For Motion Assistance Humans Prefer to Rely on a Robot Rather Than on an Unpredictable Human

IEEE Open J Eng Med Biol. 2020 Apr 16:1:133-139. doi: 10.1109/OJEMB.2020.2987885. eCollection 2020.

Abstract

Objective: The last decades have seen a surge of robots for physical training and work assistance. How to best control these interfaces is unknown, although arguably the interaction should be similar to human movement assistance. Methods: We compare the behaviour and assessment of subjects tracking a moving target with assistance from (i) trajectory guidance (as typically used in robots for physical training), (ii) a human partner, and (iii) the reactive robot partner of Takagi et al. Results: Trajectory guidance was recognised as robotic, while the robot partner was felt as human-like. However, trajectory guidance was preferred to assistance from a human partner, which was recognised as less predictable. The robot partner also was felt to be more predictable and helpful than a human partner, and was preferred. Conclusions: While subjects like to rely on predictable interaction, such as in trajectory guidance, the control reactivity of the robot partner is essential for perceiving an interaction as human-like.

Keywords: Physical interaction; haptic Turing test; human-human; human-robot; performance and perception.

Grants and funding

This work was supported in part by the EU H2020 ICT-644727 COGIMON, FETOPEN-829186 PH-CODING and ITN-861166 INTUITIVE projects, by the UK EPSRC EP/N029003/1 MOTION and EP/R026092/1 FAIRSPACE grants as well as by the German DAAD.