Thursday, August 31, 2006

Robots for the Real World

Professor Rodney Brooks, Director of the MIT Computer Science and Artificial Intelligence Laboratory, gave a talk at the ANU in Canberra on Tuesday on Demographics, Labor Markets, Robotics, and ICT:
The bulge of baby-boomers moving through the population of the west has in many ways determined priorities in medecine, science, and technology for the last 60 years. It will continue to do so for the next few decades. We will see demand for increased productivity in jobs that have traditionally involved manual labor. Robotics provides one way to meet those demands, and the critical research areas in robotics and more general information and communication technologies will be driven by those demands. Australia provides interesting test cases in certain areas while other regions of the world provide complementary cases.
The professor provided an entertaining talk with videos of robots at different stages of development. He argues that low cost flexible robots which can be easily set up by workers (not robot experts) are becoming feasible. Robots need not be able to do everything a human can do to be useful, they can do the easy (for them) more repetitive tasks.

He used the example of the Roomba robot vacuum cleaner, which several researchers have written about.

However, all his robots seemed to be very mechanical. One place where robots have become very successful is on the Internet. Web 'bots search the web and respond to routine email inquiries. These work well because the cyberspace world was specially built for them. Perhaps robots and people can more easily interact in a cyberspace enhanced real world workplace.

As an example a robot with a camera can see how many parts are in a bin next to a production line. But a cyberspace enhanced robot will also be able to see how many parts are in the warehouse next door, by checking the warehouse database. If the human operator is equipped with virtual reality glasses (or just an ordinary computer screen), they will also be able to see this as well and better interact with the robot.

Building robots needs some mechanical sophistication. As an example Professor Brooks discussed a touch sensor made of a dimpled sheet of rubber fitted with magnets and hall effect sensors. When the rubber dimple is deflected the movement of the attached magnet is detected by the hall effect sensor. But this seems a very complex design.

It might be simpler to adapt a cheap membrane keyboard:
...membrane keyboard basically consists of three layers; two of these are membrane layers containing conductive traces. The center layer is a "spacer" containing holes wherever a "key" exists. It keeps the other two layers apart.
... when the top layer is pressed down (with a finger), it makes contact with the bottom layer. ... The switch is now "closed", and the parent device registers a keypress.
If a tiny version of this was made with different thickness rubber buttons, which button was pressed would indicate the force applied. This would make a cheap and simple pressure sensor without the need for magnets or hall effect devices.

No comments: