Positioning systems answer where things are continuously rather than at read points. They are the most capable and most sensitive sensing option.
The technologies
Ultra-wideband, which gives sub-metre accuracy and needs dense infrastructure.
Bluetooth beacons, cheaper, with room-level rather than position-level accuracy.
Wi-Fi positioning, using existing infrastructure, with poor accuracy in racked environments.
Vision-based, covered separately, with its own issues.
Inertial and dead reckoning, on vehicles, which drifts and needs correction.
Accuracy requirements should be derived from the question, not maximised. Zone-level accuracy answers most operational questions at a fraction of the cost.
What it enables
The gaps between scans become visible. Travel, searching and waiting can finally be distinguished.
Congestion mapping, by location and time.
Actual travel paths rather than inferred ones, which reveals routing that the task sequence does not imply.
Equipment location, which removes searching for a truck.
Safety applications: proximity detection between vehicles and people, which is the strongest justification available.
Search time for stock, measurable for the first time.
The infrastructure reality
Racked warehouses are hostile to radio. Metal, density and height all degrade accuracy, and vendor coverage estimates are made in easier environments.
Coverage is a building project: anchors, cabling, power, mounting at height, and survey.
Calibration decays as racking and stock change.
Test in your own facility, fully loaded, before committing. A pilot in an empty aisle proves nothing.
The surveillance question, immediately
Positioning worn or carried by people is continuous location tracking of identified individuals, and it must be treated as such.
It records break locations, toilet visits, conversations and time spent anywhere.
Obligations apply in most jurisdictions: notice, proportionality, purpose limitation, consultation, retention limits.
Settle the position before deployment, not after — the full treatment is in the worker monitoring section.
Tag the equipment rather than the person wherever the question can be answered that way, which is most of the time. A forklift's position answers congestion and utilisation questions without tracking anyone.
Designing to limit exposure
Tag assets, not people, as the default.
Zone resolution rather than position where zones answer the question.
Aggregate at ingestion, discarding individual traces once the zone occupancy is computed.
Exclude sensitive areas from tracking entirely: welfare facilities, break rooms, medical rooms. This should be a hard technical exclusion, not a policy.
Short retention for any individual-level trace.
Each of these reduces both the legal exposure and the trust cost, and almost none of them reduces the operational value.
When it is worth it
For safety, particularly vehicle and pedestrian separation, where the case is strong.
Where search time is a demonstrated major loss.
Where the scan gaps are large and unexplained and the operation has already exhausted what scan data can tell it.
Not as a first project, and not because a vendor demonstrated a moving dot on a floor plan. Establish what the existing data cannot answer, then decide whether the answer is worth the infrastructure and the trust.
Testing in a loaded facility
Vendor coverage estimates are made in conditions unlike a warehouse.
Pilot in a fully racked, fully stocked aisle, not in an empty one.
Test at the heights that matter, since accuracy varies vertically.
Test with equipment moving, which introduces multipath and interference.
Test at peak occupancy, because bodies and metal both absorb signal.
Measure the actual accuracy achieved against what the questions require, and accept zone resolution where zone resolution answers them.
Tagging the asset instead
Most questions attributed to people tracking are answerable from equipment tracking.
Congestion: where equipment queues.
Utilisation: where machines are and for how long.
Travel: the paths equipment takes.
Search time for stock, by tagging pallets rather than people.
Safety proximity, by detecting equipment approaching a pedestrian zone rather than tracking the pedestrian.
Ask, for every proposed person-tracking use, whether the equipment answers it, because it usually does and it removes the entire obligations problem.