Embedded artificial gestures: Modulating behavioral coupling in human-machine interaction via motion modification

With the accelerating advances in technology the need to understand human-machine interactions for the purpose of enhancing cooperative activities has become an important theoretical and empirical problem. This study uses Virtual Partner Interaction (VPI) to examine how behavioral coupling between humans and humanoid robots can be modulated through the use of embedded micro gestures or small parametric noise. VPI is a behavioral dynamic clamp constructed from established principles of live two-people interaction during performance of a coordination task. The gestures are akin to “motioning” behaviors and are in the form of parametric noise applied to either the damping or the restoring term of the oscillator dynamics of the virtual partner. To increase the probability of correctly detecting and measuring the individual performances (VP vs. subject), a conflicting task is used: subjects coordinate their rhythmic movements with the VP inphase while the VP is programmed to coordinate nominally only in antiphase. Preliminary results show that that the probability of human subjects switching to the VP's preferred pattern (anti-phase) are enhanced by parametric perturbations on either the damping or the restoring term of the VP oscillator as compared to when no perturbation is applied. The results have implication on how one may enhance human-humanoid robot coordination during performance of cooperative tasks.