Aircon Motion Sensor: What It Detects and What It Misses

Two different devices are sold under one word, and they do not know the same things. One reports movement in front of the unit; the other measures temperature across the room and works out where a body is sitting. Which one is fitted decides everything the feature can do afterwards, and the name on the brochure is no help because both are sold as a sensor that knows whether you are there.

Daikin's operation manual states its position plainly: intelligent eye is described as the infrared sensor which detects the human movement. The same manual divides the room into two areas so the unit knows which side moved. Movement is the entire input. The unit never learns that a person is present, only that something changed in area one or area two, and it counts the minutes since the last change.

Mitsubishi Electric documents a different instrument. Its manual for a ceiling cassette states that the 3D i-see Sensor operates once approximately every 3 minutes to measure the floor temperature and detect persons in the room. That is a scan on a timer, not a trip wire: the unit reads a thermal picture of the floor and picks bodies out of it. That is how the same manual can discuss where an occupant sits and how many people the room holds.

The parts behind those two descriptions work on opposite principles, and one component maker publishes both. Panasonic Industrial Devices states that pyroelectric sensors react to a change in intensity of infrared radiation. For its Grid-EYE thermopile array, the same company puts it another way: that sensor can detect temperature distribution in a two dimensional area, identifying the number and position of people from body temperature. Change against picture is the whole difference.

Four of the makers quoted here admit the same failure in their own manuals: a person who stops moving can read as an empty room. Fujitsu General goes furthest and prints it as a warning instead of a note: do not use this function when there are only infants, children, elderly, or sick persons in the room. That is a manufacturer naming the households its own feature is wrong for.

Daikin puts the warning next to the instructions. The manual states that the air conditioner may go into energy-saving operation even if there are people in the areas. It then names a reason: this may occur depending on the clothes the people are wearing if there are no movements of the people in the areas. Two conditions sit inside that sentence, what you are wearing and whether you are moving, and neither is unusual in a room where somebody is reading or working at a laptop.

Mitsubishi Electric lists the situations in which a person may not be detected. A heat source that does not move for a long time is on that list, as is a person wearing heavy clothes whose skin is not exposed. A sensor that measures floor temperature ought in principle to see a motionless body, and the maker still names prolonged stillness as a miss. Paying for the more capable instrument narrows the problem without removing it.

One condition on those lists belongs to this climate. Fujitsu General puts a number on it: a room at 30 °C or higher in summer is a case where the sensor may find nobody when somebody is there. Mitsubishi Electric lists a high room temperature the same way, and a Singapore room that has not been cooled yet routinely sits at that number or above. Panasonic Industrial Devices explains why from the component side: during summer, when the difference between the surface temperature of a human body and the background is small, sensitivity decreases.

A clock does the work the sensor cannot. Daikin acts once no movement has been detected for more than 20 minutes, and Fujitsu General publishes the same wait, while Hitachi shifts the setting at 20 minutes and again at 60. Mitsubishi Electric waits 60 minutes before its energy-saving step and allows a separate shutdown timer set anywhere between 60 and 180 minutes. Those figures are not measurements of your household; they are how long each maker waits.

What happens after the wait is where most of the industry agrees and one design does not. Daikin, Fujitsu General, Hitachi and Panasonic all respond by moving the target temperature, and none of those manuals describes the unit switching itself off. Mitsubishi Electric shifts the target, aims the vanes toward or away from an occupant, counts how full the room is, and will stop the machine outright; most designs nudge a number, while one takes decisions.

Daikin's manual describes the louvers adjusting so that leftward or rightward airflow will not be directed at the people, and Mitsubishi Electric turns the same idea into a choice. Indirect means an occupant is less exposed to direct airflow, and Direct means mainly the vicinity of an occupant is cooled. An Even setting learns where an occupant spends most of the time. Once the vane moves the stream is an ordinary cold jet, with the coanda effect deciding how far it runs along the ceiling.

A bedroom gives this feature the least to work with. Stillness is the point of the room. It is also the condition these manuals list as a reason a person goes undetected. The machine must then separate an empty bed from an occupied one using the signal a sleeping body is least able to produce, and every other room in a flat is an easier problem.

Bedding removes another margin. Mitsubishi Electric lists a person wearing heavy clothes whose skin is not exposed among the cases its sensor may miss, and a blanket presents the same problem. The same manual names positions the sensor cannot cover: along the wall it is installed on, directly under it, and anywhere an obstacle stands between person and unit. Bedrooms in flats here put the bed in one of the first two more often than not, and a wardrobe supplies the third.

On some units the feature also takes something away. Daikin's Singapore manual for one wall-mounted range states that night set mode will not go on during use of intelligent eye operation. A household already running the night setting is choosing between two features rather than adding one, and the manual is the only place that says so.

This is where the feature arrives as a complaint instead of a benefit. Cooling that fades in the small hours, or a unit found off in the morning while the remote still shows it running, both have an occupancy explanation to rule out before anyone talks about gas. Mitsubishi Electric states the second case outright: even when the unit has stopped on its no-occupancy timer, the remote display still indicates the unit is in operation.

The saving belongs to your household, not the machine: it comes entirely from the share of running hours in which the room is genuinely empty. A family that switches the aircon on when somebody walks in and off when they leave has already collected that saving by hand. The households where the share is large are specific ones: a living room unit left running through an afternoon while people move in and out. So is a study that empties for errands and fills again after dinner, or a unit switched on before anybody is home.

The failure also runs in the other direction, and Daikin prints the list: its manual states that the sensor also mistakenly detects pets, sunlight, fluttering curtains and light reflected off of mirrors as passersby. Fujitsu General publishes a matching list and adds an oscillating fan to it. A room with a cat, an afternoon sun patch, or a curtain that lifts under the airflow may never register an absence at all, so the feature costs nothing and does nothing.

No saving percentage appears on this page because none of these documents carries one; the manuals state what the machine will do and stop there. The energy label will not settle it either, since NEA sets the tick rating from a weighted COP built from full-load and half-load measurements in a laboratory. A feature whose value depends on when your rooms empty falls outside what that test can report. What any of it takes off a bill depends on hours only your household can count.

• •The model number, not the brand name. Which instrument is fitted varies from one model line to the next, and changes again between production years.
• •What the manual says the sensor detects. Movement and floor temperature are different claims, and the feature name rarely settles which of the two is fitted.
• •What the unit does once it decides the room is empty, and after how long. A setpoint shift and an automatic shutdown are different things to live with.
• •Whether it can be switched off, and whether turning it on blocks a setting the household already uses.
• •Where the sensor sits on the indoor unit, and what stands between it and the seats people actually use.
• •Whether that room is ever occupied and then empty inside the same running session. If it never is, the feature has no work to do.