As one of the latest and most exciting safety features offered on Google Pixel Watch, fall detection can be a useful safeguard in case of emergency. Launched in February¹, this feature detects when users have taken a hard fall and haven’t moved for around 30 seconds. The device will then vibrate, sound an alarm and send a notification asking the individual if they are okay. If the user does not move or respond to the notification after about a minute, the watch will automatically call emergency services and share your location².

How exactly does Pixel Watch determine if users have fallen down a flight of stairs and need help, or if they are just chasing their grandchildren in the yard? From motion sensors and AI algorithms to stunt doubles, here’s a look at how fall detection works and how the team at Google built it to be as accurate as possible.

Pixel Watch is full of sensors that collect data to inform all the helpful features, such as the sensor that tracks your heart rate to determine your sleep quality. For fall detection, Pixel Watch uses motion sensors to look for the body’s instinctive reactions to a fall, such as moving your arms on the way down, and the impact force of a fall.

But there’s a catch: a number of the movements people make each day have similar motion patterns to a serious fall. For example, you might take a small stumble while walking with a friend or move your arms differently than normal while taking a swim — neither of which constitutes an emergency. In order to maximize efficacy and minimize false alerts, the fall detection team had to get creative with their testing process.

To be as helpful as possible and minimize false detections, the Pixel Watch engineering team created AI algorithms to differentiate between a hard fall where users might need emergency assistance and a movement that only mimics falling but isn’t serious. Step one in creating those algorithms: collecting the data.

Early on, they partnered with labs that were already researching what falls look like to motion sensors. Multiple factors can impact how sensors interpret falls. For example, if someone takes a fall on a crash pad, it doesn't look just like a real-life fall because the impact is less intense. To capture a more realistic fall, the Pixel Watch team turned to computer vision for help visualizing a variety of tumbles. First, they worked with another research team at Google to analyze videos of real-life falls. Then, they simulated those physics while also accounting for factors that may change what a fall looks like — from different limb lengths and body types to what surfaces people may land onto.

After assembling all this data and incorporating it into the fall detection algorithms, the AI needed to be tested in real-life scenarios. The team set out to Hollywood to put their work to the test with a group of stunt doubles, directing them to take different types of falls while wearing Pixel Watches.

Once the team was confident the fall detection feature worked well, they tested it on several hundred people for over a year to ensure it didn’t inundate emergency services with false alarms. This group of testers wore the watch while doing high-energy activities — like jumping, swimming and burpees — to ensure those movements didn’t trigger alerts. The emergency dispatch community also helped test that the watch’s automatic calls were as clear and helpful as possible.

Falls, big or small, are never planned and can sometimes leave individuals unable to help themselves. Fall detection can bring you and your family peace of mind, knowing no matter when or where a fall happens, help can be on the way. Fall detection is only one of the many ways Pixel Watch can offer safety. From the Emergency SOS feature that users can enable when they feel unsafe to alert emergency services or a trusted contact to Atrial Fibrillation (AFib) detection, Pixel Watch and Google are here to support users.

About the authors: Paras Unadkat is a Fitbit Product Manager working on wearable health sensing AI and the future of wearable fitness. Edward Shi is a Product Manager at Google, specializing in personal safety for Android and Pixel. Boyan Bonev is an engineer specializing in machine learning, personal safety and wearable devices. 

1. Fall detection is not available in all countries and is dependent upon network connectivity and other factors. Your watch may not be reliable for emergency communications. Fall detection may not detect all falls. To call emergency services on a Google Pixel Watch without 4G LTE, your paired phone needs to be nearby. See g.co/pixelwatch/personalsafety for more details.

2. Users must grant location permission for fall detection to share location externally.