Tunnels, cuttings, parking structures and urban canyons put a guaranteed GNSS dead zone on every route. OSCP builds photonic and MEMS IMUs, ITAR-free and made in Canada, that keep dead reckoning honest when satellites and cameras cannot.
The challenge
For road and rail alike these dead zones are on every run, not rare faults. During an outage the stack dead-reckons, and inertial drift accumulates over time, so the IMU decides how far the position estimate has wandered by the time the fix returns.
Rail has one advantage worth using. The route is fixed, so every tunnel, cutting and station box on it can be enumerated in advance. The worst denial window is not a guess, it is a measurement, and the position error it produces can be sized before a train ever runs. Road cannot do that, which is why road programs carry a worst-case assumption instead.
What both share is the failure mode. Perception-based localization degrades exactly when it is needed most. Rain, snow and glare hurt lidar and camera matching, tunnel walls give a scan matcher very little to work with, and long featureless track offers nothing to fix against. The IMU measures the vehicle’s own motion and does not depend on seeing anything.
And both shake. MEMS gyros are g-sensitive, so vibration and hard manoeuvring leak into heading error. Rolling stock delivers that vibration continuously, for the life of the vehicle.
How OSCP helps
Drift you can put a number on. Measured drift behaviour feeds directly into the position-error budget you carry for a worst-case outage. Unaided inertial error is never bounded, but it accumulates slowly and predictably enough to size against, which is what a fixed rail route needs.
Optical sensing without g-sensitivity. The MK2E2 measures rotation via the Sagnac effect in a photonic core with no moving parts in the sensing path, highly resistant to vibration and EMI, and free of MEMS g-sensitivity on rough pavement or continuous rolling-stock vibration.
Drops onto the vehicle bus. Both units speak RS-422 and CAN-FD at up to 500 Hz on a 12 to 34 V supply, so they connect to automotive and rail vehicle architectures without protocol bridges.
Fleet-ready hardware. Built to withstand harsh environments, which covers rolling stock through a full winter. Vibration passed at 2, 4 and 8 g and shock at 20 g and 40 g, both to DO-160 style profiles. Class A emissions and ESD passed. ITAR-free and made in Canada.
Recommended configuration
| Product | Grade | Choose it when |
|---|---|---|
| MK2E2 | Tactical Grade Photonic IMU, 0.5 °/hr min in-run bias, typically under 1.0 °/hr | Your error budget has to survive multi-minute outages: long tunnels, extended urban running, or continuous vibration on rolling stock. The lead choice for production programs. 78 x 63 x 36 mm, 230 g, 2 W. |
| MK2M2 | Tactical Grade MEMS IMU, 0.5 °/hr min in-run bias, typically under 1.0 °/hr | Development fleets, validation vehicles, instrumentation cars and short outage profiles. 40 x 40 x 25 mm, 75 g, 1.2 W. |
Both share < 15 µg accelerometer bias stability and the same RS-422 or CAN-FD interface, so shared interfaces keep integration work common across both. The MK2E2 is the larger unit, so it is not a mechanical drop-in.
Proof
We drove 21 minutes through Montreal with GNSS positioning denied, heading from the gyro alone with a 1 Hz speed input for distance and no EKF. The MK2E2 finished 5.7 m from ground truth. The MK2M2 finished 20.5 m. Both ran identical aiding, so the 3.6x gap is set almost entirely by heading drift.
It is a road test, and a street is not a tunnel. But the mechanism is the one that governs both: heading error accumulates with distance travelled, and the gyro decides how fast.
FAQ
How do I turn IMU drift into a position budget?
Inertial error accumulates over time, so the number depends on outage duration and what else is aiding the solution. Start from the worst denial window on your route, the longest tunnel or parking structure, and the lateral or along-track error you can tolerate at its exit. Our measured anchor is 5.7 m after 21 minutes of GNSS denial for the MK2E2, with a 1 Hz speed input. Tell us your route profile and we will work it through with you.
Are these units certified for safety-critical train control?
No. We hold no SIL certification and no EN 50126, EN 50128 or EN 50129 assessment, and we will not imply otherwise. The MK2 family is an inertial measurement unit that feeds your own positioning or odometry solution. If your application is safety-related, the certification burden sits with the system you build and we will support that work with test data rather than claims.
We already have wheel tachometers and balise corrections. Why does gyro grade matter?
Because heading and along-track error accumulate between corrections. Wheel-based odometry loses accuracy on slip and wear, and balises only correct you where they are installed. Between those fixes the inertial sensor carries the estimate, and a lower-drift gyro means the error at the next correction is smaller.
Does the IMU replace lidar or camera localization?
No. It carries the solution between corrections. The IMU measures the vehicle’s own motion and does not depend on visibility, which is why it holds up in a tunnel or when perception degrades.
