The Still Person Problem: Why the 2026 Smart Home Senses Presence With Radar, Not Motion

The sensor that gives up when you hold still

Sit at a desk and read for twenty minutes, and a conventional motion sensor decides the room is empty. The light clicks off over your keyboard, and you wave an arm to bring it back. That small daily annoyance is the whole reason presence detection is being rebuilt in 2026, and the fix is not a better motion sensor but a different physics entirely. The passive infrared sensor that has run home automation for three decades is being displaced by millimetre-wave radar, and the change turns the word occupied from a report about movement into a report about the body itself.

From a heat gradient to a radar chirp

A passive infrared, or PIR, sensor is exactly what its name says: passive. It emits nothing. Behind its faceted white lens sit pyroelectric elements that generate a small voltage only when a warm body crosses from one lens segment to the next, sweeping a moving heat gradient across the field. That design is cheap, draws almost no power, and reacts in a fraction of a second, which is why it still belongs in a hallway or on a porch. But it can only see change. A person who stops moving stops producing the gradient, and to the sensor the room has emptied.

Millimetre-wave radar works the opposite way round. It is active: it transmits a faint radio signal in the 24 or 60 gigahertz band and listens to the echo. The technique is FMCW, frequency-modulated continuous wave, meaning the transmitter sends a rising tone, a chirp, whose frequency climbs steadily over each sweep. Because the frequency at any instant is known, the delay of the returning echo shows up as a frequency difference, and that difference converts directly into distance. Send thousands of chirps a second and compare them, and the tiny phase shifts between successive echoes reveal motion far below what any lens could register.

Why a rising chest counts as movement

This is the mechanism that matters. The Doppler phase comparison between successive chirps is sensitive to sub-millimetre displacement, so the rise and fall of a chest during breathing, and even the faint pulse of a heartbeat, register as genuine motion. The radar does not need you to walk across a room; it needs you to be alive. A person reading, sleeping, or watching a film keeps the room firmly occupied because their breathing never stops. The false clear that plagues PIR simply disappears, and automations can hold a light, a heater, or a sleep scene for as long as a body remains, not merely as long as it fidgets.

That single capability rewrites the logic of a smart room. A bedroom can hold a sleep mode all night because the bed reads as occupied by breathing alone. A bathroom fan can run until the room is truly empty rather than timing out on a preset guess. Climate control can condition only the rooms that hold people. And because radar carries no camera and forms no image, it reports presence and position without identifying anyone, which is the privacy line that lets it belong in a bedroom at all.

Range or resolution: the two bands

Not all of these sensors are the same, and the frequency printed on the box decides what a room gets. The 24 gigahertz band penetrates further and reaches across a large open plan, up to roughly nine metres, but its angular resolution is coarse, so it is happiest reporting a simple occupied-or-empty state and mapping a room into broad zones. The 60 gigahertz band trades range for a wider bandwidth and finer resolution, which is what makes precise tricks possible: posture and fall detection, and the breathing and heartbeat tracking used in sleep monitoring. Choose the band by the job. A living room that only needs to know someone is on the sofa wants 24 gigahertz; a bedroom meant to track sleep wants 60.

The catch: a beam that sees through walls

Radar solves the false clear but introduces a problem of its own, and it is a design problem, not a wiring one. The same signal that reads a chest through a blanket also passes through a thin partition, so a sensor aimed carelessly will faithfully report the neighbour in the next room or a person walking the hallway, and switch a light on for no one present. Moving fan blades and billowing curtains read as motion too. A radar sensor is therefore not aimed the way a PIR is; it is bounded in software.

Designing the room around the beam

This is where the smart home becomes an interior-design decision. Good practice is to mount the sensor high, facing down from the ceiling or at around two and a half metres, point it toward an exterior wall rather than a shared interior one, and then draw exclusion and detection zones in software so the beam watches the sofa, the bed, or the desk and ignores the doorway and the window. A short confirmation delay stops a passing shadow from triggering a scene, while a long vacancy delay suits rooms where people sit still. One well-zoned radar can count occupants and cover the work of two or three motion sensors, but only once someone has decided which part of the room should count as inside.

What actually changes

The buried limit of the smart home was never movement; it was the assumption that a still person is an absent one. Millimetre-wave radar retires that assumption by listening for the body rather than watching for the walk, and in doing so it moves the design work off the ceiling and into the floor plan, where the real question is no longer where to put the sensor but where, precisely, the room begins.