People with autism have a considerable difficulty in picking up cues in the social set up. Not noticing that people get bored or confused during the conversation can have damaging effects, as most people tend to avoid any sort of communication with autistic individuals.
El Kaliouby is developing the device named as 'emotional social intelligence prosthetic', along with his colleagues at the Massachusetts Institute of Technology (MIT). The device consists of a camera that can be pinned to the side of pair of glasses, that can in turn be connected to a handheld computer, equipped with image recognition software, in addition to a software capable enough of reading the emotions displayed. The software then signals the wearer through a vibratory movement, signaling that the listener is not involved.
The researchers had previously demonstrated at the University of Cambridge that her software is capable of detecting emotions associated with agreement, disagreement, concentration, thought and level of interest of the listener. Similar computer programs designed previously have so far been able to detect the basic emotions such as fear, anger, happiness, sadness, disgust and surprise.
This new device is superior to the conventional ones as such complex emotional are most likely to come up in a conversation. Furthermore, such emotions are difficult to detect as they are communicated in a sequence of movements rather than a solitary expression.
She trained her machine to detect such subtle changes in human expression by showing it more than 100 video clips of actors, expressing different emotions over an 8-second period. No doubt the device is sensitive enough to pick up the emotions right in actors and normal people with an accuracy of 90% and 64% respectively.
Movements of eyebrows, lips, nose, head movements such as nodding, shaking and tilting head are being detected by the device which it associates with the emotion that has been stored in its memory (as shown by the actor). As a reward for its excellent performance, it is now being trained on excerpts from movies and footage captured by webcams.
The device would be tested on autistic individuals who would participate in a voluntary study, to be conducted this week. Identification of a high resolution camera that can be worn comfortably by the user and training autistic individuals to view the camera while its captures the listeners facial expressions represent some of the significant challenges to be overcome before it can be approved and marketed.
The device can also function as a teaching aid, says Timothy Bickmore. 'I would love it if you could have a computer looking at each student in the room to tell me when 20 per cent of them were bored or confused,' he concluded.