Core Faculty and Researchers
Kaiyu Guan, Shenlong Wang, Evan DeLucia, Lisa Ainsworth, Jonathan Coppess, Bin Peng, Sheng Wang
Representative Projects
3D reconstruction of plants and farmlands
Team members: Shenlong Wang, Kaiyu Guan, Sheng Wang. The ability to automatically build 3D digital replicas of plants and farmlands from images has countless applications in agriculture, environmental science, robotics, and other fields. However, 3D reconstruction of plants remains to be a challenging problem for computer vision due to heavy occlusion and the complex geometry of plant structures. The goal of this project is to leverage scientifically grounded knowledge on plant structure to build 3D reconstructions that are complete, accurate, and realistic. The development of such methods will enable automatic, large-scale monitoring of plants for agricultural and scientific purposes. The collected data can offer decision support for farmers while also aiding the design of new agricultural techniques, increasing crop productivity and alleviating the rising food crisis of today’s world. Plant shape reconstruction methods will also pave the way forward for intelligent manipulation of plants by autonomous robots, preparing them for tasks such as crop harvesting and pruning. Eventually, our goal is go even beyond shape and build models of mechanical and optical properties, bringing us closer and closer to truly complete digital twins of plants.
Modeling the agroecosystem sustainability of the future US Midwest (funded by NSF)
Team Members: Kaiyu Guan, Bin Peng, Sheng Wang
This project aims to answer the following grand questions: could the U.S. Midwest remain as the global food basket in the next 100 years? How can we ensure co-sustainability of food production and environmental quality in this landscape? Carbon (e.g. crop growth), hydrology (both water quantity and quality), and nutrient cycles in the U.S. Midwest are closely intertwined from the field/headwater scale to the whole river network and continental scales. Human activities and practices do not singly affect one component, but the whole interconnections. Their complex feedbacks and non-linear interactions require a “system” view when assessing the U.S. Midwest landscape and potential adaptations. This project adopts a “system” view to holistically model and quantify the coupled “food-water-nutrient nexus” for the U.S. Midwest agroecosystems, aiming to significantly advance the process-based understanding and predictability of this agroecosystem, with two management practices (i.e. controlled drainage, and nutrient management) and under future climate conditions.