Defense and critical infrastructure: persistent surveillance with unmanned systems

Protecting critical infrastructure (pipelines, power corridors, dams, ports, remote facilities) is a problem of coverage per unit of cost. Ground patrols cover little and arrive late; fixed cameras cover points, not corridors. And there is a feature of the Colombian context worth stating up front: the same infrastructure you want to watch is protected by the aviation framework that restricts drone flights near it, so operating here requires, first, complying with that regulation.

What do unmanned systems contribute?

• Coverage of linear corridors. A scheduled flight covers tens of kilometres and returns images comparable with the previous flight.
• Change detection, not object detection. What matters is not seeing a truck: it is knowing that an excavation, a diversion or a new structure appeared where there was nothing last week.
• Prioritised response. When the system classifies the change by severity, the field team goes where it needs to go.
• Traceable evidence. Each detection is dated and georeferenced for legal or reporting use.

What framework must be met before flying?

Aerocivil’s (Colombia’s civil aviation authority) RAC 100 (in force since 2023, replacing RAC 91 Appendix 13) requires a minimum separation of 2 km from military bases, power plants, dams and other critical infrastructure marked on the Visor Geográfico UAS (Aerocivil’s UAS geographic map viewer), unless expressly authorised. In practice, this means a surveillance programme over critical infrastructure does not start with the flight: it starts with the authorisation.

Two things are almost always triggered in this type of operation:

• Specific category and BVLOS. A corridor of tens of kilometres exceeds visual line-of-sight operation almost by definition. That requires a pilot with a BVLOS rating, a SORA risk assessment and membership in a UAS operator registered with Aerocivil.
• Flight authorisation per operation. Each flight requires its own Aerocivil authorisation, with approved dates, equipment, pilots, areas and altitudes; it is not a single permit that covers the whole campaign.

This is not an incidental obstacle: it is the filter that separates a serious operation from an improvised one, and it should be treated as part of the product, not as fine print.

How do you design a realistic programme?

1. Segment the asset. Divide the perimeter or corridor into sections, each with its own risk level.
2. Define frequency per section. Not everything needs daily surveillance; historically affected sections do.
3. Set alert thresholds. Without explicit thresholds, the analytics generate noise and the operator stops looking at them.
4. Close the loop. Every alert must end in a recorded action: verified, dismissed or escalated.

What role do governance and human control play?

In this domain, automation has a clear and deliberate limit: the system reads and carries out collection and comparison tasks; the operational decision remains human, with a record of who decided what and on what evidence. That traceability is as important as the detection itself.

Why does interoperability matter?

An isolated surveillance programme ages quickly. Aerial collection must coexist with ground sensors, field reports and existing control systems. In practice, this means an orchestration layer that normalises different sources and exposes a single view of the asset’s status.

Frequently asked questions

Can you fly a drone near a dam or substation in Colombia without a special permit?

Not freely. RAC 100 requires 2 km of separation from critical infrastructure marked on the Visor Geográfico UAS, unless Aerocivil expressly authorises it.

What is needed to operate long corridor flights (BVLOS)?

A UAS pilot certificate with a BVLOS rating, a SORA risk assessment, membership in a registered UAS operator, and a specific Aerocivil flight authorisation for each operation.

Want to assess a pilot over a specific corridor? Write to us at hello@bluberri.io.