Ground beats air

When people picture robotic security on a large site, many picture a drone. It is a natural image. A drone sees a lot of ground quickly, it looks impressive, and aerial footage of a perimeter is easy to understand.

MoLogica builds on the ground instead. The short reason is persistence. A drone spots. A ground robot sits on the problem.

What a security patrol actually needs

Strip a night patrol down to its jobs and most of them are slow, close and repetitive:

• Be present on the perimeter for the whole night, not in bursts.
• Get close enough to a gate, padlock, seal or cage to say whether it is secure.
• Stop at the same checkpoints on every pass and photograph them.
• When something is detected, hold position and keep watching until people arrive.
• Do it again tomorrow night, and the night after.

Very little of that rewards speed or altitude. Almost all of it rewards staying power and getting physically close.

Four reasons the ground wins for patrol

It’s quieter. Rotors are loud, and a drone overhead announces itself. A wheeled robot moving along a fence line at night makes far less noise, carries no headlights and has no engine. People planning to cut a fence have less warning that it is coming.

It loiters indefinitely. A drone spends energy just staying in the air. A ground robot that stops spends almost nothing to hold its position. When a robot reaches a detection and holds, it can stay there, watching, for as long as the control room needs. When its battery runs low it docks, and another robot in the rotation takes its place.

It carries more sensors. Weight is the constant constraint on anything that flies. A ground body can carry a full sensor kit: optical and near-infrared cameras with illuminators, depth sensing, RTK GPS, two-way audio and, as an upgrade, long-range thermal. The industrial body MoLogica designs around carries a 50 kg payload. Nothing on that list has to be traded away to stay airborne.

It gets physically close. An inspection verdict is only as good as the image behind it. To say “padlock on” or “trailer sealed” with confidence, the camera has to be close and square to the object. A ground robot drives up, stops, faces the door and takes the photograph. That close, deliberate check is what produces a verdict like “Roller door 3: CLOSED. Slats all the way down to the ground, no gap.”

Spot, then sit

The ground approach still uses range. The night sensor kit is designed so that thermal imaging, as an upgrade module, can detect a person beyond 200 m. Thermal spots the heat signature at a distance. The robot then drives over, and the near-infrared cameras identify the target up close.

From there the robot does the part a drone struggles to do for long. It holds position, keeps the target in view and keeps the control room informed with images, while the response team attends. It never pursues. The rules of engagement are designed around exactly this: alert first, hold, protect the guard, log everything.

Inside buildings, wheels are the natural fit

Indoors the case for the ground is stronger still. Warehouse aisles, data centre rack rows, plant rooms and the floors of a multi-storey building are cramped, full of obstacles and often GPS-poor. A small wheeled robot drives them comfortably, checks cage doors and fire exits, reads rack labels on the same circuit and docks itself to charge. Where there are stairs or rough ground, legged bodies take over, carrying the same brain and sensor kit.

Where drones still help

None of this means drones are useless. They do some things very well, and MoLogica uses what they produce.

The clearest example is the digital twin. Before any hardware is ordered for a mine perimeter, MoLogica’s approach is to build a simulated copy of the site from satellite and elevation data, with fence-level detail from the drone or survey scans most mines already have. Aerial mapping is a strong input to planning a ground patrol.

So the line is not “air bad, ground good”. Aerial survey suits mapping and occasional overviews. A persistent night patrol that inspects at close range suits the ground.

The trade-offs of staying on the ground

Ground robots have limits worth naming.

• They are slower over open ground. A perimeter of several kilometres needs several robots on randomised routes to keep gaps short. Fleets are sized to the loop, the speed, the duty cycle and the randomisation overhead.
• Terrain matters. Wheels do most of the work, and legs only go where wheels cannot. Mine fleets are designed with industrial wheeled bodies on the fence line and legged bodies on rough sections.
• They need docks. Ground robots self-dock on chargers placed around the site and rotate so several are always on patrol. Self-docking is proven in simulation. The dock hardware and the perimeter fleets are designed.

Those are fleet design problems with known answers. The limits of flight time, payload and noise are harder to design around for an all-night patrol.

Persistence is the product

Security on a long perimeter is about presence over hours, not glimpses over minutes. Every night, randomised, checkpoint by checkpoint, with a verdict and an image at each stop. A robot that can stop, wait and look closely fits that job. A ground robot is built to stay.

Where this stands

Randomised patrols, close-range inspection verdicts and self-docking are proven in simulation, in a simulated warehouse depot on indoor robot bodies. Thermal detection at range, RTK GPS, industrial wheeled bodies for perimeters and the rules of engagement are designed. Legged bodies and mine perimeter twins are on the roadmap. The first physical robot is on the bench.

Book a demo to see a ground patrol and its inspection verdicts in simulation.