Robot Power With Human Pliancy
The benefits of robot workers for business are enormous. Robots are efficient, tireless and far less likely to make mistakes than are human workers. But they also have a major disadvantage: They have difficulty adapting to unexpected problems or unanticipated situations.
The solution: Combine robot power and performance with human judgment.
To accomplish this, the Massachusetts Institute of Technology has set up the Center for Telerobotics. Its purpose is to investigate technology that can integrate human sensing, dexterity and decision-making abilities into automated work processes for jobs that are complex and non-repeating (as opposed to assembly line tasks) and that take place in unstructured environments. The key to doing this is “telerobotics,” or setting up a system in which a human operator works directly with a robot or group of robots.
Today robots are generally unable to deviate from a set, relatively limited routine. By dividing responsibility for control of the robot between the robot and a human operator, the number and type of tasks that the robot can handle would be greatly increased.
Haruhiko Asada, professor of mechanical engineering at MIT and proposed director of the Center for Telerobotics, says the center’s program “will develop basic techniques for communicating the state of the task to the human operator, for helping the robot to understand the human operator’s intention and for sharing control for the task between the operator and the machine.”
Research on telerobotics thus far has focused primarily on outer space and underwater applications. But the technology’s potential also includes areas ranging from the construction and maintenance of buildings to robot surgeons, guards and firefighters. Of particular interest is the development of telerobotics systems that could perform labor-intensive tasks in environments that are unsafe for humans.
Wavelets Viewed as Key to Use of Picture Phone
Digital technologies such as the picture phone are expected to benefit from advances in the field of “wavelets,” a branch of mathematics that allows scientists to break down complicated data into their most fundamental pieces, reassemble them and transmit them quickly without losing important details.
Wavelet research “represents building a bridge between mathematics and engineering,” says Ronald Coifman, a mathematics professor at Yale University.
For example, in electronically transmitting the image of a person’s face, wavelets make it possible to amplify key parts of the signal, such as the corners of the mouth, while reducing less important, redundant information, like the background of the picture.
“Standard telephone lines presently can’t carry enough electronic information to transmit both voice and quality video signals simultaneously,” says Tony Fitzpatrick of Washington University, where mathematicians are conducting a three-year study of wavelets. But soon “wavelets may be able to compress this information better than other techniques, so that voice and image are clear enough for a modern-day picture phone to allow a view to recognize and understand nuances of facial expression.”
Researchers at Washington University are working to develop a high-speed, fiber-optic communications network that, by utilizing wavelets, can transmit voice, data, video and high-resolution images all on the same system.
In addition to enhancing communication technologies, wavelets could have important medical applications. Wavelets “may allow analysts to develop more exact information about the body from CAT scans and nuclear magnetic resonance,” Fitzpatrick notes.
