Summary prepared by Brian Kay, Selena Kang, and Mark Jolly
Our visit to AI DuPont included a general tour of the Applied Science and Engineering labs. In these labs, engineers and researchers work together to study persons with disabilities and see how they can be helped though technology. This directly applies to cognitive science as scientists strive to incorporate artificial intelligence into robots in order to make them more intelligent. One example is the development of speech synthesizers for children with speech impediments. Scientists have been working on enhancing synthesizers to have the emotion and basic likeness of a child's voice. Another project deals with gesture and recognizing and synthesizing sign language and facial features for the deaf and motor- impaired.
Our first stop was the robotics lab in which the speech synthesis department interacts with the robotics and technological development department. Mr. Randy Glass explained his project of robots attached to wheelchairs. Advantages of these robots include the fact that you can take them with you, and that they are commercial and available. The disadvantages include price and complex mechanical and electrical design. The biggest deficiency is with the user interface, and it is hard to figure out how the machine operates. You may get feedback, or you may get put to a sublevel. The MONESS robot is operated by computer keyboard and the analoging of outputs. A computerized pointer gives the advantage of 3-dimensionality and one is able to input a gesture and represent a motion.
Our next stop was vocational robotics, machines in the workplace for people with manipulation control. With a set of 3 controls, one particular robot was based on the Cambridge University Language system. Though it works 30 times slower than the average human, it is still faster than humans with disabilities and proves advantageous in the workplace allowing people, such as muscular distrophy patients, to earn a living. Another robot we saw was an arm that could be moved off a wheelchair and worked with a camara and a workspace using a laser light pointer to perceive what it sees. An actual action is easy, but object recognition is a higher level task. Thus the intelligent user must share its intelligence with the machine. Though it has good speech recognition, one drawback is that 4 computers must be synchronized. The vision system gives it imformation about an object and functions on an xyz coordinate system and limited grammar. The database is n hierarchy, in which nothing is specified.
Next, we toured the natural language processing lab in which different ways of making communication more intelligent are being explored. The rate at which a computer comprehends is being increased with a telegraphic device which selects words on a grid. The machine expects humans to understand what it means, and does this by installing the listener's processing into the device. Semantics are taken in to fill in what makes it grammatical. There are many language patterns that don't change. Because some dialogue is the same, language is scripted into a template. Language for speakers of another language is also studied. One example is sign language, which has a synatctic structure similar to Chinese. Though it has no audio feedback, it uses knowledge of experiences, language, and spatial organization.
Another lab was the gesture lab, in which computer-human interaction is incorporated into the study of American sign language. A device called a cyberglove with 18 sensors on it is used. Information gathered from the gloves becomes input. It recognizes fingerspelling, handshakes, movements and location. It synthesizes virtual signs by means of the sensors and recieves a notation of how to make individual signs.
Finally, we saw the virtual physical therapy lab, which is built on understanding mathematical interactions with things. We saw a virtual reality physical therapy simulation.
In closing, we were told of many other experiments that take place on the premises in order to enhance the lives of the disabled. One example is visual representation in the blind and how we can allow the blind to experience it by a variety of ways. Overall, there are many differences in cognitive development, and technology today serves to alleviate some of the problems which arise from them.