Now two US researchers are building a robot that is as sensitive to emotional states as its human masters.
The robot will have no emotions of its own but will be a sympathetic helper that responds to its owner's mental wellbeing.
Before making the robot helper, the pair of researchers are creating the system that can reliably distinguish between separate human emotions and work out what someone is feeling.
Stress levels
Robot maker Nilanjan Sarkar and psychologist Craig Smith from Vanderbilt University in Tennessee are collaborating on the project to make the sensitive machine.
The pair are not concerned with giving their robot emotions. They just want it to understand what humans are feeling.
Early work is concentrating on ways of accurately sensing what people are feeling, but Professor Smith acknowledged the difficulty of the task facing them.
"The hard fact is that different individuals express the same emotion rather differently," he said.
To get a better idea of how people respond to different situations, the pair are fitting experimental subjects with small wearable sensors that monitor their reactions while they play video games.
The sensors capture information about heart rate to measure stress and anxiety levels.
Analysis of the data captured during stressful and stress-free gaming sessions has revealed a reliable way to measure stress levels.
By combining the heart rate analysis with measurements of changes in skin conductivity and the amount of facial muscle movement, the researchers can get a good idea of when someone is feeling stressed.
Being bored
Information about these indicators of stress levels have been used to create a series of behavioural rules for a small mobile robot.
When someone is feeling stressed, the robot responds by moving towards the subject and asking if it can help in any way.
Now the researchers are designing new tasks to bore and frustrate their subjects in the hope of revealing reliable indicators for levels of these emotional states.
The physiological effects of these emotions will be analysed and used to increase the behavioural repertoire of the small robot.