University of Tokyo and Kubota Develop AI-Powered Drone System to Predict Potato Yields Before Harvest

Researchers from the University of Tokyo Graduate School of Agricultural and Life Sciences and Kubota Corporation have developed a new phenotyping method that uses drones, artificial intelligence (AI), and crop growth modeling to predict potato yields before harvest. The technology enables researchers and growers to estimate underground tuber biomass without digging up plants, offering a non-destructive approach to yield forecasting and precision agriculture. 

The research combines drone-based remote sensing, machine learning, and a time-series growth model to estimate potato tuber development throughout the growing season. The project was carried out under the joint Kubota Todai Lab initiative and demonstrates the potential of AI-driven field phenotyping for crops with underground harvest organs.

Pipeline for Predicting Tuber Yields

Drone Imagery Combined with Machine Learning

As part of the study, drones equipped with RGB and multispectral cameras regularly photographed potato fields during the growing season. Researchers extracted several crop growth indicators from the images at the plot level, including:

• Plant cover ratio 
• Canopy height 
• Color indices 
• Vegetation indices

These image features were combined with underground tuber biomass measurements collected through field sampling to train a machine-learning model. Once trained, the model estimated tuber biomass in unharvested plots using only drone-derived image data. 

According to the researchers, this approach provides a practical alternative to traditional destructive sampling methods that require digging up plants to estimate yield.

observation of potato fields using drone images Using RGB and multispectral images from drones, growth indicators such as vegetation rate, community height, and vegetation index were obtained at the field scale.

Growth Curve Model Enables Yield Forecasting

To predict final harvest yields, the estimated underground biomass data were integrated into a Gompertz growth curve, an S-shaped mathematical model commonly used to describe biological growth over time. 

By applying the time-series estimates generated through machine learning to the growth model, researchers were able to forecast potato yields before harvest while accounting for crop development throughout the season. 

The study demonstrates that combining remote sensing with growth modeling can improve the accuracy of pre-harvest yield estimation and support more informed crop management decisions.

Two-Year Field Trials Delivered Promising Results

The research was conducted during the 2023 and 2024 growing seasons at the University of Tokyo Field Science Center in Nishi-Tokyo City. Multiple treatment plots with different planting densities and seed potato conditions were evaluated to test the system under varying cultivation practices.

According to the research team:

• Tuber biomass estimation achieved a correlation coefficient of more than 0.8. 
• Final yield prediction using the growth curve achieved a correlation coefficient exceeding 0.7.

These results confirmed that underground potato yield can be accurately estimated before harvest using above-ground drone observations combined with AI-based analysis.

Verification of yield prediction accuracy using growth curves Comparison results between tuber weight estimated by growth curves and true values measured at harvest.

Supporting Smart Agriculture

Potatoes are one of the world's most important food crops, but because tubers develop underground, monitoring yield during the growing season has traditionally relied on destructive sampling. The newly developed method offers a non-destructive alternative that captures spatial variation across entire fields while preserving the crop.

The researchers believe the technology could support a range of precision agriculture applications, including:

• Pre-harvest yield forecasting 
• Optimization of cultivation management 
• Improved field monitoring 
• Harvest timing recommendations 
• AI-based crop phenotyping

The team also noted that the approach has the potential to be applied to other crops with underground harvestable organs, expanding the use of drone-based remote sensing and AI in smart agriculture.

Research Team

The study was led by doctoral student Yuto Imachi, Professor Hiroyoshi Iwata, and Associate Professor Wei Guo from the University of Tokyo, together with researchers from Kubota Corporation's Next-Generation Research Department, Masahiro Okada of Sarabetsu Prediction Co., Ltd., and Pieter M. Blok, formerly a project assistant professor at the University of Tokyo and now affiliated with Eindhoven University of Technology. 

The researchers expect the technology to contribute to more accurate pre-harvest yield forecasting, improved cultivation management, and the continued advancement of AI-powered precision agriculture for potato production.