Apple Explains How Crash Detection Was Created in New Interview
Apple’s Vice President of Sensing & Connectivity Ron Huang and Vice President of Worldwide iPhone Product Marketing Kaiann Drance sat down for an interview with TechCrunch last week where the pair delved into the inner workings of Crash Detection.
Crash Detection is a new safety feature available on iPhone 14, iPhone 14 Pro, Apple Watch Series 8, and Apple Watch Ultra. It is designed to detect when the user has been in a serious car crash and automatically contact emergency services and contacts.
Crash Detection is made possible by a slew of different sensors and data points. Chief among these is an improved 3-axis gyroscope and an all-new high g-force accelerometer on iPhone 14 models and the latest Apple Watches.
The feature also receives input from the GPS to determine when the user is travelling at high speeds, the microphone for sounds related to a crash, a barometric pressure sensor for changes in air pressure associated with an airbag going off, and Bluetooth and CarPlay to determine when the user is in a car.
However, not all of these data points have to indicate a crash for the feature to be triggered. Huang and Drance said Crash Detection is powered by a “dynamic algorithm,” so there’s no one set rule governing how many data points need to be collected for the feature to be triggered.
“There’s no silver bullet, in terms of activating crash detection,” Huang told TechCrunch. “It’s hard to say how many of these things have to trigger, because it’s not a straight equation. Depending how fast the travelling speed was earlier, determines what signals we have to see later on, as well. Your speed change, combined with the impact force, combined with the pressure change, combined with the sound level, it’s all a pretty dynamic algorithm.”
The feature compares notes from each of the systems to determine whether or not there has actually been a crash. Crash Detection is configured this way to increase precision and prevent it from going off in a minor accident or, for example, when the user drops their iPhone or Apple Watch while driving.
“I actually had a rear-end fender bender when I was in New York earlier,” Drance said. “My crash detection did not go off, because it’s just one of those minor things where you just get out of your car and keep going. That’s part of the sensor fusion and accuracy, because we don’t want to be doing a lot of false calls to 9-1-1 when they’re not necessary.”
Unfortunately, this is where Apple’s accounting of Crash Detection trails off from real-world performance. As we reported on Sunday, many iPhone 14 and Apple Watch users are reporting false positives with Crash Detection, to the point where it can even be triggered by rollercoaster rides.
Huang and Drance said a lot of work went into tuning Crash Detection. Apple worked with several crash labs and performed real-world testing with more than “dozens” of car crashes and “thousands” of its own devices to create a dataset representing serious accidents.
The company also used data from the Department of Transportation and the NHTSA (National Highway Traffic Safety Administration). Even so, it looks like Apple has more work to do on Crash Detection if the feature is being triggered by rollercoaster rides and phones falling out of moving vehicles.
The two Apple execs went on to note that Crash Detection can also take advantage of the company’s new emergency SOS via satellite feature to hail emergency services if it is triggered in an area without network coverage. You can check out the full interview on TechCrunch.