Edge Telematics and Speed-Limit-Aware Safety Scoring for 150+ Tankers
Driver speed was judged against a blanket policy, not the road's posted limit. We fitted on-vehicle edge devices reporting every five seconds and matched every reading against the limit in force at that place and time.
- Service
- Software Solutions
- Industry
- Transportation & Logistics
- Region
- United States
- Engagement
- R&D Lab (joint innovation)
Outcomes
- 150+ tankers instrumented
- 5-second telemetry
- Recognition for safe drivers
Stack
Raspberry Pi · Python · Telematics APIs · Geofencing · Dashboards
Written byOriental Nippon Team
A US oil and gas distributor runs a fleet of more than 150 tankers. Safety matters to it commercially and legally, but its picture of driver behaviour was coarse: one general speed policy, with no reference to the limit on the road actually being driven.
It wanted to hold drivers to real road limits, and to be able to recognise good driving, not just react to incidents.
The problem
- Blanket policy. Speed was judged against a general rule, not the posted limit.
- No route rules. Lower limits on specific stretches could not be enforced.
- Low-resolution data. Position and speed were not captured often enough to see behaviour.
- Punish-only. There was no basis for rewarding good driving.
| Before | After | |
|---|---|---|
| Fleet covered | 150+ tankers | 150+ tankers, fitted |
| Data frequency | Coarse | Position and speed every 5 seconds |
| Speed reference | One general policy | Limit in force at that location |
| Route restrictions | Not enforceable | Geofenced |
What we built
On-vehicle devices
Raspberry Pi units across the fleet, reporting position and speed every five seconds.
A data path
Telemetry pushed into telematics and speed-reference services, so each reading is compared with the limit at that exact location.
Geofencing
Route-specific restrictions enforced as their own limits.
A dashboard
Driver and vehicle behaviour in a form operations can act on.
Every five seconds, for every tanker:
Technology and the role it played
| Technology | Role in this project |
|---|---|
| Raspberry Pi | On-vehicle capture devices |
| Python | Device software and the data pipeline |
| Telematics and speed-reference APIs | Matching each reading to the posted limit |
| Geofencing | Route-specific restrictions |
| Dashboards | Driver, vehicle and route views for operations |
How we delivered it
Our R&D lab worked with the client's operations team to define "unsafe" in measurable terms before any hardware was fitted.
- Frame the question. Measures and success threshold.
- Prototype. Devices on a handful of vehicles.
- Prove the data path. Capture frequency, transmission reliability and accuracy of limit matching.
- Fit the fleet in stages. Validating each stage's data before the next.
- Build reporting. Only once the data was trusted.
01Framing
1–2 wks
Definition of the measures, success threshold
02Prototype
3–4 wks
Devices on a handful of vehicles, data path proven
03Geofencing and limit matching
2–3 wks
Route restrictions, reference-data integration
04Fleet fitting
6–10 wks
Staged installation with per-stage data validation
05Dashboard
3–4 wks
Operations reporting, driver and route views
Risks and how we handled them
| Risk | How we handled it |
|---|---|
| Unclear definition of unsafe | Measures and thresholds agreed before fitting hardware |
| Unreliable data from the road | Data path proven on a handful of vehicles first |
| Problems multiplied across 150+ vehicles | Staged fitting with data validated at each stage |
| A system drivers resent | Designed to recognise good driving as well as poor |
Outcome
The distributor can measure safety against the limit that actually applied at that time and place, including route restrictions, and use the same basis to recognise good driving as well as address poor driving.
Recognising good driving is what the client wanted the system for. A monitoring system that can only punish rarely survives contact with a workforce.
What the client keeps
- Device configuration and firmware
- Pipeline and integration code, in their own repository
- Geofence definitions
- The dashboard
Sources and further reading
- Speeding, NHTSA
- Raspberry Pi documentation, Raspberry Pi
Facing something similar? This is the kind of work our Software Engineering & Cloud team delivers every day. Talk to us about your stack.
