The Bathroom Light Turns Off While You Are Still There: What Sensor Position Misses

A bathroom light that switches off while someone is still inside is often blamed on a short timeout. The deeper problem may be spatial: a passive infrared sensor does not detect “a person” in the abstract. It detects changes in infrared radiation across zones in its field of view. A seated user may move very little, while doors and full-height or partial-height cubicle partitions can interrupt the sensor’s view. Sensor position, partition geometry and control logic therefore form one architectural system.

What the primary study tested

Shoaib Azizi, Ramtin Rabiee, Gireesh Nair and Thomas Olofsson examined how the position of a multi-sensor device affected occupancy and indoor-environment measurements in three single-occupant offices. Eighteen devices were installed, six in each office. Every device combined passive infrared motion detection with temperature, carbon-dioxide, humidity and illuminance sensing.

The comparison matters because many projects treat a central ceiling position as neutral. In the study, it was not. The PIR sensor positioned in the middle of the ceiling achieved only 60% presence-detection accuracy, while a device installed under the desk reached 84%. The desk position was closer to movement from a seated occupant’s lower body; the ceiling position had a wider view but was less reliable for presence under the tested conditions. Combining selected PIR positions improved performance further, although the most suitable location for motion detection did not necessarily produce the best temperature or illuminance measurement.

A second field study confirms the stationary-occupant problem

Guy R. Newsham and colleagues collected data every 15 seconds for up to 31 days in 28 single-person offices. A combination of webcam-based motion detection and keyboard and mouse activity produced more than 90% daytime occupancy accuracy and substantially fewer false negatives than a commercial ceiling-mounted PIR sensor. In a nine-day proof-of-concept control test, the combined system reported energy savings of 15–68% with no reported false-negative events.

Those results do not nominate cameras or computer activity for toilets. They show that the apparent simplicity of an occupancy signal hides a choice of sensing method, placement and evidence. A technology that works while someone walks into a room may still fail when the same person becomes quiet.

What the studies did not test

Neither study was conducted in a public or residential bathroom. Azizi and colleagues studied only three single-occupant offices, so the 60% and 84% values belong to those rooms, occupants and device positions. The study compared multi-sensor locations, not toilet cubicles, disability-accessible layouts or different privacy partitions. Newsham and colleagues relied partly on computer activity and a webcam-derived signal, sources that are inappropriate or unavailable in a bathroom. Its nine-day demonstration was a small proof of concept, not a long-term estimate of maintenance, privacy or user acceptance.

PIR performance also changes with distance, movement direction, sensor optics, ambient temperature, clothing, timeout and obstructions. An ultrasonic, thermal-array or millimetre-wave sensor may detect different kinds of stillness, but each introduces its own commissioning, privacy, cost and false-positive questions. The evidence therefore supports testing the installed system, not declaring one sensor type universally correct.

Architectural application: design the occupied zone

For a cubicle-based bathroom, ArchUp proposes that the lighting-control drawing identify three zones separately: approach and entry, standing activity at basins or urinals, and seated occupancy inside each enclosed cubicle. The proposal is an architectural inference from office evidence, not a tested bathroom standard.

• Plot the sensor’s real field of view after doors, partitions, bulkheads and accessories are fixed, rather than over an empty reflected-ceiling plan.
• Check whether the seated body lies within a detectable zone when the cubicle door is closed.
• Separate the minimum light needed for safe use from the general room scene, so one missed presence event does not create complete darkness.
• Compare one central sensor with zoned, dual-technology or cubicle-level detection using the same privacy and maintenance criteria.
• Provide a clear recovery path, such as manual override or immediate reactivation, without requiring exaggerated body movement.

A longer timeout can mask false absence, but it also delays energy savings and does not correct a blind zone. A second sensor can improve coverage while increasing commissioning burden. The design choice should therefore be evaluated as a control system rather than as a single catalogue item.

How to verify the installation

1. Build or mark a representative cubicle with the specified door, partition height and ceiling geometry.
2. Record sensor events and light state during entry, door closure, five to fifteen minutes of seated stillness, normal hand movement, standing and exit.
3. Repeat the sequence with different users, clothing, mobility patterns and cubicle positions, including the accessible compartment.
4. Measure false-off events, false-on events, detection delay, recovery time and the effect of the programmed timeout.
5. Repeat the test after final accessories and ventilation grilles are installed, because they can alter sight lines and sensor mounting.

A project can then state a performance requirement: no uncommanded darkness during the defined occupied sequence, a maximum reactivation delay, and a documented manual fallback. This is more auditable than the instruction “provide an occupancy sensor.”

✦ ArchUp Editorial Insight

Energy codes and operating budgets reward sensors that switch lights off quickly, while procurement rewards a device that covers a declared area from a standard ceiling point. Those incentives shape the decision before a user enters the room. The consultant selects from a coverage diagram, the contractor proves operation by walking beneath the sensor, and the facilities team later lengthens the timeout when complaints arrive. In a toilet cubicle, however, partitions break the field of view and seated movement is small; the occupant experiences the control failure, although they were absent from the commissioning scenario. This is the same governance problem examined in ArchUp’s account of Wi-Fi as an architectural sense: sensing technology becomes a rule about who counts as present. A testable project response would move the requirement from device type to observed performance: maximum false-off events during a defined seated-use protocol, mapped blind zones after cubicles are installed, and a manual or automatic recovery path that preserves privacy. This is an ArchUp hypothesis, not a result of the office studies. It would make the sensor, partition layout, timeout and control zone one acceptance package. The party choosing the cheaper or more aggressive control would then remain connected to the consequence, instead of transferring nuisance, safety risk and repeated maintenance calls to users and operators after handover.

Prepared by ArchUp Research Lab.

Editorial correction — 14 October 2026: Removed the image-pending notice after confirming that the featured image is attached. ArchUp Editorial Committee.

References

Shoaib Azizi, Ramtin Rabiee, Gireesh Nair and Thomas Olofsson, “Effects of Positioning of Multi-Sensor Devices on Occupancy and Indoor Environmental Monitoring in Single-Occupant Offices,” Energies, 2021.

Guy R. Newsham, Henry Xue, Chantal Arsenault, Julio J. Valdes, Greg J. Burns, Elizabeth Scarlett, Steven G. Kruithof and Weiming Shen, “Testing the Accuracy of Low-Cost Data Streams for Determining Single-Person Office Occupancy and Their Use for Energy Reduction of Building Services,” Energy and Buildings, 2017.