This story was originally published in my weekly newsletter on May 5, 2023. To subscribe, visit www.staceyoniot.com/newsletter.
Currently it feels like every element of the smart home is getting me down. This week on the podcast, I admitted that Matter has been a huge disappointment all around, as it’s making life not easier but more complicated for folks trying to buy smart home devices. And at home, during my test of the much-anticipated Aqara FP2 presence sensor, I found myself frustrated by the lack of interoperability between what was the most promising device I saw at CES and my existing smart home gear.
I want to give up on Aqara’s $82.99 sensor, but instead I’ll just lay out which aspects of it work and which ones don’t. In short, this device works well in its own ecosystem, but when paired with other smart home ecosystems it performs pretty much like any old motion sensor. Which means that unless you are in deep with Aqara gear, you’re not going to get most of the value this device has to offer.
The FP2 is a dedicated millimeter wave presence detection sensor. To detect bodies in a space, it uses radio frequencies (RF) as opposed to infrared, as most current motion detection devices do. This enables it to offer finer resolution, detect multiple people in a room, and detect gestures, including falls. I’ll talk about fall detection specifically in the second story.
All of which translates into a presence sensor that can theoretically be used for use cases such as detecting where a person is in a room and using their nearness to a door, a bed, or a reading chair to trigger an action. It might also be used to trigger an event (or avoid triggering one) when multiple people are in a space. And for fun, it includes a lux sensor that detects the brightness of light in a room.
So with the right gear and this sensor, you could set up an automation that — as long as no one is within a pre-defined sensor zone, like a bed — opens the blinds at a certain time. Or you could trigger lights to automatically come on in a room if people are in the room and the brightness in that room hits a specific threshold.
This is pretty powerful stuff for automations, and I’d love to be able to take advantage of it. But I can’t. And here’s where the smart home’s lack of interoperability strikes again. The sensor works with Apple’s Home, Amazon Alexa, Google Home, IFTTT, and an open source digital assistant called Alice. But only some of the functionality is carried over to each of these platforms.
I tested the Aqara with Google Home, Amazon Alexa, and IFTTT. Unfortunately, Google Home recognized the device in the Home app, but I couldn’t do anything with it when it came time to try to build a routine. That may be because Google is in the process of beta testing a new version of the Home app with support for more deep sensor capabilities, but it was still frustrating. Although according to Aqara’s website, I should be able to use the multi-zone presence with Google Home, I couldn’t see how. Maybe it will arrive after I/O next week. Notably, Google doesn’t support the light sensing or fall detection.
Next, I tried to use the sensor with Amazon Alexa and got a little further. I was able to set up the sensor in my bedroom (more on the set-up process in a moment) and delineate a zone that represented the bed. After linking the Aqara FP2 with Alexa, I could use presence in both my bedroom and the bed area as triggers.
So I set up a routine to turn off the lamp in the living room when two people were in the bed area. When my husband and I tried it, the lamp turned off within a second or two. It also triggered consistently. That said, neither the brightness sensing capabilities of the sensor nor fall detection are supported by Alexa.
But with IFTTT, I found I could do a lot more, such as use light levels and motion triggers together to trigger an action. I can also set time parameters, so if I want to trigger lights to turn on when a person is detected and light levels are low but avoid doing that in the middle of the night, I can set the trigger to work only between the hours of 6 am and 10 pm. But unlike the other two, IFTTT requires a subscription.
So if these very real and very frustrating limitations don’t bother you, then let’s talk about the process of setting the sensor up, how it performs, as well as some of its other limitations.
Aqara is a Chinese company, and there are some awkwardly phrased elements in the app, but nothing that an English-speaking user wouldn’t be able understand. I also had to go into my profile to change the time zone to reflect daylight savings time, as it was turned off by default. Getting the time zone right matters if you use the app to try to set those time barriers.
Before installing the sensor, know that if you want to use it for presence sensing and multi-zone detection you won’t be able to use it for fall detection. And you’ll either need to install it on a wall or set it up on a stable surface near the edge of a room. From there, it will start sensing presence right away.
You can also go into the app to set up different areas of a room as special zones. Creating the zones is easy. Simply enter editing mode and then start using your finger to color in squares that will delineate the zone. To create my zone, I walked around my bed, noting where the person-shaped icon in the room was.
The sensor needs to be installed about 6 feet off the floor for best results. Once installed it can detect presence in a range of about 20 feet from the sensor edge, which is pretty significant. And you can tailor both the room depth and sensor orientation in the app. I sometimes got an artifact of two people icons when using the app, like there were two people in the room instead of one, but it didn’t seem to trigger automations that required two people in the room. So it was likely a glitch.
Another glitch was the ability to reliably detect presence to trigger automations in different rooms. In my bedroom it did well detecting people, but when I moved it downstairs to the larger family room, it suffered a bit with reliability. Sometimes I questioned whether the app had simply frozen.
I did find the zones to be pretty sophisticated, and the ones I created with Alexa as a way to trigger events worked without a hitch. Lights turned on when I was within about a foot of the outlined space and turned off less than a minute after I left. I had to create two separate automations for that with Alexa.
Sometimes my dog set off the presence detection, and sometimes she did not, which was a little frustrating. The Aqara website does say that it will release a pet filtering option via an over-the-air update, but it’s not clear when that might happen. My dog is a fluffy 20-pound mutt that stands about 16 inches tall, so it’s not as if she’s anywhere close to human-sized when tagged with a radar.
Other issues to note include the fact that this device needs power, so it must be plugged in. That isn’t a big deal if the sensor is sitting on a shelf or mounted on a wall, but if you want to use it to detect falls, it needs to be mounted on the ceiling, which does lead to a less attractive setup. Someone asked me if they could point the FP2 outside in order to detect wildlife, but the mmWave signals don’t pass through glass, so that’s not an option. Aqara says the FP2 is safe for use in humid environments, like bathrooms and under covered porches, but I didn’t test that claim.
Overall, my testing showed a device that has a lot of potential. But for now, that potential is limited to those with Aqara gear in their homes or those with a paid IFTTT account. You can get some of the value associated with delineating certain zones using Alexa, but it seems like a lot to pay for that privilege.
This story was updated on May 9 to correct an error. The FP2 is not the first dedicated mmWave sensor.
Any info on how the sensor integrates with HomeKit or Home Assistant? Disappointing to hear how integration has fallen short so far, hoping either of those platforms might expose richer automation capabilities.
The always thorough HomeKit News site did a deep dive review of this device last month, including the exact differences between what you can do in the aqara app and what you can do in the HomeKit home app. It should answer most questions.
https://homekitnews.com/2023/04/30/aqara-presence-sensor-fp2-review/
For home assistant, you have to bring it in through the HomeKit controller option, so it has the same features that HomeKit has. (Derek Seaman’s Tech Blog had a detailed article on using it with home assistant last week.)
Short answer: in homekit, you can get multiple zones, multiple people, as an occupancy sensor. And also a separate lux sensor. You don’t get the fancier stuff like being able to tell the difference between someone entering a zone and leaving a zone: those currently require setup in the aqara app.
The FP2 is not the first-ever presence sensor by a mile. Even Aqara has the FP1 that came out a year prior to this, and both Yeelight and LifeSmart both have presence sensors that precede the FP2.
Based on the screenshot, it looks like you didn’t set up the edges of the room, which is necessary for more accurate detection.
Finally, the FP2 can detection presence/motion through glass. This is one of the features that sets it apart from traditional motion sensors.
It might also be helpful to mention some of the base technology differences between mmWave and standard PIR (passive infrared) sensors.
PIR
Because it uses passive infrared, it will not detect through glass or walls. The good news is that makes it very easy to limit detection to a single room.
With the right power management, initial detection will be very quick, typically one to two seconds.
PIR is really a heat detector, so it doesn’t detect shape or direction.
Detects relatively large movements, like a person walking. Does not detect “fine” movements like typing or eating from a bowl or “very fine” movements like breathing.
Has difficulty detecting multiple changes in a short period of time. A typical “cooldown” period is 3 minutes after the initial detection during which nothing is reported.
Not typically good at people counting.
Easy to find models under $50, and some under $30.
Easy to find batterypowered models.
Setup is typically easy if you use the default parameters.
MMWAVE
Can detect through glass and thin walls. The good news is that makes it an excellent choice for bathrooms with shower stalls. But also means you have to manually define the “edges” of the room in an app.
Initial detection can be slow, as much as 6 seconds, while it collects data and goes through its algorithms to determine what it has detected. This may make it unsuitable for turning on lights quickly, such as when a closet door is opened.
Can detect shape, direction, and velocity.
Great at detecting both “fine” and “very fine” movements such as breathing. This tech has been used in prisons and hospitals for over 5 years to monitor people sleeping while allowing for privacy.
Very good at detecting multiple changes in a short period of time, although most inexpensive devices still use a 1 minute minimum “cooldown” period.
Using the app, can typically divide an area into as many as 30 zones, allowing for different rules for each.
Excellent for people counting.
Historically these devices cost between $300 and $600 each from companies like Vaymar, AltumView, and TI (although those were sometimes combined with a PIR motion sensor to allow for a quicker initial detection to turn lights on, and some models included cameras which were turned on after large motion or falls were detected). In the last 18 months Aqara, Tuya and LifeSmart have all introduced mmWave models under $150 for home use, with some models under $100.
All the ones I’ve seen have required mains power.
Setup can be quite complex as you manually define edges, zones, and detection parameters.
And although the mmWave technology is certainly capable of things like distinguishing between a dog and a person and tracking multiple people within a zone, not all of the apps include those algorithms.
TI had an excellent video in 2019 showing how mmWave sensors “view” a room and track multiple people.
https://youtu.be/njhRwijx_HY
COMBINATION
A simple rule of thumb is that PIR sensors are typically better at turning lights on or intruder detection (noticeably quicker initial trigger time and more natural “room” definition) and mmWave sensors are typically better at deciding when to turn lights off, people counting, and reporting on specific kinds of movement such as falls.
PIR sensors are also typically much simpler devices with lower power needs, so many batterypowered models are available.
So for an intruder alert security system, PIR sensors may not only be cheaper, but better.
For fall detection, detecting movement through a glass shower stall, or knowing someone is still in the room even if they are sitting still or sleeping, mmWave is definitely better.
For automating lights, a combination may work best, especially if you want the lights to come on quickly.
Why does this article begin with the pitfalls and disappointments of Matter and proceed to talk about a sensor that is NOT presently using Matter?
All my Eve Matter devices are working beautifully with Homekit and Alexa (with recent firmware and app updates).
Why not review the performance in Homekit where it arguable supports the most types of events?
Stacey mentioned on the podcast that at her house the FP2 was unable to detect through windows, but also that she has some kind of high tech windows. That’s a very good point.
If your windows have a metallic coating, which is true of most “thermal windows,” “low – E windows,” and UV-reducing windows, the MMWave sensors probably can’t detect through those.
However, that kind of material is not typically used for glass shower stalls or bathtub enclosures, so, again, detecting that a person is in the shower in order to keep the lights in the bathroom on (or to detect falls) is likely to be a lot easier with an MMwave sensor than a PIR sensor. Still, there are some shower enclosures made with low-e glass and some made with “low iron” glass and either of those might also block signals. So it’s just something to be aware of.
Also, stained glass can contain a lot of different materials, many of which are metals, and that can also block signal.
So as with pretty much anything involving radio frequencies, the details matter. Knowing that MMwave technology is able to detect through clear glass and PIR technology is not doesn’t mean that a particular MMwave sensor will be able to detect through a specific window. Sometimes you just have to try it and see.