Robots moving out of the laboratory - detecting interaction levels and human contact in noisy school environments

Abstract

To achieve natural human-robot interaction, robots need to distinguish humans from other parts of the environment. We investigate how infrared sensors currently being used on a mobile robot can be used to distinguish human interaction. Different from the previous work, that had been conducted under laboratory conditions involving selected children, the current study took place in noisy school environments with a mix of children. Also, while in previous work each child was only exposed once to the robot in the current longitudinal study, each child encounters the robot five times. The technique that we developed previously for detecting human contact still proved to be reliable, however, results are not as clear-cut, due to noisy and rather unstructured environments that interfered with the robot's sensor readings. We discuss expected as well as unexpected results in light of the challenge to develop robots that can operate under real-life conditions.