Progress 08/15/24 to 08/14/25
Outputs
Target Audience:This project targets the following audiences: Agriculture industry in Utah. Sensors research community and industry. Robotics research community and industry. Smart agriculture research community. Agriculture Industry in Utah This is a key audience for two reasons. First, working with the local agriculture community provides a means of translation for the results of our work into real practice. Second, feedback from local farmers helps inform our research direction. For example, at a recent field day hosted by USU, several participants noted that automated tools for water management for small farms growing high-value crops are not readily available. This led to informative discussion on the reasons for this lack of availability and potential solutions. We primarily leverage the Utah State University (USU) extension offices and relationships to reach this community. We reach this audience through the extension efforts of PI Zesiger and through outreach events such as field days hosted at USU extension sites. Sensors Research Community and Industry Through interaction with the sensors research community, we hope to engage the community to work on opportunities in agriculture. Although there is a large research effort on agricultural sensors, presenting our work at more general sensors conferences such as IEEE Sensors and IEEE MEMS can spur the community to think about sensing solutions for smart agriculture. Secondly, we are developing innovative passive sensing solutions that could be profitably applied to a wide range of industries. Engagement with the sensors community provides a means for dissemination of our results that hopefully informs research advancements elsewhere. PIs Roundy and Young are both highly involved in the sensors community. Robotics Research Community and Industry As with the sensors research community, the dual reason for reaching the robotics community is to engage their efforts toward agricultural applications and disseminate the new methods we develop, particularly in the area of target localization which can be useful for across many robotics applications. As an example of our engagement with the robotics community, PI Leang recently arranged for the donation of a ground robot from Inert Products for our research. The supplier will benefit by getting a first look at the methods we develop using their robotic platform. PI Leang is highly involved in the robotics community through conferences such as the International Conference on Robotics and Automation (ICRA). Smart Agriculture Research Community The larger agricultural research community is clearly an important audience both for feedback on our work and further dissemination of our research findings and system development. This audience is reached primarily through presentations as conferences such as the annual meeting of the American Society of Agricultural and Biological Engineers (ASABE). PI Zesiger and his colleagues regularly interact with this research community. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?This project funds 1 PhD student and 1 MS student in mechanical engineering and 1 PhD student in electrical engineering. Additionally, one undergraduate student is working on the project. The PhD student and undergraduate student in mechanical engineering are working on the robotic drone and the localization system. The MS student in mechanical engineering is working on the soil water depletion model. He has developed several semi-custom soil sensors to gather data from this system. The PhD student in electrical engineering is working on the passive sensors. All students participate in bi-weekly group research meetings and all students regularly collaborate with PI Zesiger on designing and implementing field studies at the USU extension site in Kaysville Utah. In addition to their core disciplines, the students have learned extensively about the others' work as they need to closely collaborate. Furthermore, they are learning about agricultural technology and practice. The USU extension site hosts regular field days. On August 5th they held a field day for urban and small farmers. Approximately 100 people attended from the community. PI Zesiger is one of the primary organizers for the event. PI Roundy gave a presentation on his work and PI Young gave a demonstration on his work. How have the results been disseminated to communities of interest?To date, the primary means of disseminating results have been through submission of conference papers, presentations at those conferences, and presenting at the USU field day. We have submitted two conference papers and one conference poster presentation to IEEE Sensors, IEEE PowerMEMS, and Modeling, Estimation, and Control Conference (MECC). These conferences are focused on sensors (IEEE Sensors), small scale power devices (PowerMEMS), and robotics and control (MECC). These presentations will bring more awareness of agricultural research to these communities. As mentioned, PI Zesiger helped organize a recent field day at which PIs Roundy and Young presented some of their work. Although the research is not at a level of maturity that can be immediately used by field day attendees, the presentations expose them to forward-looking technologies. Importantly, the attendees had excellent and probing questions that will help inform the research. As noted, attendees mentioned that automated soil monitoring solutions for small farms are not available, which led to discussions on the primary reasons for this absence and potential solutions. What do you plan to do during the next reporting period to accomplish the goals?The overall goal for the next year is to complete a full system demonstration in the field. In addition, we plan to gather a larger set of soil moisture data that can be used to train and validate the soil water depletion (SWD) model. Then, in the third year of the project, we plan to deploy the system including an autonomous robotic drone and our passive soil moisture sensors. Using the data from the full system we will demonstrate the soil water depletion map as an output. During the upcoming year, the following tasks will be undertaken with regard to the passive soil sensors. The current design will be fully characterized and calibrated. We plan a revision to the design of the spherical sensor to be robust to orientation. For the above ground transmitter/reader, we plan to incorporate an off-the-shelf amplifier that can be run from a battery and be mounted on the robot. (Current tests are being performed with a benchtop linear amplifier that requires a 120 volt AC power supply.) Finally, custom data acquisition hardware needs to be designed and incorporated into the robot mounted system. This data acquisition unit will measure the current signal across the drive coil at very a very high rate (~ 50 MHz). This signal will both inform the final localization of the transmitter/reader and contain the information about the soil moisture encoded as a change in self-resonance frequency. The following tasks will be undertaken with regard to the robotic drone and localization. First, the localization algorithm will be fine-tuned and validated in the lab. The mobile robotic platform that we have contains a robotic arm that will hold and make final adjustments to the location and orientation of the transmitter/reader. The localization algorithm needs to be integrated with motion planning for the robotic arm. Once the system is fully mounted on the mobile robot, lab and field tests will be conducted to validate the ability of the system to locate buried sensors and collect accurate soil moisture readings. For example, we need to characterize the accuracy of the soil moisture readings versus the distance between the transmitter/reader and the sensing coil. This system demonstration is a key outcome for the upcoming year. The following tasks will be undertaken with regard to soil water depletion estimation and mapping. After the current growing season, we will apply the data we have gathered toward a SWD model. As mentioned, we are pursuing three modeling paradigms: a pure physics-based model, a machine learning model, and a physics-informed machine learning model. We are leaning toward the third option. However, we will evaluate all three approaches using the data we gathered during the 2025 growing season. The goal for the 2026 growing season is to deploy a much larger set of soil moisture sensors to validate the model and mapping approach. We likely will not have a full set of custom passive sensors by this time, so we will build out a much larger set of soil moisture sensors using a combination of off-the-shelf sensors with custom loggers and our previously developed powered sensors. This larger set of sensors will be deployed during the summer of 2026 in a field managed by USU extension and irrigated with either a wheel line or center pivot. As part of this effort, we will also evaluate how robust the model is to anomalous sensor readings and missing sensors. If our year 2 efforts are successful, we will be ready for a full system deployment in year 3.
Impacts
What was accomplished under these goals?
The overarching goal of this project is to develop an automated method to monitor soil conditions, starting with soil moisture, that doesn't require installation of fixed infrastructure or significant labor to maintain and operate. The three specific objectives of this project build toward this overarching goal. This year we have made significant progress under each of the three objectives. All available soil sensors are active devices that require a power source. This power source requires either installed infrastructure (e.g., electrical lines, solar panels) or must be periodically replaced (e.g., batteries). The first objective of this project is to develop passive soil moisture sensing coils that require no power and can stay buried in the soil for years at a time. The coils are energized by an above ground reader that periodically passes over them and collects the soil moisture reading in much the same way as smart credit card readers collect information from credit cards. The self-resonance frequency of the coil changes as a function of soil moisture. Thus, the above-ground transmitter/reader calculates the soil moisture by sensing the shift in resonance frequency of the passive buried sensing coil. PI Young's group has developed a spherical passive capacitive resonance soil moisture sensor. The spherical sensor configuration increases sensing area while enhancing sensing uniformity and sensitivity. The prototype sensor has been characterized in both laboratory and field environment with consistent sensing performance. In addition, a planer wireless passive soil moisture sensor was built and characterized in laboratory and field environment for a comparison study. The study shows that the spherical sensor achieved a higher sensitivity than the planar sensor. Specifically, the sensitivity was measured to be 1% of volumetric moisture change. Furthermore, reliable communication was maintained through 15 cm of soil. A further sensitivity improvement for the spherical sensor can be achieved by optimizing the sensing electrodes design. During this year we also demonstrated a multi-coil-based wireless stimulation and sensing architecture for interfacing passive soil moisture sensors. An increased communication distance has been demonstrated compared to the traditional two-coil system under the same stimulation power dissipation. The research results have been submitted to the IEEE Sensors Conference-2025 and IEEE PowerMEMS Conferene-2025 for publication. In order for the automated soil monitoring system that we are developing to operate autonomously, robotic drones need to be able to locate the buried soil sensors and wirelessly gather data. Therefore, our second major goal is to develop motion planning and localization methods for robotic drones. The robotic drones can approximately locate a buried sensor with GPS coordinates. However, in order to energize and read the moisture signal from the sensor, the above-ground reader needs to be precisely aligned above the sensor more accurately than is possible with only GPS. Therefore, PI Leang's lab focused on several tasks related to drone navigation and precise localization of sensors. First, the team focused on developing a dual extended Kalman (DEFK) filter to simultaneously estimate the states and unknown parameters of the inductively-coupled system, comprising the robot-carrying coil and buried passive sensor. The estimated parameters consist of the mutual inductance, which correlates to the spatial position of the buried sensor relative to the robot, and the self-capacitance of the buried sensor, which relates to the soil moisture level at the sensor's position. The DEKF algorithm has been shown to successfully estimate these parameters on both simulated and real data gathered in a laboratory setting. The estimated mutual inductance parameter will be used within a non-parametric Bayesian estimation process to localize the sensor. The Bayesian estimation process produces a probability model of the sensor's position, which will guide the robot's motion planner toward the most probable location of the sensor. Second, a mobile robot platform has been identified and will be used as the system to carry the sensing system and on-board computer for implementing the estimation and localization algorithms. The team expects to physically implement the algorithms in the spring of 2026 at the Utah State Univ. extension agriculture test side. The results of the DEFK algorithm will be presented at the upcoming 2025 Modeling, Estimation, and Control Conference (MECC) in October. The third component of the full system is a spatially dense soil water depletion map from the data gathered by the passive sensors that can be used to inform an irrigation system or irrigation practices. PI Roundy's group is pursuing three modeling approaches to build the soil water depletion map: a physics-based model, a machine learning model, and physics informed machine learning (i.e., symbolic regression). They have installed a series of off-the-shelf soil moisture sensors combined with a custom logging system to provide initial data from which to investigate the different modeling approaches. The sensors are installed in a field with drip irrigation. The water to the drip irrigation system is metered and catch cups are being used to measure precipitation events. At the end of the season, when irrigation is turned off, they will use the data to apply the different modeling approaches to determine the best approach moving forward. When all three components: the passive soil sensors, the robotic drones, and the soil water depletion model are working together, the result will be a low-cost fully automated system to manage irrigation more effectively. Furthermore, the platform could be used for other types of soil sensor for more complete soil management.
Publications
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2025
Citation:
J. M. Anderson, M. A. Mrotek, S. Ding, D. Young, S. Roundy, K. K. Leang, Simultaneous State and Parameter Estimation of Inductively-Coupled Buried Sensors for Soil Moisture Monitoring in Precision Agriculture, 2025 Modeling, Estimation, and Control Conference (MECC), October 5-8, Pittsburgh, Pennsylvania
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2025
Citation:
Ding, S., Roundy, S., Leang, K., Zesiger, C., Young, D.J. (2025) Passive Spherical Capacitive-Based Resonant Sensor for Soil Moisture Monitoring. IEEE Sensors, Vancouver, Canada, October 19-22, 2025.
- Type:
Conference Papers and Presentations
Status:
Under Review
Year Published:
2025
Citation:
Ding, S., Zhang, Q. Roundy, S., Leang, K., Zesiger, C., Young, D.J. (2025) Multi-Coil-Based Wireless Stimulation and Sensing for Zero-DC-Power Passive Spherical Soil Moisture Sensor and System Design. PowerMEMS 2025, Albequerque New Mexico, December 15-18, 2025.
|