For farmers, challenges such as pest infestations and irrigation inefficiencies are ongoing concerns. Traditionally, they relied on broad solutions like spraying entire fields with pesticides, irrigating on fixed schedules or applying uniform fertilizer rates, even when only certain areas required attention. These approaches, while practical in the absence of precise tools, often wasted resources. ICICLE is prototyping the use of agricultural drone imagery to allow for a more targeted approach.
Using drones equipped with high-resolution cameras and sensors, the team captures thousands of images during a single flight. These images are then analyzed by AI to create infrared maps, which visually represent areas of concern by identifying temperature differences in crops that indicate stress or disease. These heat maps could help farmers make precise, informed decisions to address specific problems in their fields.
The ICICLE team’s research combines precise but time-consuming on-the-ground measurements with more scalable, cost-effective data sources like drone imagery and weather patterns to provide a more comprehensive understanding of farming conditions.
“This approach saves time and resources by identifying specific issues rather than applying generalized solutions,” Subramoni said.
While agricultural drone imagery cannot solve every agricultural challenge, its AI-driven analysis could help address key issues like determining soil moisture levels, detecting crop disease, measuring harvest yield and classifying growth stages. Through infrared mapping and phenotyping, drones could create a holistic view of a farm’s health.
Processing the high number of images per flight requires significant memory and computational power, which is where OSC plays a critical role.
“OSC resources are vital due to the required memory being more than what would be possible on a personal computer. OSC resources allow for multiple flights worth of data to be processed at the same time throughout the center,” said Matthew Lieber, a software engineer on the project.
Through OSC’s computing power, the project can process large datasets quickly, which would allow farmers to get timely insights that would otherwise take far longer to compute. In addition, OSC staff helped set up an app to assist the researchers with data capture. This faster data analysis could create a more responsive farming operation that can adapt to challenges in real time.
ICICLE’s work extends beyond agriculture. The team is developing workflows that make complex technical processes actionable across diverse fields, ensuring AI technologies are adaptable to a broad range of needs.
“This isn’t just for individual users it’s about fostering partnerships,” Lieber said. “Crop consultants, for example, could use these tools to provide more precise guidance, and similar workflows could be adapted for applications in environmental monitoring or urban planning.”
By integrating AI into practical applications, ICICLE is also training a new generation of professionals to navigate a data-driven future.
“This project is about empowering people,” Subramoni said. “Through efforts like crop growth stage analysis, ICICLE is innovating while ensuring its benefits are shared broadly.”
OSC serves as a core collaborator on the larger ICICLE initiative, now in its fourth year, offering computing power, data storage and research software engineering expertise. As part of the ICICLE leadership team, Karen Tomko, director of research software applications, leads OSC’s efforts on the NSF grant, and various OSC staff members have provided assistance to research projects.
Source : osc.edu