A digital simulation and analysis platform, informed by real-world data collected from the smart farm, would allow researchers to more quickly run experiments and gather data. Because many experiments can be run in parallel within a simulation, researchers can pursue ideas that may not be feasible to test on a real farm due to cost or scale.
ESFSim is a campus-wide cloud-model-based simulation testbed that will allow broad and auditable sharing of data and compute resources among any campus researchers as long as they need access and propose pilot research projects on ESFSim. ESFSim’s capabilities will grow with researcher needs. Currently, we are focused on building the foundation for ESFSim such that it can be expanded upon in the future.
Of course, bespoke digital simulations have drawbacks of their own. They are initially costly and difficult to build. By creating a common foundation, more researchers will be able to make use of compute resources by lowering the effort required to deploy simulator-based experiments. They can instead focus their efforts on customizing the platform to suit their specific needs and performing directly productive research activities. Such efforts can be reused by other researchers to further lessen the effort required to use the system.
ESFSim focuses on creating a small-scale ESF simulation infrastructure to simulate collecting, transmitting, preserving, and processing data, in support of measuring, evaluating, and acting on short- and long-term processes and outcomes pertinent to sustainable farming activities, environmental monitoring, and workforce training. In the proposed infrastructure, we will simulate our high-risk/high-reward research ideas and build a smaller exemplar farm CPS system.
Users will be able to perform training tasks within the ESFSim platform. Such tasks can make use of available datasets or data from simulation results.
To allow researchers and stakeholders to use ESFSim, a web-based interface will be constructed. This interface will allow users to interact with data, perform training AI training tasks, and run simulations.
AI Training Image: Copyright Fuzheado via Wikimedia | User Interface Image: Copyright Linux Screenshots via Wikimedia