Interpreting the emotions of a person from their gestures, their posture, their intonation or their expression is difficult even for humans, evolved to capture the smallest details and micro-expressions during a talk. To suppose that a robot can do the same with our current technological level is daring. But we are taking steps in that direction.
In 2015, the Japanese firm SoftBank Robotics launched Pepper, an android that can tell by the tone of voice if the person speaking to it is angry, sad or happy . It not only detects these emotions and the meaning of certain postures, but it also emulates them by combining voice, eye movement and certain head turns. Despite its price – in addition to an initial payment of almost 2,000 euros, it is necessary to pay 200 more over 36 months to cover insurance, updates and technical service – the robot sold out the minute it went on sale.
Pepper speaks several languages and is being used, above all, as a claim in Japanese stores, where he talks to customers or tells them where what they are looking for is. And it entertains children: it tells stories and displays educational games on its screen .
Although limited, Pepper is a glimpse of the robots to come in a few years thanks to advances in artificial intelligence and 3D vision systems. These technologies have simplified the analysis of emotions based on gestures or the volume and intonation of sentences . Since with these machines we will interact mainly with voice and non-verbal communication, it is vital that they learn to understand us as a human does.
The applications of these techniques go beyond robotics. The conversational assistants that we use on mobiles and computers, such as Siri and Alexa, could soon detect our mood when talking to them, and evaluate whether or not they change the meaning of their messages. They will give more effective answers or they will notice if we lose patience because the results they offer do not convince us.
Image: Jake Curtis / SoftBank Robotics
