ForkGuard Log
Impact, near-miss and pre-shift-check evidence for small forklift fleets
Logs every impact, every reversing near miss with a person and the daily pre-shift check on each forklift, with an advisory cab chime as a driver aid.
- Who buys it
- Operations or EHS managers at warehouses, 3PLs, builders' merchants and plants with 5-40 mixed-brand forklifts and no OEM telematics.
- Why it pays
- A 20-truck site loses about $8-12k a year to rack and door damage found late and to unprovable paper checks. ForkGuard costs about $10.1k in year 1 and $2.2k a year after (est.).
- Target price
- $399 per truck + $9/truck/month (target)
- First-minute wow
- A floor-plan heat map of where reversing near misses happen, and a text to the supervisor a minute after a rack is hit.

What the research found
And how this design answers it
- Life-safety liability -> sold as an evidence recorder; the chime is a labelled driver aid and there is no truck-control output
- AI warning kits already sell for about $950 -> under half the price, competing on fleet data: impacts, hotspots, checks and hours
- Enterprise near-miss and telematics platforms are quote-only -> published price, any brand, any voltage, month-to-month
- Hailo per truck at about 6 W -> on-sensor IMX500 AI camera with a Pi Zero 2 W at about 3 W
- Truck power and vibration -> railway-rated Mean Well RSD-30 converters for 12-80 V, TPU isolators, M12 connector, read-only filesystem
- Cameras on workers -> no images stored or sent by default; operator names only on the check they sign
Use case
Rack upright hit during the afternoon shift
Forklift operatorShift supervisorRack inspector
- 01Trigger: truck FL-07 reverses into a rack upright at 14:20
- 02Detect: the accelerometer logs a 4.1 g, 40 ms impact in the Aisle 12-14 zone
- 03Alert: the supervisor gets a text within a minute with truck, zone and severity
- 04Action: the supervisor tags the bay out and calls the rack inspector the same shift
Outcome: Damage found in minutes instead of at the annual inspection; one upright replaced before it fails, about $500-1,000 (est.)
How it works
Architecture
- 01
Sense: Raspberry Pi AI Camera (IMX500) detects people on the sensor; LIS3DH logs impacts; isolated input reads the reverse signal
- 02
Decide: Pi Zero 2 W tracks people, sets distance bands and raises near-miss and impact events
- 03
Send: events go to SQLite first, then MQTT/TLS over site Wi-Fi, with 30 days of store-and-forward
- 04
Tell: advisory cab chime and LED bar; supervisor texts; weekly hotspot and check-compliance report
Hardware
Real parts, buyable today
Bill of materials: about $250 per truck (est.).
| Block | Part | Why this part |
|---|---|---|
| Detector | Raspberry Pi AI Camera (Sony IMX500) | Person detection on the sensor for $70 at about 1 W |
| Host | Raspberry Pi Zero 2 W | Tracking, IMU, Wi-Fi and BLE zone hints for $15 |
| Impact | LIS3DH +/-16 g accelerometer | 400 Hz with an interrupt so no impact is missed while busy |
| Power | Mean Well RSD-30G/L/H-5 | Railway-rated isolated 5 V from any 12-80 V truck |
| Cab | WS2812 LED bar, 85 dB piezo, IP67 button | Check status, advisory chime and mute in one small pod |
Mechanical
Printable CAD
10 parts, generated from code and checked as closed solids. Pick a part to inspect it, or download the STL files to print.
See the CAD parts render

Loading the CAD…
Real CAD · drag to rotate · scroll to zoomReal CAD · swipe sideways
Bought hardware fitted inside this part (checked in the CAD: nothing collides with the plastic)
- Camera unit frontcam-front.stl · ASA
- Camera unit backcam-back.stl · ASA
- Camera gasketcam-gasket.stl · TPU 95A
- 35 degree guard clampguard-clamp.stl · PA12-CF or ASA
- Clamp backer barclamp-backer.stl · PA12-CF or ASA
- Isolation pad (x4)iso-pad.stl · TPU 85A
- Cab pod frontcab-front.stl · PETG
- Cab pod backcab-back.stl · PETG
- Converter trayconverter-tray.stl · PETG
- Pre-shift QR tag holderqr-tag.stl · PETG
Open risks we are working on
- Buyers or insurers may read it as a safety system; labels, contract and insurance must say it is a driver aid
- Commodity AI warning kits may add cloud logs and squeeze the price
- IMX500 recall on crouching or partly hidden people in cluttered aisles is unproven
- Pi Zero 2 W and microSD reliability under forklift vibration and key-off power cuts
Design stage only: nothing on this page is for sale. Specifications and prices are targets.