Radio imaging technology for plant measurement in specialty crops

Overview

This research proposes a new method for measuring plant structure using characteristic signals obtained from RF signals passing through vegetation. The goal is to improve yield forecasting in specialty crops, where existing technology-assisted methods have performed no better than manual approaches. By leveraging the dielectric properties of biomass, the approach aims to enable direct measurement of plant mass and volume rather than relying on indirect or assumed statistical relationships such as leaf-area index, which may not hold in diverse production environments. Instruments mounted on farm machinery could offer a dramatic improvement over current methods, benefiting specialty crop growers who depend on accurate yield estimates.

Technical specifications

The core approach uses RF signals transmitted through vegetation, with the resulting signal characteristics analyzed to infer plant structure and biomass. Key technical elements include:

• RF propagation through biomass: The dielectric properties of plant material allow RF signals to carry information about mass and volume when properly processed.
• Signal processing and computational imaging: Contemporary techniques are applied to extract meaningful measurements from complex RF signals, building on foundational work that demonstrated RF-based biomass estimation is feasible.
• Commercial off-the-shelf components: Initial validation confirmed that biomass presence between transmitter and receiver antennas is measurable using standard network analyzers, with signal complexity increasing alongside physical complexity of the biomass.
• On-farm integration potential: The system is designed for mounting on farm machinery, enabling in-field measurement during normal operations.

Technology readiness level

The technology is at an early feasibility stage. Initial measurements obtained in October 2019 confirmed that RF behavior with respect to plant biomass is measurable using commercial off-the-shelf components. Planned next steps include specifying the radio system, modeling radio propagation, building a laboratory prototype, and conducting proof-of-concept testing with fruit and vegetation. Signal processing and machine learning feasibility will be evaluated by collecting training and validation data from laboratory observations and training vegetation and fruit mass estimation models. Agronomic field validation is recognized as necessary but falls outside the scope of the current project.