UCLA researchers create a wearable glove that translates sign language in real-time
Bioengineers at UCLA have designed a glove that can translate American Sign Language into English speech in real-time using a smartphone application. The goal of the project is to facilitate an easy way for people who use sign language to communicate directly with those who don't understand the sign language without needing a person there to translate.The team also believes the glove could help more people learn sign language. The system is a pair of gloves with thin, stretchable sensors that run up the length of each of the five fingers. The sensors are made from electrically conducting yarn and can pick up hand motions and finger placements that stand for individual letters, numbers, words, and phrases in sign language.
The device can turn the finger movements into electrical signals that are sent to a dollar-sized coin circuit board that's worn on the wrist. The board can transmit signals wirelessly to a smartphone that translates them into spoken words at the rate of about one word per second. The researchers also placed sensors on the faces of those testing the gloves. Sensors were placed between the eyebrows and on one side of their mouths, to capture facial expressions that are part of American Sign Language.
The researchers say the previous wearable systems to translate sign language were limited in usefulness because they were bulky and heavy, or uncomfortable to wear. The lightweight wearable system created by the UCLA team is made from inexpensive materials that also last for a long time. The electronic sensors are also flexible and affordable.
The team says the system performed well in testing with four deaf people who use American Sign Language. Each participant made hand gestures 15 times each, and the custom machine-learning algorithm turned the gestures into letters, numbers, and words that represented. The system can recognize 660 signs, including each letter of the alphabet and numbers zero through nine. A commercial model would need to be improved to support a more extensive vocabulary and faster translation time.