Somewhere in a warehouse right now, a manager is staring at a screen and asking one of operations’ oldest questions:
Where is everything?
Where is forklift 14? Where is the pallet that was definitely here yesterday? And where is that very expensive piece of equipment that appears to have developed free will?
Hamlet had his skull. Operations teams have missing carts.
And both are faced with a difficult question:
To tag, or not to tag?
When organizations want to track people, vehicles, or assets in real time, they usually consider two main approaches.
One uses cameras and computer vision. The other uses radio tags and real-time location systems, or RTLS.
Both promise visibility. But they arrive there by very different routes.
A camera looks at the world and tries to work out what it is seeing.
A tag simply says:
Hello. I am forklift 14. I am over here.
That difference matters more than it may first appear.
The Camera Makes a Strong Entrance
Computer vision is impressive.
Give a system enough cameras, enough processing power, and enough training data, and it can recognize people, vehicles, pallets, safety equipment, and all sorts of other things moving through a facility.
It can do this without attaching anything to the object being monitored, which is understandably attractive.
No tag. No badge. No device.
Just cameras, software, and the reassuring glow of artificial intelligence.
For visual verification, incident review, security footage, or checking whether someone is wearing a hard hat, this can work very well.
The trouble begins when the camera is asked not merely to see something, but to remember exactly what it saw, follow it everywhere, distinguish it from nine identical objects, and never become confused by darkness, dust, glare, racks, doors, people, or reality.
That is quite a demanding job description.
A Camera Can Only See What It Can See
The great weakness of a camera is surprisingly literal.
It needs to be able to see.
If a pallet moves behind a rack, a worker steps behind a machine, or a cart enters a storage area outside the camera’s field of view, the system may lose track of it.
In computer vision, this is called occlusion.
In a warehouse, it is called Tuesday.
Industrial facilities are full of shelves, columns, vehicles, containers, doors, walls, machinery, and people moving in inconvenient directions. One thing is almost always blocking another thing.
Additional cameras can reduce blind spots, but this often leads to more mounting, more cabling, more network traffic, and more places where someone must eventually climb a ladder.
A tag-based system does not need a perfect view. Radio signals can continue to communicate when an asset moves behind equipment, around a corner, or into an area where a camera cannot follow.
The pallet may disappear from sight.
It has not disappeared from the system.
Cameras Have Opinions About Lighting
Cameras are also rather particular about their working conditions.
They prefer good lighting, clean lenses, limited glare, clear air, and objects that behave politely in front of them.
Warehouses and factories are not famous for providing these things.
A loading dock can be dark in the morning, flooded with sunlight at noon, and covered in dust by the end of the shift. Cold storage creates condensation. Outdoor sites bring rain and fog. Industrial environments bring steam, dirt, shadows, and the occasional forklift parked directly in front of the camera.
Computer vision depends on image quality.
Tags do not.
A radio signal does not care whether the lights are on. It does not become dazzled by sunlight. It does not need someone to wipe dust from its lens.
At three in the morning, in a dim warehouse, forklift 14 is still forklift 14.
Identical Assets Are a Vision System’s Version of Twins in a Soap Opera
Recognizing an object category is one thing.
Maintaining the identity of one particular object is another.
A camera may correctly identify three forklifts moving through an aisle.
But which one is forklift 14?
Which one is due for maintenance?
Which one belongs to the loading area?
Which one has been idle for four hours?
Now imagine that the forklifts cross paths, disappear behind a rack, and reappear in another camera view.
At this point, the system is not merely observing. It is solving a logic puzzle.
The same problem appears in hospitals with wheelchairs, infusion pumps, beds, and carts. Many assets look exactly alike because, inconveniently, they were designed that way.
A tag avoids the problem altogether.
Every tag has a unique identity.
The system does not have to infer whether it is looking at cart 14 or cart 15. Each cart introduces itself.
This makes it easier to support applications such as:
- Asset search
- Utilization monitoring
- Maintenance tracking
- Movement history
- Zone alerts
- Inventory visibility
- Chain-of-custody workflows
The system is not thinking, “That looks like cart 14.”
It knows.
Cameras Watch People. People Notice.
There is also the privacy question.
Cameras can capture faces, behavior, conversations, screens, visitors, and all sorts of information that has nothing to do with the original tracking objective.
Even when facial recognition is not being used, employees may still feel uncomfortable being watched throughout the day.
A badge or wearable tag is different.
The system tracks the location of an assigned device, not a person’s face. The data can also be anonymized or restricted according to role and purpose.
This does not remove the need for clear policies, transparency, and responsible data use.
But “the system knows badge 207 entered zone B” is usually a simpler proposition than “the system recorded several hours of video and used AI to interpret what everyone was doing.”
One feels like location technology.
The other can feel like an audition nobody agreed to attend.
Cameras Generate Footage. Tags Generate Identity.
This is the central difference.
A camera sees an image and tries to interpret it.
A tag provides a direct signal linked to a known ID.
That identity can belong to:
- A forklift
- A medical device
- A container
- A worker badge
- A patient wearable
- A tool
- A vehicle
- An infant security tag
Once identity is built into the system, location data becomes much more useful.
You are no longer asking:
“Is there a forklift in this area?”
You can ask:
“Where is forklift 14?”
“Has it entered a restricted zone?”
“How long has it been idle?”
“How often does it travel through this aisle?”
“When is it due for maintenance?”
That is the difference between spotting an object and managing an operation.
Tags Can Also Do More Than Report Location
A tag is not necessarily just a location beacon.
Depending on the device and application, it may also include:
- A panic button
- Motion detection
- Immobility alerts
- Tamper detection
- Temperature sensing
- Battery monitoring
- Zone-entry alerts
- Two-way communication
A camera cannot be pressed during an emergency.
It cannot travel with a worker into a blind spot.
It cannot report its own battery level because, being a camera, it considers that someone else’s problem.
Tags can become active parts of safety and operational workflows, not merely passive objects on a map.
This Does Not Mean Cameras Are Useless
Cameras are very good at showing what happened.
They provide visual context. They can help investigate incidents, verify events, observe behavior, and understand how a process unfolded.
Sometimes a picture really is worth a thousand words.
But a picture is not always worth a unique asset ID.
For many operations, the strongest approach may involve different technologies doing different jobs.
Use cameras when you need visual evidence.
Use tags when you need reliable identity and continuous location.
The mistake is assuming that because a camera can see something, it can always track it accurately, identify it consistently, and follow it through a complex environment.
That is a little like assuming that because someone saw your suitcase at the airport, they know where it is now.
Where Litum Fits In
Litum has spent more than two decades building real-time location systems for environments where objects disappear behind racks, people move between zones, lighting is inconsistent, and equipment rarely stays where someone left it.
In other words, real environments.
Litum uses technologies such as ultra-wideband, Bluetooth Low Energy, LoRaWAN, and GPS depending on the use case, required accuracy, facility, and scale.
Ultra-wideband can provide high-precision indoor location.
Bluetooth Low Energy can support room-level or zone-level visibility.
LoRaWAN can extend connectivity across larger areas.
GPS can support outdoor tracking.
The point is not to force every application into one technology.
The point is to use the right one for the job.
Tags, anchors, gateways, and location software work together to provide live location data, zone alerts, geofencing, heatmaps, movement history, utilization metrics, and other operational insights.
This can help a hospital find a wheelchair before a patient transfer is delayed.
It can help a manufacturer locate tools without sending three people to search for them.
It can help a warehouse understand forklift traffic.
It can help a safety team receive an alert when a worker presses a panic button or remains immobile.
It can help a maternity ward detect when an infant tag approaches an unauthorized exit.
None of these situations requires a camera to recognize a face.
They require the right tag, the right infrastructure, and a system that knows exactly what it is listening for.
So, To Tag or Not to Tag?
Computer vision asks software to look at the world and make a very educated guess.
A tag introduces itself.
In controlled spaces with good visibility, cameras can be useful. But in busy facilities with poor lighting, blind spots, identical equipment, privacy concerns, and assets that refuse to remain photogenic, tags provide a more dependable foundation for real-time location.
Hamlet never got a particularly satisfying answer to his question.
Operations teams can.
To tag.
