Effects of Augmented Feedback and Motor Learning Adaptation on Human–Automation Interaction Factors

Abstract

Increased interest in body–machine interfaces necessitates understanding how to train users to use nontraditional inputs. In this study, a control task driven by subject-activated surface electromyography was developed as a testbed to observe the effects of automated training methodologies on the development of performance, workload, and trust. Forty-eight subjects learned to use a surface-electromyography-based command system to perform a Fitts’s-law-style cursor-to-target task with 120 training trials and 40 evaluation trials. Subjects were divided into four groups: control, concurrent feedback, terminal feedback, and adaptive threshold. The control group trained and learned through repetition using the visual feedback of the cursor position. The concurrent feedback group received additional concurrent visual feedback during command input, and the terminal feedback group had supplementary visual feedback after command input. The adaptive threshold group did not have any additional feedback, but experienced changes in the cursor control designed to induce motor learning adaptation. The results indicate that 1) additional visual feedback improves task performance, workload, and trust during training, and 2) the groups converged in their command proficiency by the end of training.