Accuracy is a humanitarian technology as much as a military one. A system that reaches exactly what it was directed at, and nothing else, is the difference between a proportionate action and an indiscriminate one. That is the engineering problem we work on.
The Case For Precision
Accuracy is restraint.
§ 01 / RATIONALE
Before guidance, hitting one specific point meant delivering a great many unguided munitions and accepting whatever else was within the error radius. Precision inverted that arithmetic: fewer effects, smaller warheads, less collateral damage, and a far greater chance that what was intended is what actually happens.
This is why we regard precision as an accountability technology. A commander can only be answerable for an outcome they could reasonably predict. Tightening the error radius is what makes prediction — and therefore responsibility — possible. Our work here is guidance and accountability engineering; a person authorises every action, always.
How Guidance Works
The accuracy chain.
Guidance is not one technology but a chain of them, each covering the phase where the others are weakest. These are the publicly understood principles the field is built on.
§ 02 / PRINCIPLES
01
Inertial navigation (INS)
An inertial measurement unit senses acceleration and rotation and integrates them into a position estimate. It needs nothing from the outside world — which makes it unjammable — but its error grows steadily with time, so it cannot work alone over long flights.
02
Satellite navigation (GNSS)
A satellite fix corrects inertial drift and holds accuracy roughly constant over distance. The weakness is the opposite of the INS: the signal is weak, and can be jammed or spoofed by an adversary.
03
Coupled INS/GNSS
Fusing the two covers both weaknesses. Satellite fixes bound the inertial drift; the inertial solution carries the system honestly through periods when satellite navigation is unavailable or untrustworthy.
04
Midcourse guidance
The long phase where the system travels toward the area of interest, flying a planned trajectory on the coupled navigation solution and updating it as conditions change.
05
Terminal guidance
The final phase, where a seeker looks at the actual scene and refines aim onto the specific designated point rather than a set of coordinates — the step that separates good accuracy from precision.
06
Circular error probable (CEP)
The standard measure: the radius within which half of all deliveries fall. Every element of the chain contributes error, and the total is what CEP expresses.
Terminal Seekers
Looking at the scene, not the coordinates.
Coordinates can be slightly wrong, and things move. A terminal seeker resolves that by sensing the actual scene in the final phase. Each seeker type trades differently between weather, cost and the kind of designation required.
Semi-active laserHomes on laser energy reflected from a point designated by an operator or an aircraft. Very precise, and it keeps a human explicitly in the loop — someone must actively designate.
Imaging infraredForms a thermal image and matches it against the intended aimpoint. Works at night and needs no external designation, at higher cost and with some weather sensitivity.
Millimetre-wave radarSees through cloud, dust, smoke and rain where optical seekers struggle — the all-weather option in the family.
Multi-modeCombining seekers so that the failure conditions of one are covered by another, which is the direction most modern work takes.
Guidance ChainPhases
LaunchAlignment and handover
MidcourseCoupled INS/GNSS navigation
TerminalSeeker refines onto the aimpoint
MeasureCircular error probable (CEP)
AuthorityHuman authorisation, always
StageResearch & development
Our Work
Where we contribute.
We are an autonomy and guidance software company. Our contribution is the navigation, perception and decision layer — the same core that runs our unmanned aerial and tactical systems — applied to the accuracy chain.
Navigation that survives denialGuidance is only as good as its position estimate. Our GPS-denied navigation work applies directly here.
Perception for terminal phasesRecognising the designated aimpoint reliably, in poor conditions, is a perception problem before it is anything else.
Honest uncertaintyA guidance system that knows how confident it should be can abort rather than proceed on a bad solution. We treat that as a required feature.
Abort and authorityThe ability to withhold, redirect or abort is designed in from the start, not added afterwards.
Our ContributionSoftware
We buildNavigation, perception, decision software
Applied toThe accuracy chain
Shared withUAV and tactical programmes
ConstraintHuman authorises every action
StageResearch & development
Non-Negotiables
Limits we design in.
These are constraints on what we will build, not options a customer can configure away.
§ 03 / LIMITS
01
A person authorises
No system we build selects and engages on its own authority. Authorisation is human, explicitly and always.
02
Abort is always available
The ability to stop or redirect is a first-class requirement, not a fallback.
03
Refuse on low confidence
A system unsure of its solution should decline rather than proceed.
04
Reconstructable afterwards
Every decision recorded so an action can be examined and accounted for.
This page explains publicly documented guidance principles at a conceptual level. We do not publish designs, parameters, performance figures or any implementation detail. All work is under active research and development; nothing has been fielded. Engagements are subject to end-use and end-user assessment, sanctions and export-control screening — see security & compliance and human rights.
A system that uses guidance to reach a specific designated point with a far smaller margin of error than an unguided equivalent. That margin is normally expressed as circular error probable, or CEP — the radius within which half of all deliveries fall.
They fail in opposite ways. An inertial measurement unit needs nothing from outside and cannot be jammed, but its error grows steadily with time. Satellite navigation holds accuracy roughly constant but the signal is weak and can be jammed or spoofed. Coupled together, satellite fixes bound the inertial drift and the inertial solution carries the system honestly through periods when satellite navigation cannot be trusted.
The final phase, where a seeker senses the actual scene rather than trusting coordinates. Common approaches include semi-active laser, which homes on energy reflected from a point a person actively designates; imaging infrared, which matches a thermal image; and millimetre-wave radar, which works through cloud, dust and smoke.
No. A person authorises every consequential action. We do not build systems that select and engage on their own authority, abort is always available, and a system unsure of its solution is designed to decline rather than proceed. This is a limit on what we will build, not a setting a customer can change.
No. We explain publicly documented guidance principles at a conceptual level only. Designs, parameters and performance figures are never published, and technical material is shared with government programmes and evaluators only after end-use and end-user assessment.