Earlier this fall, a British scientist was explaining to a London audience why the maid, the butler, and others around the house would be here for some time to come.
''The perfect home robot is a long way off,'' said Matthew Cowan, director of the Electricity Council Research Center, adding that human beings were so wonderfully flexible that the designers' task was almost impossible.
It would be prohibitively expensive, he said, to design a robot that didn't step on the cat, apprehend the homeowner instead of the burglar, or drown the baby while bathing it.
In industry, too, the robot revolution - hordes of mechanical men invading factories worldwide - has run into trouble.
The total stock of programmable robots in the West has skyrocketed in the past decade, from about 3,000 in 1974 to about 37,000 last year.
According to the British Robot Association, the Japanese still lead the way - with some 16,500 robots today. The Americans have an estimated 8,000. The West Europeans boast another 12,500. In Europe, the West Germans have the most by far with 4,800. The French follow with about 2,150, and the British, Swedes, and Italians have about 1,800 each.
But many experts say the heady days of the past five years - which have seen the number of robotmakers jump from 12 to some 250 - have ended, especially for robot builders.
Dr. Joe Engelberger, the ''father'' of industrial robotics, says the business as a whole has reached a plateau. ''I did a great job of selling robotics,'' he says, ''but I sold the idea to competitors, and not to customers.''
Officials at the British Robot Association expect the production of robots to grow for a while, but agree that installation of robots in workplaces may level off.
''There's certainly an oversupply,'' says the association's assistant general secretary, Brian Rooks, ''and I certainly don't see it going away.''
Only Sweden, which already boasts the highest number of robots in the manufacturing sector per employee (about 30 per 10,000), continues to report rising sales. In other Western countries, sales have flattened.
Dr. Rooks says some forecasters widely overestimated the potential market for robots, leading manufacturers to build thousands more than they could sell.
The manufacturing industry, moreover, has discovered that robots can perform only the simplest jobs - spot welding, for instance - and that using robots for anything other than the most straightforward assembly tasks remains science fiction. Manufacturers have also found that robots can't be left alone to work.
Experts at the Paris-based Organization for Economic Cooperation and Development (OECD) predict that the use of robots in the West will increase until late next year as work posts where the technology has been well proven are converted.
''But after these uses reach saturation level,'' according to a recent OECD study, ''some slowdown is anticipated as robots are gradually applied to the complex assembly tasks.'' The study says that even if the introduction of robots into industrial plants causes job losses, ''The magnitudes are not likely to be large in terms of the economy as a whole.''
OECD experts estimate the percentage of jobs likely to be affected by robot installation in industrial plants up to 1985 is 1.5 for Japan and Sweden (rising to 3 by 1990); 0.4 in West Germany and the US (1.5 and 1, respectively); and under 0.2 in France and the United Kingdom (0.5). But these will be offset in part by the jobs created to produce the equipment and maintain it. The problem is that the qualifications for those jobs will be considerably higher than for those lost because of the robots.
Until now, the automobile industry has been the main user of robots. General Motors' West German subsidiary, Opel, placed what was probably the biggest single robot order of 1983 - $20 million.
The chief obstacle to more widespread use of robots in other industries - such as electrical engineering and metalworking - has been their inability to see or feel. Scientists are making inroads with such machines. But so far robots have been used principally for isolated tasks or simple assembly operations. When robots are more capable of identifying parts and choosing the right one from a random selection, and when they can maintain a high degree of repeatability and what experts call ''positional accuracy,'' say in the early 1990s, they will be widely applicable on the assembly line.
Many leading robot manufacturers - from Unima-tion in the US to ASEA in Sweden to KUKA in West Germany - are attempting to develop a vision system that will enable their machines to ''see'' by capturing a ''still'' of a scene from a monochrome television camera. But that could more than double the cost of a robot. These machines will no doubt be the wave of the future if the ratio in performance terms between the price of labor and the real cost of robots moves in the robots' favor.