable remote connectivity to cellular networks. The integration will also encompass autonomous flight mission planning at the edge by the system controller, a fully functional drone garage with robust electromechanical components capable of successfully deploying a drone, and a custom drone with a custom payload that can execute an autonomous flight mission. The process includes imaging the field based on user preferences, successfully returning to FieldDock, landing autonomously, offloading raw imaging data via WiFi to the edge computer in the drone garage, and performing an edge-computed image processing procedure. This procedure derives the necessary data products that can provide guidance for real-time breeding -- all powered by solar energy. Each of these components has been successfully demonstrated or has been deemed feasible based on the results of research and development to date. The development of the front-end/back-end is nearly complete and is functioning as designed. Remote mission planning will be developed and deployed by utilizing already widely-adopted, de-risked technologies such as QGroundControl, MavLink, and Pixhawk. Remote connectivity to cellular networks can be achieved using global SIM cards, or multi-carrier SIMs, to auto-connect to the best available network at any given moment. Companies like SIMetry offer a Tier 1 multi-carrier SIM that connects to Verizon, AT&T, and T-Mobile -- the three largest networks in the US -- as well as global SIMs that connect to the dominant networks in over 190 countries. The integration of a cellular modem has already been accomplished in this project and continues to send/receive data without issue. The drone garage will be one of the last components to be developed, as well as the power generation/distribution system, because there are still components of the FieldDock system that need development, and all related power requirements must be defined. However, a drone garage concept has already been fully designed and built to better understand the complexities of such a structure. This concept tested various electromechanical components, such as using a single "lead screw" to drive multiple moving parts, enabling the drone to take off and land safely from the platform/landing pad. The individual steps of the edge computer image processing pipeline have all been developed and successfully demonstrated. Work is currently underway to combine these processes into a single process, with completion anticipated soon. The main challenge for this FieldDock component is power consumption. The FieldDock team is exploring the integration of Field-Programmable Gate Arrays (FPGAs) to offload as much image processing work from the Nvidia Nanos as possible to conserve power. If integration proves too complex and resource-intensive, a larger power system/photovoltaic array will need to be installed at each FieldDock site to handle the power requirements of charging the drone, powering the electromechanical hardware components and edge processing, sensor measurements, network connectivity, and cloud data reporting.
Impacts
What was accomplished under these goals?
During the FieldDock project's third year, we made major adjustments in the development of fundamental components to better accommodate autonomous systems and enhance complex feature interoperability. Consequently, we hired an early-career engineer and a master's level associate at the Danforth Center. These strategic shifts improved and expedited the development of key features. We've enhanced hardware and software interoperability, streamlined system communication, and made considerable progress in autonomous flight and landing. We've successfully identified and tested all wired sensors, which are now fully functional. The development of wireless sensors, based on early research and development results from the FieldDock project, has commenced and is progressing rapidly. We expect to start testing these wireless sensors early in the upcoming project year. Further, a novel system controller, which manages all sensor-based technology, has been developed and validated through numerous indoor and outdoor experiments. We've begun developing a new iteration of the FieldDock backend cloud infrastructure, designed for flexibility in managing diverse data types, including incoming sensor data, fixed imaging data, processed drone flight data, system diagnostic data, and all other data captured by FieldDock. We are concurrently building a new FieldDock GUI, that is currently being integrated with the backend. This frontend software offers users control and a basic view of FieldDock features. In addition, we're developing a backend management interface for FieldDock administrators, and enhancing the frontend with additional data visualization components to improve user interaction with sensor data. FieldDock Sensors:Ambient air measurements will now be recorded at the FieldDock station using a high-accuracy central air sensor, with remote air measurements enabled by wireless sensors. These wireless sensors will measure the same parameters as the FieldDock station, serving as a means of validation and comparison between conditions outside and inside the crop canopy. Our standard air measurements include temperature, relative humidity, barometric pressure, and LUX for relative intra-canopy light measurements. Wind measurements will be recorded using a high-accuracy ultrasonic anemometer, and rainfall will be gauged by two high-accuracy rain sensors. Indirect validation of the rain measurements will be done by comparing them with soil moisture measurements at the FieldDock station and in the field where wireless sensors are deployed. Soil probes at the FieldDock station and those connected to the wireless sensors will provide detailed data on local soil moisture, temperature, and salinity conditions in each experimental plot. A unique system controller has been developed with a "Root" (Master) and "Leaf" (Slave) architecture. This controller will manage all FieldDock sensors and allow for easy integration of new sensors and an increase in the number of sensors reporting back to the nearest FieldDock. The system controller uses a combination of single board computers (SBC) for the Root and microcontrollers for the Leaves, facilitating an expandable, flexible hardware/software architecture. We have developed and successfully deployed prototypes for this system architecture. It can be utilized not only for remote sensor stations reporting back to the nearest system controller, but also for various electromechanical hardware components of the FieldDock Drone Garage. FieldDock Station Hardware and Power Distribution and Management:The overall design and development of the FieldDock structure, including the sensor tower and drone garage, was paused for the majority of the third year due to a shift in our approach to several key FieldDock system components. Our focus shifted to the edge image processing pipeline and the hardware it requires, system controller and sensor network, system interoperability, and the development of both the frontend and backend as well as autonomous drone flight. We also put the development of power distribution systems and hardware on hold as we focused on developing other key components. We have recorded power requirements for all new systems to guide the development of a remote solar power system in Year 4 (current NCE phase). Image processing at the Edge:Multispectral imaging data from the drone is now being automatically extracted and processed on the edge computer. Processes such as orthorectification, radiometric calibration, mosaicking, auto plot segmentation, calculation of plant multispectral indices, and automated plot scoring, which are all parts of the broader edge computing image process pipeline, have been individually developed, tested, and proven to be largely successful. We are currently working on combining these processes into a single, automated pipeline. This is being developed and tested with various edge computing concept systems. Each edge computing concept system leverages both novel (FPGA) and existing (Nvidia Nano Jetson) technologies, enabling us to identify trade-offs and determine the best path forward for achieving an optimal power consumption-to-graphics processing speed ratio. UAV Hardware:In Year 3, we decided to employ custom drone frame kits to expedite and enhance the testing of various facets of autonomous flight during the Software in the Loop (SITL) and Hardware in the Loop (HITL) testing phases. The use of small, affordable drone kits, adaptable for varying payload weights and types, has significantly accelerated the development of autonomous drone flight. The drone payload has largely remained as identified in the second year, with the exception of the wireless charging. The integration of the WiBotic charging platform was not successful because the power transfer between the transmitter and receiver requires extremely precise alignment and proximity. These requirements are not feasible for remote autonomous operations, primarily guided by GPS coordinates, and in some experiments, RTK. Consequently, we are currently researching and testing different methods for drone battery charging, which potentially include automating a mechanical connection between the drone and the charger upon landing. Autonomous Avionics Development:We have achieved success with autonomous drone landing under various heights, approaches, and outdoor conditions (high vs. low wind). This progress will serve as a foundation for full flight mission planning and execution. Testing on user-defined mission flight parameters through the GUI will soon commence. These parameters will be sent to the system controller on FieldDock via cellular connectivity. Subsequently, a mission will be autonomously planned and executed based on several factors, including GPS coordinates of the field of interest, size of the field, required altitude for desired imaging, drone battery life in relation to flight time, landing, wireless data offloading to the edge computer, and weather conditions. FieldDock Graphic User Interface (GUI): TheFieldDock GUI has been redesigned and redeveloped for better compatibility, serviceability, and ease of optimization with contemporary cloud infrastructure modules, software, and services. The previous GUI concept, which utilized a relatively unknown and infrequently used framework called Phoenix/Elixir, was initially chosen due to its purported reliability and inherent security. However, due to its limitations and developmental challenges, it was decided to abandon this framework for a more comprehensible one with a broader array of online resources. The original FieldDock GUI, constructed with HTML, CSS, and JavaScript, was repurposed and seamlessly integrated into the React framework. All intended features are operating as designed, and a substantial portion of the data collected by the drone and remote sensors is now streaming into the GUI.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Dilmurat, Kamila, Vasit Sagan, and Stephen Moose. "Ai-Driven Maize Yield Forecasting Using Unmanned Aerial Vehicle-Based Hyperspectral and Lidar Data Fusion." ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences 3 (2022): 193-199.
- Type:
Journal Articles
Status:
Accepted
Year Published:
2022
Citation:
Bhadra, S., et al. "Automatic Extraction of Solar and Sensor Imaging Geometry from Uav-Borne Push-Broom Hyperspectral Camera." ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences 3 (2022).
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2023
Citation:
Gano, Boubacar, Nurzaman Ahmed, and Nadia Shakoor. "Machine learning-based prediction of sorghum biomass from UAV multispectral imagery data." 2023 4th International Conference on Computing and Communication Systems (I3CS). IEEE, 2023.
|
Progress 06/15/21 to 06/14/22
Outputs
Target Audience:Due to covid restrictions in 2021, extension and outreach efforts were mostly limited to virtual event presentations. PI Shakoor presented and discussed the FieldDock project at the Ohio State Graduate Research Symposium hosted by the Horticulture and Crop Science Graduate Student Association (April, 2022). The project was also presented at the CROPs conference in Huntsville, Alabama in June, 2022. Both of these invited presentations were aimed towards project outreach and wereattended by a diversity of scientific trainees (graduate students and postdoctoral associates). In May, 2022, PI Shakoor hosted 3 young women high school students from St. Joseph's Academy in Saint Louis, MO for their 2022 Senior Service Projects. These students volunteered in the Shakoor lab for 80 hours each and gained quite a bit of hands-on experience in plant phenotyping, engineering and research. One young minority woman from the service project stayed on for a paid summer internship and is currently working in the lab on this project and related research. PI Shakoor is also currentlyhosting a summer REU female undergraduate student intern for 2022. Changes/Problems:The project was and continues to be significantly challenged by the global supply chain disruptions. Several critical components of the FieldDock hardware system were unavailable and continue to be very slow to procure. We have started sourcing alternative parts that we can use for the initial prototype version of the FieldDock, however these systems will not have optimal design-to-manufacture features. As part of our TTP goals, we are using design-to-manufacture principles to ensure that the FieldDock prototype built by the end of the project will be optimized for commercial development. Our engineers have been in contact with component manufacturers who expect these supply issues to ease somewhat in late 2022 but there will be shortages off and on into 2023 and 2024 for some components. What opportunities for training and professional development has the project provided?Through participation in several conferences and seminars in the field, this project has provided training and professional development for four postdoctoral researchers, one computational scientist, one research technician and several summer interns. How have the results been disseminated to communities of interest?Due to covid restrictions in 2021, extension and outreach efforts were mostly limited to virtual event presentations. PI Shakoor presented and discussed the FieldDock project at the Ohio State Graduate Research Symposium hosted by the Horticulture and Crop Science Graduate Student Association (April, 2022). The project was also presented at the CROPs conference in Huntsville, Alabama in June, 2022. Both of these invited presentations were aimed towards project outreach and were attended by a diversity of scientific trainees (graduate students and postdoctoral associates). In May, 2022, PI Shakoor hosted 3 young women high school students from St. Joseph's Academy in Saint Louis for their 2022 Senior Service Projects. These students volunteered in the Shakoor lab for 80 hours each and gained quite a bit of hands-on experience in plant phenotyping, engineering and research. One young minority woman from the service project stayed on for a paid summer internship and is currently working in the lab on this project and related research. PI Shakoor is also currently hosting a summer REU female undergraduate student intern for 2022.Results of this project have also been disseminated via presentations at conferences and webinars, including the Salk Institute Suberin Club Seminar Series (Mar, 2022) and CROPS Conference (June, 2022). What do you plan to do during the next reporting period to accomplish the goals?In year three of the project, we will aim to make significant progress in the integration of the FieldDock's edge computing device for GWAS/Modeling. The FieldDock prototype and minimally viable product (Version 1) will be nearly complete as well. We will continue FieldDock system optimization for autonomous UAV development and inter-platform data analysis, and the first autonomous FieldDock flight is planned for September, 2022. We will establish the data transfer and feature extraction pipeline to process raw data and generate derived data products, and we will generate calibrated and mosaicked data processing methods for the drone collected imagery.
Impacts
What was accomplished under these goals?
At the end of year 2, the project is well on its way to successfully developing a prototype 'smart farm' system, or FieldDock. The project is on track to conduct its first autonomous flight in a crop research field in September, 2022. If successful, the FieldDock project will validate the possibility of remote autonomous UAV platforms capable of capturing and processing large amounts of data daily/weekly facilitated entirely by solar power. This weekly capture of unbiased information will give tremendous insight into how crops respond to environmental conditions in real time throughout their life cycle. Researchers and crop breeders will have direct and daily access to remote field and crop conditions that could accelerate and improve global crop breeding strategies.The development of FieldDock's capabilities, namely UAV technical specifications and power needs, solar charge and power requirements, onboard edge computational hardware/software and wireless communications have been progressing steadily. UAV Garage Structure: Design/development of the UAV garage has also begun and the initial prototype will be made of 80/20 aluminum. This will allow the UAV garage to be easily adjusted without having to do a complete redesign in the event of a UAV model/size change later in development. The garage will open as a "drawer" with a scissor lift to elevate the UAV platform up above the UAV garage roof line to ensure the maximum amount of UAV landing surface area, while maximizing the efficiency and simplicity of all moving parts. Minimizing power needs for these moving parts is essential due to the system's solar power requirements.Development and integration of the UAV garage components such as the Edge computer, Drone induction charger, Hanger controls, Edge computer battery box, Drone induction charger battery box, and the Hanger controls battery box are ongoing and nearly ready for a full field test. Power distribution and management testing: Charge controller components crucial to operations are still on backorder until a later date. Delivery is expected July 2022.Lithium ion battery size/type and 100W solar panel selection have been identified and a custom mount has been developed and integrated to the roof of the UAV garage structure. Custom PCB's for solar charge control and overall system power distribution have been designed and are scheduled for delivery in July 2022. UAV communication integration with FieldDock software/hardware and testing: Telemetry collection software is complete. The FieldDock team is able to receive and record autopilot and mission telemetry via WiFi during test flights. UAV Payload Tests: Static drone tests are in process. These include testing payload power supply (+5V for mission computer, +12V for Micasense Altum camera), using a Wibotic charger to charge the drone battery, and optimizing the location of the payload CG (Center of gravity) relative to the drone CG. Sensor integration and testing: Due to supply chain shortages and time restrictions (weather conditions conducive to flight tests), the FieldDock engineers are focused on autonomous flight requirements in order to meet our project deadline for the first autonomous flight. Further integration of other sensors will happen in year 3 once other system components are brought online. Progress on the wireless sensor network is yet to be determined. Autonomous avionics development: Hardware in the loop (HITL) testing with the Aegis UAV has been ongoing and will continue up to and beyond the in-field autonomous flights when they begin in September 2022.Software development continues on the Mission computer and the MissionControl software, as well as the integration of system components such as the Nvidia Nano edge computer that will retrieve images from the Micasense multispectral camera, and the System Controller that will manage five separate daemons, BackendComms, RS485Comms, FlightOps, Camera, and SysLocalComms, and all communications with the FieldDock Backend. The primary focus here recently has been on MissionControl and a standalone telemetry app in order to prepare for a real flight test, completely under computer control by September 2022. Work also continues on BackendComms and the FieldDock Backend as we continue to add functionality to both ends to transfer software updates and download log files. Graphic User Interface: An updated, optimized version of the Graphic User Interface (GUI) for FieldDock has been completed and shared with the team. Overall design responsiveness has been optimized for multiple breakpoints on multiple platforms (smartphone, tablet, desktop, etc.). Additional items on multiple pages were built into the design that will act as a placeholder for future features. Completion of these features will largely depend on time and resource availability beyond the scope of the project priorities. Further, flight configuration parameters via the GUI are in progress. The configuration parameter settings are data that do not change very often, but may need to be configured for specific installations. Certain read-write parameters can be changed by authorized users, and these parameters are grouped into System Configuration, Measurement Configuration, Pre-Flight Operations Sequence, and Post-Flight Operations Sequence settings.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Sagan, Vasit, et al. "Data-driven artificial intelligence for calibration of hyperspectral big data." IEEE Transactions on Geoscience and Remote Sensing 60 (2021): 1-20.
|
Progress 06/15/20 to 06/14/21
Outputs
Target Audience:Due to covid restrictions in 2020-2021, extension and outreach efforts were limited to virtual event presentations. In February 2021, the FieldDock project was presented at a Science in St. Louis Public Seminar series hosted by the Academy of Science Saint Louis, St. Louis County Libraries and Zonta Club of St. Louis's Girls Can STEM Initiative. The presentation was recorded for YouTube and shared with the Saint Louis Science Center and their YES Teens youth group. There were a total of 85 recorded participants, with representation from Saint Louis community residents and primary participation from middle and high school students. PI Shakoor also presented and discussed the FieldDock project at two student career panels including the NSF REU Graduate Student Career Panel (July, 2021) and BioBash: Exploring Scientific Career Paths (May, 2021), both hosted by the Donald Danforth Plant Science Center. The REU presentation was made to a group of approximately 30 undergraduate students pursuing a diversity of STEM careers. The BioBash event was attended by scientific trainees (graduate students, technicians and postdoctoral associates) at the Danforth Center and local agtech and biotech companies. Changes/Problems:?The project was and continues to be significantly challenged by the global supply chain disruptions. Several critical components of the FieldDock hardware system were unavailable and continue to be very difficult, if not impossible to source. For example, one of the critical processors (STM32L072RZT6 ARM32) and transceiver (THVD1450DGKR) of the main environmental sensor stack continue to be unavailable at the time of this report. We have started sourcing alternative parts that we can use for the initial prototype version of the FieldDock, however these systems will not have optimal design-to-manufacture features. As part of our TTP goals, we are using design-to-manufacture principles to ensure that the FieldDock prototype built by the end of the project will be optimized for commercial development. Our engineers have been in contact with component manufacturers who expect these supply issues to ease somewhat in late 2022 but there will be shortages off and on into 2023 and 2024 for some components. What opportunities for training and professional development has the project provided?Through participation in several conferences and seminars in the field, this project has provided training and professional development for three postdoctoral researchers and one research technician. How have the results been disseminated to communities of interest?As described in an earliersection, due to covid restrictions in 2020-2021, extension and outreach efforts were limited to virtual event presentations. In February 2021, the FieldDock project was presented at a Science in St. Louis Public Seminar series hosted by the Academy of Science Saint Louis, St. Louis County Libraries and Zonta Club of St. Louis's Girls Can STEM Initiative. The presentation was recorded for YouTube and shared with the Saint Louis Science Center and their YES Teens youth group. There were a total of 85 recorded participants, with representation from Saint Louis community residents and primary participation from middle and high school students. PI Shakoor also presented and discussed the FieldDock project at two student career panels including the NSF REU Graduate Student Career Panel (July, 2021) and BioBash: Exploring Scientific Career Paths (May, 2021), both hosted by the Donald Danforth Plant Science Center. The REU presentation was made to a group of approximately 30 undergraduate students pursuing a diversity of STEM careers. The BioBash event was attended by scientific trainees (graduate students, technicians and postdoctoral associates) at the Danforth Center and local agtech and biotech companies. What do you plan to do during the next reporting period to accomplish the goals?In year two of the project, we will engage in the discovery phase of minimal computing resources needed for the GWAS/modeling scripts and aim to make significant progress in the integration of the FieldDock's edge computing device for GWAS/Modeling. The FieldDock prototype and minimally viable product (Version 1) will be nearly complete as well. We will continue FieldDock system optimization for autonomous UAV development and inter-platform data analysis, and the first autonomous FieldDock flight is planned for September, 2022. We will establish the data transfer and feature extraction pipeline to process raw data and generate derived data products, and we will generate calibrated and mosaicked data processing methods for the drone collected imagery. Project Impact:Successful development of remote autonomous UAV platforms pose many challenges. The FieldDock system requires that enough solar power be captured and stored in order to regularly facilitate and power a fully charged UAV battery capable of ~20 minutes of flight with a reasonable sensor payload, wireless and wired communication of various system operations, wireless transfer of large datasets and imaging files, edge processing of large imaging files, image processing and downsampling of large image files, environmental sensor measurements at a high temporal resolution, operations of all electro-mechanical components, and cellular connectivity to the cloud. There are also many challenges associated with remote autonomous flight that the FieldDock project is addressing such as navigation, plot identification, and UAV landing/taking off safely aided by computer vision techniques, just to name a few. If successful, the FieldDock project will validate the possibility of remote autonomous UAV platforms capable of capturing and processing large amounts of data daily/weekly facilitated entirely by solar power. Remote infrastructure requirements for plant phenotyping equipment and the need for manual human-collected measurements will no longer be necessary. This daily capture of unbiased information will give tremendous insight into how crops respond to environmental conditions in real time throughout their life cycle. Researchers and crop breeders will have direct and daily access to remote field and crop conditions that could accelerate and improve global crop breeding strategies.
Impacts
What was accomplished under these goals?
At the end of year 1, the project is well on its way to successfully developing a prototype 'smart farm' system, or FieldDock. The project is on track to conduct its first autonomous flight in a crop research field in September, 2022. The FieldDock's onboard sensor suite, UAV technical specifications and power needs, solar charge and battery capacity requirements, onboard edge computational hardware and gateway hardware have been identified and are currently being sourced and/or developed. Sourcing of FieldDock onboard sensors: FieldDock mechanical and electrical infrastructure has been largely designed and will be developed to be modular. Modularity will allow the easy swapping of system component replacements as well as accommodate the integration of existing and/or any new 3rd party sensors for the foreseeable future. Universal connectors for the FieldDock system have been developed allowing custom sensor enclosures, 3rd party sensors and other ancillary system components to connect anywhere on the system bus. Temperature, humidity and air pressure sensors have been selected and will be housed in a small, custom waterproof enclosure that will allow for optimum ambient air measurements. Multiple instances of this "sensor module" will be connected to the FieldDock system in order to validate sensor accuracy including long term sensor drift as well as demonstrating reliability and resiliency to measurements crucial to flight command. This sensor module will also be used for the wireless sensor network, taking further advantage of the developed electronic architecture.Wind, rain, PAR and soil sensors have also been identified and acquired for integration. Design/development of aluminum hardware for FieldDock shell: The FieldDock physical mechanical structure will be made of aluminum and powder coated for long term weather and corrosion resistance. All enclosures are being designed to ensure no ingress will occur of water or other natural elements that could compromise internal electronics.The pole structure that all environmental sensors, battery boxes and the system controller/gateway mounts to has been fully designed and 5 prototype units have been manufactured. These units will be used to test form and fit of mechanical manufacturing specifications, water ingress in indoor/outdoor experiments, and the form and fit of the companion electrical components. Design/development of the UAV garage has also begun and the initial prototype will be made of 80/20 aluminum. This will allow the UAV garage to be easily adjusted without having to do a complete redesign in the event of a UAV model/size change later in development. Likely, the garage will open as a "drawer" with a scissor lift to elevate the UAV platform up above the UAV garage roof line to ensure minimum water ingress while maximizing the efficiency and simplicity of all moving parts. Minimizing power needs for these moving parts is essential due to the system's solar power requirements. Design/development of system electronics: Many iterations of various FieldDock electrical system PCB's have been developed and tested such as sensor module boards, cellular modem daughter boards (for compatibility with the system controller), battery pack and solar charge controller board, sensor connector board (junction board) and signal conditioning adapter modules for sensor plug-and-play functionality. Power distribution and management testing: charge controller components crucial to operations are still on backorder until a later date. Small quantities of these components acquired before the global shortage are being used to continue development and prototyping. However, the final FieldDock system power distribution system will require additional numbers of these components. Delivery is expected Q2 2023. Lithium ion battery size/type and solar panel selection have been identified and acquired for development. Custom PCB's for solar charge control and overall system power distribution have been designed and multiple prototype iterations have been acquired and tested. UAV communication integration with FieldDock software/hardware and testing: The FieldDock edge computer system will extract images from the MicaSense Altum camera via the camera's HTTP interface over WiFi. From here, the edge computer will perform the data reduction, image mosaicking, field plot auto-segmentation and vegetative index calculations. The UAV mission computer (Pixhawk) will receive its mission file from the system controller via WiFi as well through the Mavlink serial interface. It will control the phases of flight and transmit mission telemetry while in flight during the testing phase only. Sourcing of UAV hardware: Aegis (formally Emergent RC) Intense Eye V2 Quadcopter became the choice for our preliminary drone testing/prototyping due to its compatibility with WiBotic's wireless battery charging technology. While we test our SITL (software in the loop) and eventually HITL (hardware in the loop) with this UAV, we will outfit it with temporary 3D printed components that have the same size, shape, and weight of the actual components we intend to integrate for safety purposes. Identification of payload sensors: The UAV payload with a MicaSense multispectral camera has been identified and sourced based on weight, physical size, power consumption, data resolution, project research requirements and its ability to integrate with the custom avionics box that will likely house a Pixhawk. A downwelling radiation sensor, WiFi/cellular antenna, wireless charging antenna and battery will also be mounted to the UAV. Sensor integration and testing: To date, our ambient air sensors (temperature, humidity, air pressure) have been integrated and undergone periodic testing. Further integration of other sensors will happen at a later date once other system components are brought online. Initial testing of the MicaSense Altum has been carried out during preliminary UAV flights conducted by Co-PI Sagan's group at Saint Louis University. Autonomous avionics development: Development of autonomous flight control software is underway in addition to the FieldDock electrical, power and computational system to ensure firmware and systems communications are consistent, well-understood and validated throughout the development phase. SITL and HITL testing with the Aegis UAV have begun and will continue throughout year 2 of the FieldDock project. Graphic User Interface: A preliminary working version of the Graphic User Interface (GUI) for FieldDock has been completed. The GUI is a web application demonstrating up-to-date responsive design principles so that users can view the GUI on a variety of device platforms (desktop, tablet, smartphone, etc.). The GUI will be hosted within the backend (AWS) and use the Phoenix web framework and Elixir coding language. With this initial prototype, software engineers can begin best practices for connecting all FieldDock collected data in the cloud database to the GUI for user interaction. As new FieldDock derived data products are successfully sent and stored in the cloud, they will subsequently be connected to features within the GUI for user viewing and downloading. Version 2 of the GUI will begin development in year 2 based on feedback from software engineers and is expected to be completed in Q4 2021.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Maimaitijiang, M., et al. "A fully automated and fast approach for canopy cover estimation using super high-resolution remote sensing imagery." ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences 3 (2021): 219-226.
|