Robots toddling from factories toward home and office

Joseph F. Engelberger, president of the world's largest robot manufacturer, has something in mind besides industrial robots.

He wants a robot that can scurry around his office kitchen at Unimation Inc. and serve snacks to guests. This robot is to perform tasks by itself, without the help of commands given by remote control. Mr. Engelberger is well on the way to getting a truly high-tech robot.

At this point, Unimation has a mobile robot at its West Coast research facility, perhaps a prototype for ''Isaac,'' the company president's dream. It is a nearly 4-foot-high version of Unimation's PUMA robot, says Ellen Mohr, a Unimation spokeswoman.

''Mr. Engelberger wants Isaac to respond to his voice command, come out of the closet, take orders from his guests, go to the kitchen, prepare the orders, and then serve them,'' Ms. Mohr explains. It should be able to prepare coffee, put a Danish in a microwave oven, and wash the dishes.

Though Unimation doesn't have plans to market Isaac now, the yet-to-be-developed robot will perhaps be employed as Mr. Engelberger's office waiter by 1985.

Until Isaac can perform household chores without external commands, engineers will not be impressed. For without a human around to operate them, a remote-control robot won't do much to relieve the drudgery of housecleaning. These remote-control robots are ''purely chicanery for wowing the public,'' states Carl Helmers, editor of Robotics Age magazine.

What ''real'' robot tinkerers and researchers are working toward, and what Unimation hopes Isaac will be, is an autonomous robot. ''That would involve having an on-board computer directing the actions of the robot without explicit guidance from a human,'' Mr. Helmers says. It means having a robot with artificial intelligence (AI)--a robot that can explore its environment (perhaps through TV camera ''eyes''), interpret and understand what it finds, move on its own without bumping into things, and make some decisions based on reasoning.

Relatively speaking, autonomous robots are not so far in the future as many of us may think. ''My gut reaction is that home-robot technology and price will come together for the consumer in another 10 to 15 years,'' says Hans Moravec, who works on mobile robots at Carnegie Mellon University's renowned robotics lab.

In his own Urbana, Ohio, workshop, with surplus parts and a capital outlay of about $2,000, Charles Balmer has won a little victory in the field of autonomous robots. For four years he has been working on Avatar, a robot designed to respond to voice commands, navigate in a flat environment, lift objects with its arm, and locate and plug itself into an ''energy oasis'' when it needs more power.

Mr. Balmer runs Avatar by computer programs developed separately on his homemade computer. Avatar, hooked up by a tether running from its back to the back of the computer, follows the program's commands. The plan is eventually (in the next two years) to get the programs loaded into Avatar, then disconnect the robot from the terminal so it runs on its own computer. Avatar can already locate its ''energy oasis'' and pull into it for recharging.

Avatar finds its way around by sending out light beams. If an object is blocking its path, the light is reflected off the object and back to Avatar. ''It's rather a crude form of vision,'' Mr. Balmer says, ''but TV cameras are way too expensive, use too much power, and relay too much data'' for Avatar to interpret. ''TV cameras are really the way to go, but I just couldn't do it.''

Editor Helmers believes home robots will reach commercial success by evolving as appliances and entertainment centers. He shares the ''gut'' reaction that it will take about 10 years for the autonomous, mobil robot to become a practical help at home. But he doesn't think technology is the hang-up. ''The big challenge is figuring out exactly what a home robot should do,'' he explains. ''Should it be for pleasure? Should it just work? Should it have a lot of general things it can do, or should it be justified by just one task?''

He also argues that no robot will sell if it has to operate in a modular home where furniture sizes, etc., are uniform. ''The great thing about AI is that it will allow a robot to work in different environments,'' Mr. Helmers exclaims