Drift
Every dead-reckoned estimate accumulates error over time. The question is how fast, and how you correct it.
Capability 04 · Navigation
Navigation in
denied conditions.
Knowing where you are sounds like a solved problem — until satellite navigation is jammed, spoofed or simply unavailable. For a defence system, assuming GPS will be there is a design flaw. We treat its absence as the normal case.
The Problem
Satellite positioning is
the first thing to go.
Global navigation satellite systems broadcast a very weak signal from a very long way away. That makes them straightforward to jam and, more concerning, to spoof — feeding a receiver a plausible but wrong position. A platform that silently believes a false position is more dangerous than one that knows it is lost.
So we build navigation that degrades honestly. When satellite positioning is unavailable or untrustworthy, the system should keep functioning using what it can see and sense directly — and it should know how much confidence to place in its own estimate.
The Approach
Position from
first principles.
There is no single replacement for satellite navigation. What works is combining several independent methods, each with different failure modes, into one estimate that is more robust than any of them alone.
- Visual odometryEstimating movement by tracking how the world shifts across the camera between frames.
- Mapping & relocalisationBuilding a map of the surroundings while working out your position within it, and recognising places already seen.
- Inertial dead reckoningUsing motion sensors to carry the estimate forward through gaps — accurate over short intervals, drifting over long ones.
- Fused estimationWeighting every source by how much it can currently be trusted, and flagging when the estimate is degrading.
NavigationGPS-Denied
AssumesSatellite positioning unavailable
Primary sourcesVision and inertial sensing
Key propertyHonest confidence estimate
Failure modeDegrade, never silently drift
StageResearch & development
Why It's Hard
The failure modes
that matter.
Navigation errors compound. A small mistake early becomes a large one later, which is why the discipline is mostly about managing error rather than eliminating it.
Featureless environments
Open water, desert, snow and darkness give vision very little to work with.
Spoofing
A false satellite fix is worse than none, because the system does not know it is wrong. Detecting that is its own problem.
Repetitive scenes
Places that look identical to each other cause confident relocalisation to the wrong one.
Compute cost
Vision-based navigation is expensive, and it competes with perception for the same hardware.
Knowing you're lost
Recognising that the estimate has degraded past usefulness — and saying so — is a feature, not a failure.
Everything described here is under active research and development. Nothing has been fielded, and we do not publish specifications or performance figures. Our first end-to-end mission test is planned for December 2026. Technical detail is shared directly with partners, evaluators and the armed forces under appropriate terms — get in touch.
Go Deeper
Navigation is a
founding problem.
If you work on state estimation, SLAM or resilient positioning, we should talk.