Northwestern unveils battery-free smart glasses powered by blinks
GlassTENG uses triboelectric sensors on the frame to harvest energy from blinks, smiles and jaw movements, with 93.8% action-recognition accuracy.
Researchers at Northwestern University have developed a battery-free smart glasses prototype called GlassTENG that generates its own power from facial movements, including eye blinks, smiles, jaw clenching and eyebrow raises. The research was reported by science outlet New Atlas and covered by ZDNet Korea.
The prototype relies on triboelectric nanogenerator (TENG) sensors, which produce electricity when different materials come into contact and separate. Three sensors are mounted on the frame: near the angular artery by the nose bridge, over the superficial temporal artery at the temple, and near the ear.
Signal strength varied by movement. Pulse produced the weakest signals at 50 to 100 mV, blinks and eyebrow raises produced 200 to 500 mV, and strong jaw movements reached up to 1.5 V.
Recognition accuracy and limits
A machine learning algorithm trained on signals from 20 participants identified actions with an average accuracy of 93.8%, including 99% for biting and 96% for jaw clenching. Pulse measurements taken from 10 seated resting participants differed from a chest-worn reference heart monitor by an average of 1.82 beats per minute. Speaking was identified correctly only 55% of the time, with 37% of speech misclassified as eating.
GlassTENG currently sends data to a smartphone or tablet over Bluetooth. The researchers said they aim to move processing onto the glasses themselves in the long term.
Quick answers
What is GlassTENG?
It is a battery-free smart glasses prototype developed by researchers at Northwestern University that harvests energy from facial movements such as eye blinks, smiles and jaw clenching.
How accurate is GlassTENG's action recognition?
A machine learning algorithm trained on signals from 20 participants identified actions with an average accuracy of 93.8%, including 99% for biting and 96% for jaw clenching. Speaking was recognized correctly only 55% of the time.
Does GlassTENG process data on the glasses?
Not yet. The prototype currently sends data to a smartphone or tablet via Bluetooth, with on-glasses processing a long-term goal for the researchers.