The primary goal of this research effort is to improve the effectiveness of skill transfer, rehabilitation, and collaboration via haptic devices. We hypothesize that mediating robotic interfaces (either serving as the expert or placed between a human expert and the novice) can facilitate and improve the effectiveness of skill transfer and collaboration in expert-novice pairs as well as in therapist-patient rehabilitation interactions. Various shared control system architectures for skill transfer are being studied in two phases. In the first phase, the human acts as the novice or patient, and the robot serves as the expert. Control schemes for the expert system (the haptic device) will be designed and analyzed theoretically and experimentally. The second phase of the research effort will explore human-robot-human interfaces. Here, we will focus on expert-novice and therapist-patient teams, with a robotic system acting as the mediator between the two.
Phase 1 (Human-Robot architecture)
Prior to 2009, Joel Huegel and Y. Li studied a means of progressively decreasing guidance gains in order to improve skill transfer.
In 2009, the 16-subject Haptic cues in shared-control systems (pilot study) run by Dane Powell compared the performance of groups trained using no guidance (control), summation, temporal separation, and spatial separation methods. Results showed that simple summation was in fact most effective at eliciting skill transfer in most respects. Spatial separation imposed the lowest overall workload on participants, however, and might thus be more appropriate than summation in tasks with other significant physical or mental demands. Temporal separation was relatively inferior at eliciting skill transfer, but it is hypothesized that this paradigm would have performed considerably better in a non-rhythmic task, and the need for further research is indicated.
In 2010, the 50-subject Haptic cues in shared-control systems primary study, also run by Dane Powell, compared the performance of groups trained using no guidance (control), summation, temporal separation, spatial separation, and noise conditions. Data analysis from this study was completed and published in a special issue of Transactions on Haptics (September 2012).