Backed by UzCombinator

Crop monitoring and spraying automated

Robot built to do both on its own: precise doses where they are needed, and a camera that watches for disease.

Request DemoHow it works
3D model of the CropScout V3 robot: a tracked base carrying a spray tank, with a tall twin mast of side-spray nozzles and a camera at the front.

Backed by

UzCombinator

Batch 2, 2026

Presented at

ASU Fulton Innovation Showcase

Spring 2026

Awarded

USAID agri-innovation contest

2nd place

Built at

HackHarvard

1st place

The robot

Two jobs at launch, one base

A tracked base sized for greenhouse rows and wet soil. Every tool mounts to the same top plate, so one robot grows into the farm's platform step by step.

Simulation (digital twin): the robot drives a greenhouse row, the camera checks each plant, and only the plant with leaf spot is sprayed.
/01Launch · parts ordered

Precision sprayer

Doses pesticide and fertiliser down greenhouse rows and small plots where tractors can't fit. Nozzles switch one by one, so a plant flagged as diseased is sprayed and its healthy neighbour is not.

/02Launch · retraining models

Crop monitor

A camera and sensors track plant health, disease signs and growth, and report to the farmer's phone. The vision models are being retrained on real field photos, tomato and cucumber first.

/03Next

Weeder

Automated weeding on the same base: the job farmers named most often as their biggest time cost.

See the full engineering log

Field research

Built on what 30 farmers told us

We interviewed farmers in Jizzakh, Surkhandarya and Shakhrisabz in July and August 2026, before settling the design. The results are published in full, including the answers that argue against us.

30

Farmers interviewed

Jizzakh, Surkhandarya, Shakhrisabz

14/ 30

Weeding is their biggest time cost

12/ 30

Can't find seasonal labour

18/ 30

Picked the universal robot as helpful

“When we spray by hand, more than the norm ends up on the crop. […] If every plant got the same exact dose, no chemical would be wasted, and it wouldn't be a risk to the people who eat it.”
Greenhouse farmer · cucumbers · 0.8 ha
“If we take electric robots out to the fields and there's no power out there, how would I charge them?”
Orchard and vegetable farmer · 30 sotix
“It would be good if spare parts were easy to find, service were close by, and the price were affordable for farmers.”
Vegetable farmer · 2 ha and a greenhouse

Quotes shared with consent, without names, translated from Uzbek.

Roadmap

Build, then pilot

  1. 2026

    Done

    MVP

    A four-wheel rover with seven custom PCBs, demoed at the ASU Fulton Innovation Showcase. Then 30 farmer interviews, which told us to build bigger.

  2. Fall 2026

    Done

    CAD model and Vision v1

    The first browser version of the crop-vision tool, built at HackHarvard.The tracked V3 base modelled and frozen in SolidWorks.

  3. Winter 2026

    In progress

    Fabricate and assemble

    Undercarriage, sprayer and onboard compute, built at ASU Polytechnic.

  4. 2027

    Next

    Pilot

    On real plots. The first number we owe farmers is the one they asked for: how many sotix an hour it covers.

The CropScout team standing behind the four-wheel MVP rover at the ASU Fulton Innovation Showcase, with the project poster beside them.
The MVP rover at the ASU Fulton Innovation Showcase, spring 2026.Watch the MVP demo

See CropScout in the field

We're building the V2 at ASU Polytechnic and start pilots in November 2026. Farmers, partners and investors: get in touch for a demo.

Request DemoTry Vision now