Biorobotics Laboratory BioRob

Intention Detection in Human-Robot interaction



This video shows some experiments of human-human and human-robot physical interacting while walking. The human-human experiments were performed to identify, through supervised learning, the minimal set of features that result in an accuracy above 90% in intention detection. The results were used to develop an “Interlimb Coordinator” controller for a humanoid robot, which was demonstrated in human-robot experiments.

Published in IEEE robotics and automation letters, DOI: 10.1109/LRA.2018.2864351