Problem / 01
Labor-intensive maintenance
One early orchard conversation surfaced a reported $670,000 annual manual weed-removal burden—large enough to justify testing automation immediately.
Autonomous farm robotics · 2023–2024
ZAPTrack
Building an orchard robot from the farm outward—one cheap, testable machine at a time.
20farms interviewed across multiple operating scales
1successful paid pilot
2 weeksfrom rough parts to two fieldable iterations
$50K/moletter-of-intent value captured during validation
01 / The brief
Organic farming had a scaling problem, not an ideas problem.
For organic growers, a single non-organic herbicide application can jeopardize certification, while approved alternatives can be dramatically more expensive. Manual removal works, but it does not scale cleanly with acreage. ZAPTrack began as a search for the smallest machine that could prove a better workflow in the field.
Constraint / 02
Protect the crop
Cultivators and finger weeders can be slow, operator-dependent, and risky around valuable plants. Precision mattered as much as raw coverage.
Direction / 03
Reuse what farms already understand
The project moved toward familiar mowing hardware plus an autonomy layer—reducing new mechanical invention and concentrating effort on perception, control, and deployment.
02 / Customer discovery
Talk to the person who owns the problem. Then bring hardware.
The archive documents conversations with conventional, drone-enabled, and organic farms in Cincinnati and California. Each pushed the product in a different direction: crop analytics, equipment augmentation, and autonomous physical work.
01
Measure crop health
A 1,000+ acre operation highlighted fertilizer spend and the value of repeated stand-count data for smarter allocation.
02
Upgrade existing tools
A grower already using a drone wanted better intelligence from equipment she owned—not another disconnected machine.
03
Scale organic acreage
An organic farmer working one acre of a 100-acre property needed weed control that could grow without eroding the farming method.
03 / Build philosophy
Every prototype had one job: collapse uncertainty.
The process borrowed from the same rapid-prototyping discipline used across my hardware work. Keep fidelity only where the question demands it; spend money on learning; let the ugly prototype expose the next constraint.
01
IdeaTurn a field observation into one falsifiable proposition.
02
CADModel only the geometry needed to answer the next question.
03
BuildPrint, wire, adapt, or salvage the fastest physical route.
04
TestPut the machine into dust, grass, heat, and real operator workflows.
05
IterateLet evidence—not attachment—choose what survives.
04 / Engineering build log
From sensing crops to moving through them.
ZAPTrack was not a single robot reveal. It was a sequence of increasingly specific machines, each built to retire a different product or engineering risk.
Phase 00
Mar–Aug2023
Perception before actuation
Build. iSight paired a custom drone with a Raspberry Pi 4 and Pixhawk flight controller for invasive-species detection and stand-count reporting. CropEagle explored a lower-cost, high-resolution camera attachment. CropMantis translated the research into an autonomous ground-robot direction.
Why it mattered. These concepts separated three jobs—observe, understand, act—and clarified that the most acute organic-farm need was not another report. It was removing weeds without chemicals.
Flight hardware → detection output → camera concept → ride-on sensor rigRaspberry Pi 4 · Pixhawk · high-resolution imaging
Phase 01
Apr2024
A two-axis toolhead for precise weed targeting
Build. A linear positioning mechanism drew from CNC machines and 3D printers: extrusion rails provided structure, linear rods constrained movement, a lead screw converted motor rotation into controlled travel, and a printed carriage localized the working tool.
Challenge. Targeted removal demands repeatable positioning without prematurely engineering a full vehicle. Isolating the toolhead let the team learn about motion and packaging before coupling it to autonomy.
Zapping mechanism prototypeExtrusion · lead screw · linear guides · printed carriage
Phase 02
Apr–May2024
The $50 mower hack
Build. A used electric lawn mower was electrically and mechanically combined with parts from an RC ride-on car. The donor systems supplied a cutting deck, traction, wheels, power, and steering behavior quickly enough to test the premise on a farm.
Challenge overcome. The machine moved and cut, but its geometry was wrong: the deck sat too high. That failure was useful. For roughly $50, it converted an abstract clearance concern into a hard packaging requirement.
Prototype 01 — donor mower + ride-on drivetrainCost ≈ $50
Phase 03
Jun2024
A lower, tougher platform for the orchard
Build. The second prototype leaned on a production mower chassis and was assembled with JingXiang Mo in days. A compact sensor mount, low body, and four driven wheels created a more credible machine for a large-farm trial.
Challenge overcome. Moving from improvised mobility to a purpose-built mower frame improved ground clearance, durability, and the ability to demonstrate on uneven orchard terrain—without designing every mechanical subsystem from zero.
Prototype 02 — pilot-intent orchard platformCost ≈ $300 · built in days
Phase 04
Jul2024
From custom robot to deployable autonomy layer
Build. ZAPTrack sourced a mowing base from a manufacturing partner, then focused its own work on making that platform autonomous. The vehicle interface could be driven by manipulating three signals, shrinking the integration surface.
Milestone. The team completed a $100 paid pilot on one acre, with broader testing across two acres of California almond and stone-fruit orchards, and reframed the company around software and integration that could be assembled onto incoming hardware in about one day.
Product directionStandardized base · three-signal control · one-day integration target
05 / System engineering
The robot became a stack, not a sculpture.
Each layer could evolve on its own: a commercially sourced base for dependable motion, a small electrical control interface, onboard perception, and task planning for orchard work.
01 / Mobility
Production mower base
Use a partner-built chassis to inherit traction, deck geometry, and serviceable hardware instead of re-solving the entire vehicle.
02 / Interface
Three control signals
Minimize the autonomy-to-vehicle boundary so an incoming platform can be instrumented and tested quickly.
03 / Perception
Camera + edge compute
Run live orchard imagery through an onboard computer to identify trunks, rows, and other field objects near the machine.
04 / Action
Mow, survey, target
Share one mobile platform across mowing, disease observation, harvest data, and precise weed-removal experiments.
06 / Challenges resolved
Six constraints changed the machine.
01
Which problem was worth automating?
Interviews spanned fertilizer optimization, aerial scouting, invasive-species detection, and weeding. The product narrowed toward weed removal only after growers exposed the severity and frequency of the task.
02
How do you test without hardware capital?
Use discarded and commodity systems as engineering modules. A mower, a ride-on toy drivetrain, simple wiring, and fast fabrication answered the mobility question for tens—not thousands—of dollars.
03
Would the geometry work between trees?
The first mower was visibly too tall. Instead of hiding the miss, the next iteration treated height, width, sensor placement, and under-canopy travel as primary architecture inputs.
04
How do you survive field reality?
The build moved quickly from a workshop assembly to orchard trials. Uneven ground, vegetation, dust, lighting, and transport exposed issues a benchtop test could not.
05
What should be custom?
The manufacturing-partner platform shifted effort away from commodity mobility and toward the differentiated layer: control integration, perception, planning, and farm-specific workflows.
06
Could the business scale with the machine?
The technical demonstration earned demand, including a roughly $50K-per-month LOI. It also exposed a strategic mismatch: a premium product without an established premium segment or channel was difficult for an early-stage company to carry.
07 / Outcome + reflection
The prototype proved the work. The market refined the company.
ZAPTrack reached the thing early hardware rarely reaches: a paying customer in the actual operating environment. The larger lesson was that technical feasibility, customer pain, and a scalable distribution path must mature together.
Founder reflection
Start cheap. Learn in public. Make the next machine earn its complexity.
The premium-first approach created ambition, but not the channel required to support it. Reusing existing mower platforms pointed to the more durable idea: meet buyers through equipment they already know, then deliver value through autonomy.
Jan 2023
Startup WeekendMar 2023
Cincinnati Business Achievement AwardAug 2023
Next Innovation ScholarshipJun 2024
Founders Village Tech Weekend pitch winner