Original title: Sensor-based human activity recognition using wrist and ankle wearables
Ojan Majidzadeh Gorjani, René Jaroš, Petr Bilik · September 2026
Smart home systems need reliable ways to track what a person is doing to offer real-time help. Researchers tested placing wearable motion sensors on both the right ankle and wrist. Each device tracked movement, direction, and rotation 66.6 times every second across nine daily activities. Combining data from both the upper and lower body gives the system a complete picture of motion. In tests, the best computer models correctly recognized actions 93.48% of the time on completely fresh data. The sensors also proved specialized: ankle sensors are best at spotting stair climbing, while wrist sensors catch small hand actions like typing. Adding a simple voting filter smoothed out errors while responding in under 2 seconds to trigger smart home controls.
Imagine trying to guess what someone is doing by only watching their hand movements. You might mistake writing for cooking if you only see the hand. But if you also watch their legs, you can tell if they are standing still or climbing stairs. In this paper, the system uses both wrist and ankle sensors to get a complete picture of the body's movements, making it much better at distinguishing activities.
For people who need smart homes to adapt automatically—like the elderly or those with disabilities—this means a home could adjust lighting, temperature, or alert a caregiver based on what you're actually doing, without you having to press a button or say a word. The system uses sensors you already wear on your wrist and ankle, and it works in real time with less than two seconds of delay.
The study only covers nine specific daily activities, so it may not recognize others. The sensors are placed only on the right ankle and wrist, which might not work for left-handed people or different placements. Data was collected in controlled intervals, not during continuous free-living, so real-world performance might differ. The abstract does not mention testing with diverse populations (age, health conditions).
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Sensor-based human activity recognition using wrist and ankle wearables — Ojan Majidzadeh Gorjani, René Jaroš, Petr Bilik — Frontiers in Bioengineering and Biotechnology — doi:10.3389/fbioe.2026.1926076
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.