This project aims to develop a precise (at the level of a single tree) yield-monitoring and mapping system for nut trees

The Department of Agriculture National Institute of Food and Agriculture awarded University of California, Davis $590,114 on February 15, 2026, under the Agriculture and Food Research Initiative (CFDA 10.310) to develop a precision yield-monitoring and mapping system for nut trees operable at commercial harvesting speeds.

The system uses a laser scanner to measure the volume of nuts passing beneath it during harvest, compatible with commercial almond and pistachio harvesting equipment and applicable to other nuts harvested with similar machinery. The core technical challenge addresses the mixing of nuts from neighboring trees as trees are shaken and nuts collected or conditioned into windrows. The award funds development of parametric models trained on single-tree harvesting experiments, with machine learning and optimization algorithms recovering individual tree yields and flows that best reproduce measured volume flow. The yield monitor will function as an add-on to conventional commercial harvesting equipment, supporting integration of large-scale yield maps with remote and proximal sensing data, historical production and environmental data, and enabling yield prediction and precision irrigation and nutrient management in orchards.

The award runs through February 14, 2029, with place of performance in Davis, California. The project grant is assistance type. UC Davis receives the award as the ultimate recipient and principal investigator organization.

The system will be compatible with commercial nut-harvesting equipment operating at commercial harvesting speeds. Although the system will be tested primarily in almond and pistachio orchards, IT can be applied to other nuts harvested with the same or similar equipment. The yield monitor will use a laser scanner to measure the volume of nuts passing beneath IT during on- or off-ground harvest. The main challenge arises from the fact that, as trees are shaken and nuts are collected from the ground or conditioned into neat piles (windrows), nuts from neighboring trees mix. Therefore, the stream of nuts processed by the machine and measured by the laser is the SUM of streams of neighboring trees. Recovering - or 'deconvolving' - the individual streams and yields is non-trivial. In our novel approach, parametric models will be trained based on yields and streams from single-tree harvesting experiments. Optimization and machine learning will use these models to recover the individual yields and flows that best reproduce the measured volume flow. The system will be an add-on to conventional commercial harvesting equipment. The proposed yield monitoring and mapping system will support the long-term goal of integrating large-scale yield maps with remote and proximal sensing data and historical production and environmental data to enable yield prediction and precision irrigation and nutrient management in orchards.