Knowing what 'normal' is
Flagging change requires a model of the baseline — which the system has to build for itself.
Application 01 · ISR
Intelligence,
surveillance,
reconnaissance.
ISR is the work of finding out what is happening somewhere, so that decisions can be made on fact rather than assumption. It is the single largest and most immediate application of autonomy in defence — and the one where the gap between what is needed and what is affordable is widest.
The Problem
Watching is expensive.
Surveillance is fundamentally a problem of attention. Watching an area properly means watching it continuously, and human attention degrades badly over long, uneventful periods — which is precisely what good surveillance consists of. Staffing it properly is expensive; staffing it improperly is worse than not doing it.
Reconnaissance has a different cost: to answer a question about a place, someone usually has to go there. That exposure is often the most dangerous part of an operation, and it is undertaken to obtain information rather than to achieve anything at the location itself.
How Autonomy Changes It
Persistent watch,
without persistent staffing.
An autonomous system does not get bored. It can hold a pattern for as long as its endurance allows, notice what changes, and escalate only what merits a person's attention.
- PersistenceSustained observation over long periods, at a cost that makes continuous coverage realistic.
- Reconnaissance without exposureSending a machine to answer the question instead of a person.
- Conclusions, not raw feedReporting what changed and why it might matter, rather than hours of video for someone to review.
- Attention where it countsOperators supervising several systems and spending their judgement on the exceptions.
ISRObserve · Understand · Report
PersistenceLimited by endurance, not attention
ReportingStructured findings, not raw video
LinkAssumed intermittent
DecisionAlways made by a person
StageResearch & development
What It Demands
Why ISR is a
hard autonomy problem.
Suppressing false alarms
An ISR system that reports constantly is quickly ignored, which makes precision more valuable than sensitivity.
Distance and scale
The subjects of interest are usually far away and small in the frame.
Endurance
Persistence is an energy problem before it is a software problem.
Bandwidth discipline
There is rarely enough link to send everything, so the system must decide what is worth sending.
Auditability
Anything that informs a decision has to be reviewable afterwards.
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.
Talk To Us
Evaluating
autonomous ISR?
We share technical detail with programmes and evaluators under appropriate terms.