In a new development, supposedly a breakthrough in wearable technology, a wrist-placed device that continuously tracks the human hand in 3-D designed.
Called FingerTark, the bracelet can sense and translate many positions of the human hand into 3-D. This includes 20 finger joint positions that is tracked using three to four low-resolution, miniature thermal cameras that read contours on the wrist. The use of the device includes sign language translation, mobile health, virtual reality, human-robot interaction, and other areas.
The technology reconstructs in 3-D with keen accuracy – a major discovery by the team, said the lead associate involved in the development. It is the first-ever system to be able to rebuild full hand posture based on contours of the wrist.
Featuring four Miniature Cameras, device superior than earlier ones to reconstruct hand
Prior to this, wrist-placed cameras have been considered too big and obtrusive for everyday use, and most of them could reconstruct only some discrete hand gestures.
Physically, the bracelet is lightweight that allows for free movement. The bracelet not merely captures position of the fingers that has been focus of most past research. Instead, the bracelet uses a combination of machine learning and thermal imaging to reconstruct the hand virtually. Regarding functionality of the cameras, the four miniature thermal ones – each of which is about the size of a pea – capture multiple silhouette images to create an outline of the hand.
Thereafter, a deep neural network stitches together the images of these silhouettes and rebuilds the 3-D virtual hand.
“Among a number of possible uses of the technology, the most promising one is its potential use in sign language translation,” said the lead associate behind the development.
Meanwhile, the current technology for sign language translation requires either to use a glove or have a camera in the environment – both being cumbersome.