Evaluation of Active Handrest performance using labyrinths with adaptive admittance control and virtual fixtures

The Active Handrest is a large workspace, human-machine interface that provides ergonomic support for a user's hand and arm while allowing the user to retain complete control over a grasped tool. In this paper, we introduce methods to improve the base performance of the Active Handrest's linear admittance controller (V = Kα * F) based on knowledge of user intention, through adaptive admittance, and knowledge of the environment, via virtual fixtures. Herein, we present feasibility studies to apply adaptive admittance and virtual fixtures to the motion of the Active Handrest and measure user performance. We introduce a more relevant evaluation task for the Active Handrest by conducting experiments involving the navigation of virtual labyrinths. The results of these experiments show that adaptive admittance improves a user's ability to accurately navigate a narrow labyrinth, with less cost of time than using constant gains with a linear admittance controller. Additionally, a feasibility study on virtual fixtures shows that user performance for drawing straight lines with virtual fixtures applied directly to the Active Handrest is equivalent to user performance with virtual fixtures applied directly to the grasped tool, as has been done more traditionally. These results underscore the utility of the Active Handrest as they show the potential to achieve highly accurate motions without the need for a robotically enabled or co-manipulated tool.