System and method for turning irrigation pivots into a network of robots for optimizing fertilization

Eyal Neistein, Herzlyia (IL); Ran Yifa, Kfar Hahores (IL); Yair Sharf, Aloney Aba (IL); Yaroslav Don, Ramat Yohanan (IL); Yossi Haran, Modi'in-Macccabim-Re'ut (IL); Yuval Aviel, Eshtao (IL)

USPTO: us-12622347 · published 2026-05-12 · patents · A01C 21/007· A01B 69/008· A01B 79/005· A01B 79/02· G05D 1/0212· G05D 1/0246· G06V 20/188· G06T 2207/10036

System and method for turning irrigation pivots into a network of robots for optimizing fertilization

Eyal Neistein, Herzlyia (IL) , Ran Yifa, Kfar Hahores (IL) , Yaroslav Don, Ramat Yohanan (IL) , Yair Sharf, Aloney Aba (IL) , Yuval Aviel, Eshtao (IL) , Yossi Haran, Modi'in-Macccabim-Re'ut (IL) This is my paper

Pith reviewed 2026-05-15 23:02 UTC · model grok-4.3

classification patents A01C 21/007A01B 69/008A01B 79/005A01B 79/02G05D 1/0212G05D 1/0246G06V 20/188G06T 2207/10036

keywords variable rate fertilizationNDVI leaf segmentationprecision agricultureirrigation pivot roboticsmulti-spectral imagingground-penetrating radarsegment-based dosing

0 comments

The pith

Fertilizer doses per field segment are adjusted by re-computing an eight-constant NDVI from leaf-location and intra-leaf image data after each application pass.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The method divides a field into segments and uses a mobile platform carrying ground-penetrating radar to apply an initial fertilizer amount to each segment. After a growth interval, multi-spectral images are taken and a reference NDVI value is computed for every segment by first locating leaves and then sampling specific regions inside those leaves. The same imaging and NDVI step is repeated after the interval, and the new values directly determine how much additional fertilizer each segment receives. A sympathetic reader would care because the approach claims to replace uniform field-wide rates with data-driven, segment-level corrections that respond to actual plant reflectance changes.

Core claim

The central claim is that fertilizer application can be optimized by treating irrigation pivots as a network of mobile platforms that apply an initial dose, capture multi-spectral imagery, calculate an NDVI that explicitly incorporates leaf positions and intra-leaf zones, and then re-apply a revised dose based on updated NDVI values computed with the same eight-constant formula.

What carries the argument

An NDVI formula that first identifies leaf locations and intra-leaf regions in infrared, red and green images, then plugs the resulting reflectance values A, B and C into an eight-constant expression whose output drives the variable-rate decision for each segment.

If this is right

Each re-imaging pass produces a new NDVI map that can be used to increase or decrease the next application without manual recalibration.
The same mobile platform can serve as both applicator and sensor, turning a conventional irrigation pivot into a closed-loop robotic network.
Fertilizer savings scale with the number of distinct segments whose NDVI trajectories diverge from the field average.
The method can be repeated multiple times within a single growing season as new imagery becomes available.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Because the platform already carries radar, the same passes could simultaneously map soil moisture or compaction and fold those data into the fertilizer decision.
If the eight constants prove stable across crop types, the algorithm could be embedded directly in existing pivot controllers without per-field tuning.
The leaf-location step opens the possibility of detecting early disease or stress signatures that standard whole-canopy NDVI would miss.

Load-bearing premise

The eight-constant NDVI expression derived from the three spectral bands accurately indicates how much additional fertilizer each plant segment needs, without any ground-truth calibration measurements.

What would settle it

A side-by-side field trial in which segments fertilized according to the leaf-aware NDVI show no statistically significant difference in final yield or tissue nutrient content compared with segments fertilized at a uniform rate.

read the original abstract

1 . A method of varying fertilization levels within a field, the field being divided into a plurality of segments, the method comprising: applying, by a mobile platform with a ground-penetrating radar sensor, at a first time an amount of fertilizer to each segment; capturing images of each segment in at least two different spectral ranges after a time period; calculating a reference Normalized Difference Vegetation Index (NDVI) value for each segment based on the captured images, wherein the calculation of the NDVI value incorporates a determination of one or more of: (i) specific locations of leaves within each plant, and (ii) specific regions within each leaf; and varying the amount of fertilizer applied to each segment after the time period, wherein the amount of fertilizer is varied based on re-captured images and re-calculated NDVI values for each segment after the time period, wherein the NDVI value is calculated according to the formula: ; ; where A, B and C are spectral reflectance measurements acquired from the images captured in the infrared, red and green spectral ranges, and where α, β, γ, δ, φ, χ, ψ and ξ are constants.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Patent filing on center-pivot variable-rate fertilization via leaf-aware NDVI; the system sketch is concrete but the eight-constant formula has no derivation or validation.

read the letter

The core offering is a practical integration: mount multi-spectral cameras and ground-penetrating radar on an irrigation pivot, segment leaves and intra-leaf zones from the images, feed a modified NDVI into a feedback loop that adjusts fertilizer per segment on the next pass. That combination of hardware placement and per-leaf masking is the only piece that goes beyond textbook variable-rate practice. Everything else—standard NDVI logic, pivot kinematics, and segment-wise application—is already in commercial use. The formula itself is the weak point. It introduces eight free constants (α through ξ) whose values are never derived, fitted, or tested against tissue samples or yield. Without that mapping, the claim that the index tracks fertilization need remains an assertion. The patent supplies no error bars, no calibration data, and no comparison to ordinary NDVI, so the performance gain cannot be assessed. Because this is a patent rather than a research paper, the absence of results is expected, yet it also means the document offers no reproducible result a reader can build on. An engineer looking for prior art on pivot-mounted sensing might cite the claims; a researcher needing validated indices would not. I would not bring it to a reading group and would not cite it in my own work. A serious editor should desk-reject rather than send it to peer review; the technical gap is too large for referee time to close.

Referee Report

2 major / 0 minor

Summary. The manuscript describes a method for precision fertilization in which irrigation pivots are reconfigured as a network of mobile platforms equipped with ground-penetrating radar and multi-spectral cameras. After an initial uniform fertilizer application, images in infrared, red and green bands are captured; a custom NDVI is computed that purportedly incorporates leaf-location and intra-leaf-region masks; the resulting per-segment NDVI values then determine adjusted fertilizer rates in subsequent passes. The sole concrete formula supplied is NDVI = f(A,B,C,α,β,γ,δ,φ,χ,ψ,ξ) where A/B/C are the three reflectance channels and the eight Greek letters are unspecified constants.

Significance. If the claimed mapping from the eight-parameter NDVI to actual plant nutrient demand were empirically validated, the approach could enable finer-scale variable-rate fertilization than current NDVI-based commercial systems. No such validation, calibration data, or yield-response curves are provided, so the practical significance cannot yet be assessed.

major comments (2)

[Abstract] Abstract (formula block): the NDVI expression is stated to depend on eight free constants α–ξ whose values are neither derived from first principles nor obtained by any fitting procedure against tissue-nutrient or yield data. Without this mapping the feedback loop that varies fertilizer rates has no demonstrated grounding in plant response.
[Abstract] Abstract: the text asserts that the NDVI calculation 'incorporates a determination of one or more of (i) specific locations of leaves … and (ii) specific regions within each leaf,' yet the displayed formula contains only the three reflectance channels A/B/C and the eight constants; no term or preprocessing step shows how the leaf masks enter the index.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the detailed reading. The manuscript is a patent application whose primary contribution is the system architecture that turns irrigation pivots into a mobile sensing-and-actuation network. Below we address the two concrete points raised about the NDVI formulation. We are prepared to add clarifying language in a revised version but note that a patent document is not required to contain the full empirical calibration data set.

read point-by-point responses

Referee: [Abstract] Abstract (formula block): the NDVI expression is stated to depend on eight free constants α–ξ whose values are neither derived from first principles nor obtained by any fitting procedure against tissue-nutrient or yield data. Without this mapping the feedback loop that varies fertilizer rates has no demonstrated grounding in plant response.

Authors: The eight constants parameterize a family of NDVI-like indices that can be tuned to local crop, soil, and sensor conditions. In the intended commercial deployment the constants are obtained by a one-time field calibration against tissue samples or yield-monitor data; the patent text deliberately leaves the precise fitting procedure outside the claims so that it can be performed by the end user or service provider. We agree that the current wording does not make this calibration step explicit and will add a sentence stating that the constants are determined empirically for each deployment. revision: partial
Referee: [Abstract] Abstract: the text asserts that the NDVI calculation 'incorporates a determination of one or more of (i) specific locations of leaves … and (ii) specific regions within each leaf,' yet the displayed formula contains only the three reflectance channels A/B/C and the eight constants; no term or preprocessing step shows how the leaf masks enter the index.

Authors: The leaf-location and intra-leaf masks are generated by a preceding image-segmentation stage (described in the detailed description) that isolates vegetated pixels before the reflectance values A, B, and C are extracted. The function f therefore operates only on the masked pixels; the masks themselves do not appear inside the algebraic expression. We will insert a short clause in the abstract and in the formula caption to make the preprocessing order explicit. revision: yes

standing simulated objections not resolved

Empirical validation or yield-response curves are not supplied; as a patent filing the manuscript prioritizes enablement of the claimed system over experimental results.

Circularity Check

1 steps flagged

NDVI formula with eight unspecified constants reduces fertilization variation to parameter choice by construction

specific steps

fitted input called prediction [Abstract, claim 1]
"the NDVI value is calculated according to the formula: ; ; where A, B and C are spectral reflectance measurements acquired from the images captured in the infrared, red and green spectral ranges, and where α, β, γ, δ, φ, χ, ψ and ξ are constants."

The only concrete expression supplied for the index that drives segment-specific fertilization contains eight free constants whose values are never derived or validated against yield or nutrient measurements; varying fertilizer on the basis of this index is therefore equivalent to re-using the chosen constants.

full rationale

The paper's central loop applies fertilizer, images the crop, recomputes NDVI via the supplied multi-constant expression, and adjusts the next application. Because the eight constants α–ξ are introduced without derivation, calibration data, or mapping to tissue nutrient levels, the re-calculation step is definitionally equivalent to selecting those parameters; the claimed optimization therefore collapses to the input parameterization rather than an independent prediction.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Central claim rests on the unproven assumption that leaf-aware NDVI directly indicates optimal fertilizer dose and on eight unspecified constants in the reflectance formula.

free parameters (1)

alpha to xi
Eight constants in the NDVI formula are introduced without derivation or reported values.

axioms (1)

domain assumption NDVI calculated from leaf-specific regions accurately reflects fertilization requirement
Invoked throughout the method description without supporting evidence.

pith-pipeline@v0.9.0 · 6360 in / 1049 out tokens · 60224 ms · 2026-05-15T23:02:55.223475+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith.Cost.FunctionalEquation washburn_uniqueness_aczel contradicts

?

contradicts
CONTRADICTS: the theorem conflicts with this paper passage, or marks a claim that would need revision before publication.

the NDVI value is calculated according to the formula: ; ; where A, B and C are spectral reflectance measurements acquired from the images captured in the infrared, red and green spectral ranges, and where α, β, γ, δ, φ, χ, ψ and ξ are constants.
IndisputableMonolith.Foundation.DAlembert.Inevitability bilinear_family_forced contradicts

?

contradicts
CONTRADICTS: the theorem conflicts with this paper passage, or marks a claim that would need revision before publication.

varying the amount of fertilizer applied to each segment after the time period, wherein the amount of fertilizer is varied based on re-captured images and re-calculated NDVI values for each segment after the time period

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.