Can a robot learn to understand and speak a human language? New results from researchers from the University’s School of Computer Science and published in PLoS ONE, show it can begin to develop basic language skills through conversation with a human!
In the same way that an infant picks up the frequency of sounds in speech, the childlike iCub humanoid robot called DeeChee has learnt some simple word forms. Experiments carried out with DeeChee by Dr Caroline Lyon, Professor Chrystopher Nehaniv and Dr Joe Saunders as part of the iTalk project have shown how language learning might emerge.
Like an infant, DeeChee can only babble and perceives speech as a string of sounds. But after humans speak to DeeChee as if it was a small child, the robot adapts its output to the most frequently heard syllables. It “speaks” word forms such as the names of simple shapes and colours.
Although DeeChee is learning to produce word forms, it does not know their meaning – and learning meanings is another part of the iTalk project’s research.
Teaching DeeChee to speak using methods similar to those used to teach children is a key part of the learning process of the human-robot interaction which could have a significant impact on the future generation of interactive robot systems.
The research paper “Interactive language learning by robots: the transition from babbling to word forms” can be read here: http://bit.ly/LUIeHz
A short video on iCub robot learning names of colours and shapes can be seen here: http://bit.ly/K1vSd3