Inventions ยท Feature
Many electrical devices unintentionally emit electromagnetic interference during operation. EM-Sense investigates whether these signals can be detected through the body when a device is touched. A watch-format research prototype distinguished between various objects without additional tags on the devices. This is a specialised sensing technique, not a general ability to identify every object touched reliably.
Existing interference becomes a measurement signal
The researchers behind EM-Sense exploited a property of many electrical and electromechanical objects: during operation, they generate small electromagnetic interference signals. When a person touches such an object, components of these signals can travel through the conductive body to a wearable sensor. For this purpose, the team modified a small software-defined radio and developed a demonstrator in the form of a smartwatch. The signals are analysed to assign the object being touched to a previously studied class. The approach differs from labels or radio tags because the target objects do not need to be specially equipped with them. Nevertheless, it is not a universal material-measuring device. It detects characteristic electrical emissions under contact conditions, not directly the shape, colour or complete identity of an arbitrary object.
The experiments show possibilities and limitations
The work by Carnegie Mellon University and Disney Research reports that distinguishing between several dozen objects is possible in principle. The researchers also investigated changes associated with different wearers, times and local environments. Such experiments are important because a signal that stands out in the laboratory can be altered in everyday life by other devices or operating states. An object that is switched on is also not necessarily equivalent to its switched-off state. Similar devices can also generate similar electrical patterns. The documented results therefore demonstrate feasibility within the situations tested, not error-free recognition of all household objects. For a specific application, it would be necessary to establish which object classes are relevant, how the classification is learnt and how the system handles unknown or ambiguous signals.
Contextual information needs a clear purpose
A wearable contact sensor could, for example, bring up operating guidance appropriate to the device currently being used. Such scenarios explain the usefulness of the research principle, but the demonstrator does not automatically confirm them as reliable everyday functions. The decisive question would be whether the recognised information is more accurate or practical than a deliberate selection by the person. There are also questions of data processing: some aspects of everyday activities can be inferred from the devices touched. A future system would therefore have to make clear what is detected and stored. EM-Sense should be regarded as a technical proof of concept, not as an available universal recognition watch. Its distinctive idea is to treat supposedly disruptive side effects of existing electronics as a source of information, and to make human contact itself the transmission path for the sensor signal.
Sources & further reading
- Disney Research: EM-Sense, research paperla.disneyresearch.com
- Chris Harrison, Carnegie Mellon University: EM-Sense, author versionchrisharrison.net
Written in our own words with AI assistance, based on the sources listed above. The text distinguishes research findings, technical prototypes and established knowledge. This feature is not a breaking-news report.