It is time to start building machines which can learn and be raised in the same way as humans, the authors of an article in the journal Science say.
But computer scientist Professor Juyang Weng, of Michigan State University, US, says it is time to work on systems which "live" autonomously, have bodies suited to their working environment and learn in a general sense.
"According to this paradigm, robots should be designed to go through a long period of autonomous mental development, from 'infancy' to 'adulthood'.
"The essence of mental development is to enable robots to autonomously live in the world and to become smart on their own, with some supervision by humans," he writes.
Underlying principles
Making computers copy the way humans learn should not be as difficult as it sounds if scientists can uncover the underlying principles of mental development.
A living brain does not take in every single aspect of an image it is seeing, but makes general observations about shape, colour and motion.
Experimenting with robots will also improve the chances of answering questions such as how the human brain develops a sense of the world around it, Professor Weng says.
And developmental robots will learn to do jobs which humans shun, such as working underwater and in space, or cleaning up nuclear waste, he says.
Good and bad
Professor Weng and his colleagues have already built a prototype developmental robot, called Sail.
Sail is allowed to explore the real world for itself, but a supervising human shows it toys and reinforces behaviour patterns by pressing a "good" button and a "bad" button.
More money needs to be spent on understanding the way humans and animals learn, he says.
"Biologically motivated mental development methods for robots and computational modelling of animal mental development should be especially encouraged."