Robots learn from people themselves

Can the robot future learn to express emotions like humans?

The creation of robots that can interact with humans is the purpose of a European project led by British scientists.

Feelix Growing is a research project consisting of six participating countries and 25 robotists, psychologists and neurologists. Coordinator, Dr. Lola Canamero said the project's goal was to create robots that could 'learn from people and respond appropriately on a social and emotional level'. The project of about 2.3 million euros will last for three years.

Dr. Canamero of the University of Hertfordshire said: 'The emotional world of people is complex but we react to simple signals, things that we do not notice do not pay attention to, like people moving. how.'

Touch device

The project will include creating a series of robots that can pick up sensors that are placed in the human body that they will interact with, and thereby learn their behavior.

Dr. Canamero likens robots like children to learn behavior from the mobilized and emotional world around them. Robots themselves are simple machines, and in some cases they are old machines, the interesting thing of the project is in the software.

Rome learns the behavior that always follows his 'mother'

Dr Canamero said: 'We will use very simple robots as hardware, with some machines we build heads for them. We are primarily interested in programming and developing behavioral capacities, especially in social and emotional interaction with people. '

Robots will learn from the feedback they receive from humans.'It's mostly behavioral and contact feedback. Contact and behavioral feedback, through positive reinforcement such as good words, beautiful gestures or helping robots do something if it gets in trouble. '

University workers are creating different robots for different emotional interactions.

'Detecting expressions'

The robots will receive feedback from simple optical cameras, sound, contact sensors, and sensors that can measure the distance between the machine and the human.

'One of the things we will use to detect facial expressions and movement patterns is a neural network (artificial)'.

Artificial neural networks are being used because they are very effective for receiving changing input information, in this case detecting behavior patterns, voices, movements, etc.

Dr. Canamero said, 'Neural networks learn these types from the examples it observes.'

One of the areas that robots will learn is human movement."Movement speaks a lot about the emotional state. The approximation of the human form and the robot, and the frequency of human contact, through those things we hope to detect the state of love. feeling we need. '

Robots can follow people around and learn the way people communicate with it

Dr. Canamero

Robots will not try to detect such disgusting emotional states, but will focus on states like anger, happiness, loneliness - emotions that affect the right robot how to behave

'Remember behavior'

'It is very important that robots detect that users are angry and robots have done something wrong or if people are lonely, robots need to cheer them up.'

One of the first robots the project created is showing a documented behavior, found in birds and some animals at birth.'It is attached to the object it first saw when it was born. It is always the image of a mother, and it makes them follow her mother. We have a robot model that can follow people around us and learn how people communicate with it. '

Dr. Canamero said robots that can learn human behavior are necessary when machines also play a part in human life.

At the end of the project, two robots will be built to harmonize the different aspects of machinery being developed in Europe.