By Andi Anderson
Researchers from Purdue University and the University of Arizona have developed an innovative system called VR BioTalk that allows plant scientists to analyze large agricultural datasets using virtual reality and natural language commands. The new technology aims to make complex plant research more accessible to users without programming expertise.
Modern agricultural sensors generate enormous amounts of information every day. A single field scanner can produce terabytes of data daily, creating challenges for researchers who often lack the specialized coding knowledge needed to analyze such large datasets. VR BioTalk addresses this problem by allowing users to interact with data through simple voice commands in a virtual environment.
According to Jorge Vazquez, Ph.D. Student in Computer Science at Purdue University and the paper’s first author, “Our goal was to remove the programming knowledge requirement that stands between plant scientists and their own data.”
The immersive system enables users to wear a virtual reality headset, enter a three-dimensional reconstruction of a field, and ask questions about plant traits. The platform can identify plants, calculate measurements, generate statistics, and display charts within seconds.
Professor Bedrich Benes, Lead of the Project, explained, “The challenge we are trying to solve is the domain gap.” The technology helps bridge the gap between plant scientists and computer scientists by allowing researchers to work directly with data without creating complex computer programs.
VR BioTalk operates through three main stages. First, field data are processed to identify individual plants and measure characteristics such as height, leaf area, stem size, and plant structure. Second, speech-recognition technology converts spoken words into text, while an AI language model translates the request into an analytical task. Finally, the system executes the command and presents results through interactive visualizations.
Voicu Popescu, Co-Principal Investigator on the Project, noted, “The point-cloud 3D data acquired with scanners is particularly difficult to visualize.” The team developed specialized methods to improve visualization and interpretation of these datasets.
The technology was tested with participants from agronomy, environmental science, plant biology, and computer science backgrounds. Results showed that users found the system highly usable, engaging, and easy to understand, even without previous programming experience. The platform also demonstrated the ability to understand different accents.
Professor Alejandra Magana, who leads the educational effort, said, “VR-BioTalk has the potential to support research, but also education.” Researchers believe the system can help students focus more on scientific thinking and problem-solving rather than coding skills.
Looking ahead, the team plans to develop advanced digital twins that can simulate how plants respond to different environmental conditions. This future development could unlock the potential of vast amounts of agricultural data and support faster scientific discoveries.
Photo Credit: purdue-university
Categories: Indiana, General