This story was originally published in my weekly newsletter on May 5, 2023. To subscribe, visit www.staceyoniot.com/newsletter.
This week, I was testing the Aqara FP2 millimeter wave sensor. The device uses mmWave to detect people and their actions in a predefined space, and can trigger automations or send text-based alerts. In the first story of this week’s newsletter, I focused on how it performed as a presence sensor. Because it uses mmWave technology, it can detect how many people are in a room and where they are in said room, which enables home automation nerds to create some very specific automations. (Here is the full review.)
But for this story, I wanted to evaluate the device as a fall detection monitor. It’s important to note that if you use it for that purpose neither the multi-person presence detection nor the ability to create specific zones will work; it will only detect falls. I also wanted to explore whether or not a ceiling-mounted sensor or a wrist-worn fall detection sensor, like an Apple Watch, makes more sense.
Installing the FP2 sensor as a fall detection device requires it to be mounted on the ceiling, at the edge of a room. The fall detection works best for low ceilings, specifically between eight and nine feet. The sensor also only measures falls in a radius of about six-and-a-half feet out. This, plus its ability to withstand humid environments, makes it a good option for a bathroom, but a poor option for a living room with a cathedral ceiling.
I tested it in our downstairs family room, which has eight-foot-high ceilings but is a bit larger than the recommended area size. After I placed the sensor in the room and calibrated it by moving to the rightmost and leftmost edges of the room, I also set up the app to alert me via notification when a fall was detected. And then the fun began.
Aqara specifically says you shouldn’t use the FP2 as a medical device, and after testing it I tend to agree. Possibly because I have an active family, with a 16-year-old and a dog, and the living room is also where we all work out, I ended up with a lot of fall notifications. Anytime my kid or I dropped quickly to our knees to play tug with the dog, I got a notification. Whenever someone plopped down quickly to do sit-ups, I got a notification.
Slower movements didn’t trigger notifications, nor did squats. But my fake trips and falls did. So it did a good job of letting me know when someone dropped rapidly down to the ground. But in my home that doesn’t necessarily mean a fall. It’s likely that with someone older, or someone frail, there would be fewer false positives. But then the question becomes whether a wearable makes more sense.
At $83 per sensor, and the fact that one might need several of them to cover an entire home, this sort of device could quickly get expensive. The $250 Apple Watch SE can detect falls and call actual emergency services when a fall is detected if the person also has an iPhone. And it doesn’t just work in rooms with low ceilings.
But the tradeoff with a wearable is battery life. If a person doesn’t charge their Apple Watch, it can’t detect falls. And at night, when someone is most likely to charge their wearable, is probably when they are most likely to fall. The other time is in the shower. And if the wearable is sitting on the bathroom counter, any fall will go undetected.
Perhaps a layered approach is best, or one in which the ceiling-mounted sensors are placed in bathrooms or bedrooms, where falls are more likely to happen when the wearable isn’t on. Other than on stairs, those are the places where falls tend to happen. And some people might not want a wearable. Some might not have an iPhone, which leaves them with few other options that don’t look like medical equipment.
So unfortunately I can’t say with certainty that the Aqara is the best option for fall detection. But it is an option. And if the shape and height of the rooms you want to monitor are within its sensing parameters and the price is right, it might be your best option.
Interesting article! As is my usual philosophy, I believe choice is good: I think different things will work for different households. Or sometimes just different use cases. @Ben mentioned in our previous discussion on fall detection on this site that his wife is a nurse in the dementia ward of an aged care facility, and said wearables, just don’t work in that environment. He was responding to comment that I made where I mentioned that I am a wheelchair user who does my own transfers, so falls are a major issue for me, and that I have personally found a wearable most practical for my purposes.
Your tests were interesting, and I’m sure the issue of false alarms gets complicated in your setup. I also have a big dog, and I imagine him jumping up and down off the couch Might also set off an area alarm.
On the battery life issue… As I mentioned previously, because the Apple Watch is so important to me for both monitoring and communication, I do have two of them and swap them out at bedtime. Several of my other friends who have mobility issues do the same thing. Yes, it doubles the cost: but if you can afford it and you need 24 hour fall monitoring and some of the other features, it can be a good solution. Certainly mine gives both me and my care team considerable peace of mind.
You mentioned that some types of falls are more common than others. It’s also true, that some locations are more likely than others. You already mentioned the bathroom. Beyond that, particularly for those with mobility challenges, getting in and out of bed is a common place to fall. So the ceiling monitor placed to cover the bed might be a good solution for those folks. And then there’s the sidewalk right outside your house. Particularly the path to and from the mailbox. This is obviously where wearables shine: they go where you go, and those that have a built-in communication option add even more to this particular use case.
So I think we will continue to see improvement in solutions for fall detection, and I expect we will also continue to see multiple device classes offered.
A common “fall” type for elderly people is to slide off or sideways from a sat position. It’s less likely to cause direct injury, but still leaves the person stranded.
How well does this sensor pick those out?
As someone who uses a wheelchair and does my own transfers, I’m very familiar with the slip down to the floor kind of fall. 🙁
SPEED MATTERS
Part of the fall detection algorithm is the velocity of the movement. Stacey says in her article that squats and slower movements did not trigger the fall detection on the FP2, and I think that’s typical.
In my case, if I’m sliding off the bed down to the floor and trying to stop myself, it’s likely that my movements would be too slow to trigger fall detection. If I manage to transfer off the bed far enough to get my butt off the bed but miss the wheelchair and then fall down quickly, that probably would.
Purpose-built fall detection systems deal with this issue in a couple of different ways.
Some can also be triggered by a wearable button. So the person can press the button after they fall and then initiate all the usual fall detection response actions from the system. That would be useful for someone like me who may, as you noted, fall without much impact, but then not be able to get up again. (Although in my case, I would get both fall detection and contact on request functionality from my Apple Watch.)
There are some sophisticated fall detection systems for institutional settings which can alert if someone is lying down in a specific detection zone (commonly on the floor by the bed) for more than a set period of time. That’s intended to catch the kind of falls that you describe while not alerting if someone is just getting a box out from under the bed.
The aqara device has promised this type of functionality (which they call posture detection) in a future firmware update, but the device doesn’t have it now, and who knows if it will ever be actually delivered. But even if posture detection functionality does arrive in the future, you’ll still have the same issues of Interoperability that Stacey mentioned. It’s likely you’ll be able to use it in the aqara app, but not integrated with other systems.
“ALERT BY PROXY” FOR PARTIAL INTEGRATION
For some use cases you can do an “alert by proxy” setup where you have the aqara app turn on an aqara smart plug that you don’t use for anything else, and then have that plug coming on trigger events in other integrated systems (because the plug is visible in both). You wouldn’t get as much information as you get in the aqara app itself, but it can still be a useful partial integration. This would definitely work with HomeKit, Alexa routines, or IFTTT.
————-
TL;dr : the Aqara FP2 does not currently detect slow changes in position, like sliding down to the floor. The company has said that it will offer “posture detection” in a future update in its own app, which will let you distinguish between lying or sitting on the floor and standing in the same area. You might be able to use that if it actually gets delivered as promised, but as Stacey noted, you’d still have very limited integration with other apps.
Correction: I said you could use the “alert by proxy” method with Alexa routines, and you can, but not with an aqara smart plug, because at the time of this writing, those devices can’t be used as triggers in an Alexa routine. My bad. It would definitely work with HomeKit or Ifttt, though.
My apologies for any confusion.