Sentinel
Internal Vehicle-Distance Research
Sentinel began as an internal RUC on Rails research project exploring distance measurement using a vehicle's own signals. We developed a working prototype that calculated distance locally, without GPS, cellular connectivity or a continuously connected phone.
Sentinel is retained as company research and is not being developed for commercial release. This page documents the approach and what we learned.
Privacy by Architecture, Not Policy
Our research explored whether distance could be measured using vehicle speed and elapsed time, independently of geographic position. The calculation ran on the device itself.
Location-based telemetry can include coordinate logs, mapped routes and movement histories. Those records were unnecessary for the distance-measurement approach we tested.
Sentinel's measurement protocol has no latitude, longitude or route-coordinate fields. Research recordings captured the speed, timing and diagnostics needed to examine the distance result.
Location Telemetry Can Include
Sentinel's Measurement
Hardware & Embedded Foundation
We developed Sentinel's custom firmware for the WiCAN OBD v340 dongle. Connected to a vehicle's diagnostic port, the prototype requested standard OBD speed readings over the CAN network.
Network discovery listens before diagnostic requests begin. The firmware requests vehicle readings; it does not write ECU settings or issue vehicle-control commands.
The vehicle supplied the signal. Sentinel calculated the distance.
Fully Autonomous Edge Computation
Distance was calculated on the dongle from successive speed readings and the actual time between them. The engine accumulated distance in integer millimetres while retaining fractional remainders.
No GPS Required
Distance derived mathematically from vehicle motion data rather than inferred from geographic position.
No Mobile Data
The device operates independently of cellular connectivity. Computation continues regardless of network availability.
No Cloud Dependency
Distance capture ran locally. Bluetooth allowed recorded research evidence to be retrieved separately, without a continuous server connection.
Speed Integration & Calibration
Vehicle speed was the primary distance source. The firmware also supports gradual calibration from a standard odometer reading where a vehicle supplies it. The recorded campaign used speed integration alone.
Vehicle-Speed Integration
The prototype requested standard OBD speed readings and combined each pair with the actual elapsed time to calculate distance. The campaign recorded approximately 9.4 speed readings per second.
- Standard OBD speed readings
- Actual response timing
- Fractional distance retained
- Calculation on the device
Optional Odometer Calibration
The firmware can use a supported standard OBD odometer signal as an additional calibration reference, gradually refining the distance scale while speed integration continues on the device.
- Additional calibration reference
- Standard odometer reading
- Gradual scale adjustment
- Continuous speed integration
Each calculation uses the actual timing of vehicle readings, with fractional distance carried forward between samples. Built-in timing checks keep the calculation grounded in fresh data by excluding sample gaps longer than 500 ms.
Road-Test Results
Across 35 recorded drives, Sentinel calculated 2,500.685 km against 2,515.4 km shown by the vehicle's trip meter. These results came from speed integration alone, without GPS, mobile data or odometer calibration.
Consistency Across Recorded Drives
Sentinel achieved 99.415% agreement with the combined dashboard trip readings across 35 drives. Analysis of 2,000,680 research records showed continuous record sequences, no long integration gaps and no stationary drift at repeated zero-speed readings throughout the recorded sessions.
Inside the data.
See vehicle signals become distance. Explore two example data samples showing vehicle profiles, response timing and device measurements. The figures here relate to these samples; the research totals above cover the full dataset.
Data Boundaries & Minimisation
The distance engine did not require location coordinates. Research sessions also recorded speed, timing, accumulated distance and device diagnostics so we could examine how the measurement behaved.
Bluetooth retrieved distance snapshots and recorded evidence. The calculation continued independently of that connection, without a live location feed or continuous upload to a server.
Stored on Device
Not Stored
Engineering Depth
Sentinel brought together custom embedded firmware, local distance calculation, persistent storage and research analysis. The work demonstrated a practical approach to measuring distance without relying on satellite positioning or cellular connectivity.
Real-Time Firmware
Custom firmware in C using ESP-IDF and FreeRTOS to coordinate vehicle readings, distance calculation, storage and communication.
CAN Frame Parsing
Listen-only network discovery followed by standard OBD requests. The recorded campaign used a 500 kbit/s CAN connection.
Numerical Integration
Trapezoidal integration of vehicle speed over actual elapsed time, with integer millimetres and retained fractional remainders.
BLE Communication
Authenticated Bluetooth transfer of distance snapshots and recorded research evidence, separate from the local distance calculation.
Flash Persistence
Checksummed distance checkpoints preserve saved totals between drives. On restart, the device recovers the latest valid saved record and resumes accumulation.
Sync Safeguards
Sequenced research records support verification of transferred data. Snapshot acknowledgements track transferred totals while preserving accumulated distance.
What the Research Demonstrated
Sentinel demonstrated local distance calculation using vehicle data, without collecting location coordinates or requiring cellular connectivity. The prototype, firmware and test results are retained as company research. Sentinel is not a released product, and no commercial release is planned.
Privacy-Preserving Distance Computation
Sentinel records what we learned about measuring distance from vehicle data without GPS or cellular connectivity. It remains internal research and is not being developed for commercial release.