Alan Turing posed the question of whether a machine could learn like a child in 1950. Systematic investigation into this field began at the end of the twentieth century.
Developmental robotics focuses on embodied sensorimotor skills and social interaction within changing physical settings instead of assuming advanced symbolic reasoning capabilities. This approach differs sharply from classical artificial intelligence which relies on pre-programmed solutions for specific problems.
Robots develop sensorimotor abilities such as discovering their own body structure dynamics and hand-eye coordination through active learning. They also acquire locomotion object interaction tool use and social linguistic skills including turn-taking and lexicon acquisition.
Existing techniques fail to allow real-world high-dimensional robots to learn open-ended repertoires of increasingly complex skills over lifetimes. High-dimensional continuous sensorimotor spaces constitute a significant obstacle requiring new mathematical approaches to solve effectively.
The first international meeting devoted to computational understanding of mental development occurred the 5th of the 7th of April 2000 at Michigan State University. This NSF/DARPA funded workshop established the term by since agents act actively during their own developmental processes.