Control of a Telepresence Robot Using Force data

Telepresence robots are robots intended to compensate for non-verbal information during telecommunication. However, current telepresence robots don't have sufficient functionality to send gesture information, within non-verbal information. This research aims to develop a communication system that recognizes the motion of the human and supplements the lack of gesture information by transmitting it to humanoid robots. The method proposed involves motion data acquisition using force data, gesture recognition with CNN (Convolutional Neural Network) and control of a humanoid robot with the transmission of gesture by on-line control. Finally, the proposal is evaluated by the TDMS (Two-Dimensional Mood Scale) to verify the difference from using the current telepresence robot. As a result, we recognized 6 motions with an automatic motion recognition accuracy of 77.8%. Telepresence using a humanoid robot was confirmed to improve comfortable feeling by transmitting a gesture, although a significant difference from existing telepresence robot was not confirmed. Copyright (C) 2020 The Authors.