Human Joint Transportation in a Multi-User Virtual Environment Stephan Streuber Astros Chatziastros
06/06/20162 multisensory visually (physical environment, social environment) somatosensory (information source, constraint) interesting questions: How do humans coordinate their actions with those of others in time and space? Which information and mechanisms are important in order to coordinate actions successfully? Joint Transportation Task
06/06/20163 Which information and mechanisms are important in order to coordinate actions successfully? Environment (e.g. common goal g) Sensing the interaction partner (e.g. gaze) Communication (e.g. pushing) Task representation (e.g. walking trajectory)
06/06/20164 Interaction via Environment: Individuals perform the task as if they would act alone. (no integration of social information) E.g. individuals affect each others action by creating physical constraints in the environment IF ( g < threshold ) THEN compensate; Leader (no extra); Follower (all)
06/06/20165 Interaction via Sensing: Integration of social information during Joint Action in order to maximize performance of the dyad. Non verbal communication: gaze, eye contact, facial expressions, gesture, tonality,… Experiment: only limited visual information: position, orientation, velocity, gaze Prediction: Leader (no extra); Follower (all) IF ( f(g,p) < threshold ) THEN compensate;
06/06/20166 Interaction via Communication: Exchange of information in order to communicate the common goal Experiment: Follower could push the stretcher! Prediction: Leader (share); Follower (share) Broadcast g; IF (g < threshold) THEN compensate;
06/06/20167 Interaction via Task Representation: Both individuals have a representation of the particular task. (e.g. stretcher carrying) They plan their walking paths based on this representation without necessity of integrating social cues about the interaction partner or communicating with them anymore Prediction: Leader (share); Follower (share)
06/06/20168 Collaborative Setup
06/06/20169 Experimental Questions: Which of the four introduced aspects of interaction are mainly involved in joint stretcher carrying? Which information is used? What path characteristics do humans optimize depending on whether they walk individually or jointly (connected via stretcher)?
Individual Condition (4 Trials) 12 subjects, duration experiment ca. 1,5 hours Predictions: Individual Condition (4 Trials) 12 subjects, duration experiment ca. 1,5 hours extra effort = deviation from individual walking trajectory 06/06/ Joint Condition (4 Trials) Procedure: environmentsensingcommunicationrepresentation Leader and Follower share extra effort? NO YES
06/06/ m 12 m Variables: path length compliance of a power law Experimental Methods top view onto the maze 12 m C-Segment
06/06/ Experimental Methods two third power law: Established relation between curvature and end-effector velocity hand drawing (Lacquantini, Viviani) locomotion (Vieilledent et al.) Assumption: Participants will obey a power law if they are not constraint by the interaction partner (the better they act like a single entity) correlation factor between 0 and 1 vtvt vtvt R r=0.84 v t = k∙R1/3
06/06/ Experiment 1 _ results individualjoint Power Law Compliance
06/06/ Experiment 1 _ results individualjoint Path Length Development
06/06/ Experiment 1 _ results How do individuals share the extra effort due to the task related- cooperation in the joint action condition? environmentsensingcommunicationrepresentation Leader and Follower share extra effort? NO YES Power Law CompliancePath Length Development
06/06/ Experiment 1 _ results How do individuals share the extra effort due to the task related- cooperation in the joint action condition? environmentsensingcommunicationrepresentation Leader and Follower share extra effort? NO YES Power Law CompliancePath Length Development
20 subjects, duration experiment ca. 1,5 hours Prediction: 06/06/ Joint Condition A (physical stretcher) Joint Condition B (virtual stretcher) Experiment 2 _ procedure communicationrepresentation Leader and Follower share extra effort? NOYES
06/06/ Experiment 2 _ results Power Law Compliance Path Length Development Physical Stretcher Virtual Stretcher Physical Stretcher Virtual Stretcher communicationrepresentation Leader and Follower share extra effort? NOYES
06/06/ Experiment 2 _ results Power Law Compliance Path Length Development Physical Stretcher Virtual Stretcher Physical Stretcher Virtual Stretcher communicationrepresentation Leader and Follower share extra effort? NOYES
06/06/ Conclusions: The availability of visual or haptic information about the interaction partner did not effect the joint action behavior Subjects planed their actions based on a stable representation of the task. We did not find indication for social learning: subjects shared extra effort due to the cooperative task evenly! high similarity of individual behavior and stretcher “behavior”
06/06/ OEgsToPDskJg1xpmma3dCxzZQAO0jNdK&sk=fYamgWVtaQotgVD66oJMEwC&fs= 1&title=Comedy football first aid OEgsToPDskJg1xpmma3dCxzZQAO0jNdK&sk=fYamgWVtaQotgVD66oJMEwC&fs= 1&title=Comedy football first aid 2....
06/06/ Thank you for your attention!
06/06/ Subject Order Effects: Observing the interaction partner performing the task caused an improved performance in the own action! initial role influenced behavior of leader and follower (behavioral pattern carried over into new role) P = 0.14 joint condition P < 0.05 individual condition Experiment 1 _ results
06/06/ Experimental Methods Velocity curvature relation in walking examined before - what's new here? neither predefined path, nor totally free movement no simple shape of movement (ellipse), no periodic motion analysis of connected dyads Interpretation: higher level control law, central movement planning peripheral biomechanics optimization strategy (related to minimum jerk / maximization of smoothness) Calculation: correlation coefficient log(Radius); log(Velocity) describes how accurate any power law applies r=0.84
06/06/ Outlook