The Task-Dependent Efficacy of Shared-Control Haptic Guidance Paradigms

Abstract

Shared-control haptic guidance is a common form of robot-mediated training used to teach novice subjects to perform dynamic tasks. Shared-control guidance is distinct from more traditional guidance controllers, such as virtual fixtures, in that it provides novices with real-time visual and haptic feedback from a real or virtual expert. Previous studies have shown varying levels of training efficacy using shared-control guidance paradigms; it is hypothesized that these mixed results are due to interactions between specific guidance implementations ("paradigms") and tasks. This work proposes a novel guidance paradigm taxonomy intended to help classify and compare the multitude of implementations in the literature, as well as a revised proxy rendering model to allow for the implementation of more complex guidance paradigms. The efficacies of four common paradigms are compared in a controlled study with 50 healthy subjects and two dynamic tasks. The results show that guidance paradigms must be matched to a task's dynamic characteristics to elicit effective training and low workload. Based on these results, we provide suggestions for the future development of improved haptic guidance paradigms.