EM-Sense is a novel sensing technology for object detection, triggered only when objects are physically touched. Our approach exploits unintentional EM noise emitted by many everyday electrical and electromechanical objects, such as kitchen appliances, computing devices, power tools and automobiles. These signals tend to be highly characteristic, owing to unique internal operations (e.g., brushless motors, capacitive touchscreens) and different enclosure designs, material composition and shielding. When a user makes physical contact with these objects, electrical signals propagate through the user’s body, as it is conductive. By modifying a commodity software-defined radio receiver, we can detect and classify these signals in real time, enabling robust, on-touch object detection. Our approach utilizes low-cost, commodity hardware and is small enough to be worn on the wrist or, in the near future, integrated into smartwatches. Unlike existing approaches requiring object instrumentation (RFIDs, barcodes, BLE beacons, etc.), EM-Sense can identify objects solely on their EM signatures, without the need for additional tags or hardware.

This research was done in collaboration with Disney Research.

Research Team: Gierad Laput, Chouchang Yang, Robert Xiao, Alanson Sample, Chris Harrison

Awards: Best Talk Award