Boeing engineers designed the 747 here in the Puget Sound area in the late 1960s. They ran the aerodynamic and structural math on room-sized mainframes that cost millions and needed their own climate-controlled buildings.
The watch on your wrist outruns those machines by orders of magnitude. It also carries something they never had: sensors that measure the physical world directly, all day, from your body.
I build mobile apps on those sensors at LINC Innovations. This post covers two things: 1) what published research says the hardware can actually do, and 2) where sensor projects actually fail. It's almost never the hardware.
What the research says the hardware can do
Consumer phones and watches carry MEMS inertial measurement units (IMUs): a three-axis accelerometer plus a three-axis gyroscope, typically sampled at 50 to 100Hz. A 2023 review in Biomimetic Intelligence and Robotics tracks how far activity monitoring and motion control on these parts has come (Wang et al.).
The clinical validation work is the striking part.
A 2023 study in IEEE Journal of Translational Engineering in Health and Medicine used a wearable accelerometer and gyroscope to estimate essential tremor severity against the Fahn-Tolosa-Marin rating scale. That's the same scale clinicians score by eye.
A 2024 scoping review in Frontiers in Digital Health looked at resting heart rate measured by pressing a fingertip against a phone camera. Agreement with ECG across the reviewed studies: correlations from r = 0.98 to 1 (Mather et al.).
A 2025 study in Applied Sciences went further and estimated heart rate from a six-axis IMU alone, no optical sensor involved, by detecting the small vibrations each heartbeat sends through the body.
Add the rest of the parts bin (microphone, barometer, magnetometer, LiDAR on the Pro iPhones) and a stock consumer device is a measurement platform researchers validate against clinical equipment. No custom hardware involved.
Where sensor projects actually fail
The gap between a lab demo and a product people rely on is the whole game. Three places it goes wrong.
Time sync. Fusing two sensor streams only works if their clocks agree. My own example: SniperPulse syncs a 100Hz watch IMU with 240fps phone video. That took me three attempts. The third one worked.
Battery. Continuous sampling is a power-budget decision, and a 2024 IEEE Access paper on real-time sensor libraries for Wear OS makes the point well: how you collect the data matters as much as what you collect. Users don't file bug reports about battery drain. They uninstall.
Device fragmentation. The same code behaves differently across vendors. Samsung's battery management will silence a background sensor stream that runs fine on a Pixel (dontkillmyapp.com documents this per vendor). The only answer I trust is physical hardware: every release of my apps runs on a bench of real phones and watches before it ships.
Receipts
I've shipped 9 Kotlin Multiplatform apps to both stores since 2024, most of them sensor apps. They share sensor and math code across iOS and Android from a single codebase.
A few:
- SniperPulse reads a watch IMU during rifle dry-fire training and scores your hold in the seconds before the shot. I compete in NRL22 and built it for my own training.
- PrecisionLevel turns the phone IMU into a spirit level.
- DawnCup tracks workout intensity on Wear OS.
- LoomHaptic, the newest project, classifies environmental sound on-device and turns it into wrist haptics for deaf and hard-of-hearing users. The audio never leaves the device.
The gap is the opportunity
The hardware is ready and the research checks out. What separates apps people rely on from demos is the unglamorous part: clocks, batteries, and the long tail of real devices.
That part is measurable. I spend most of my weeks measuring it.
References
- Wang, X., Yu, H., Kold, S., Rahbek, O., Bai, S. "Wearable sensors for activity monitoring and motion control: A review." Biomimetic Intelligence and Robotics 3(1), 2023. DOI: 10.1016/j.birob.2023.100089
- "Wearable Accelerometer and Gyroscope Sensors for Estimating the Severity of Essential Tremor." IEEE Journal of Translational Engineering in Health and Medicine 12:194-203, 2023. DOI: 10.1109/JTEHM.2023.3329344
- Mather, J.D., Hayes, L.D., Mair, J.L., Sculthorpe, N.F. "Validity of resting heart rate derived from contact-based smartphone photoplethysmography compared with electrocardiography: a scoping review and checklist for optimal acquisition and reporting." Frontiers in Digital Health 6:1326511, 2024. DOI: 10.3389/fdgth.2024.1326511
- "Research on Improving the Accuracy of Wearable Heart Rate Measurement Based on a Six-Axis Sensing Device Integrating a Three-Axis Accelerometer and a Three-Axis Gyroscope." Applied Sciences 15(14):7659, 2025. DOI: 10.3390/app15147659
- "Real-Time Energy-Efficient Sensor Libraries for Wearable Devices." IEEE Access, 2024. ieeexplore.ieee.org/document/10600468