Position-Based Control of Under-Constrained Haptics: A System for the Dexmo Glove
Sebastian Friston, Elias Griffith, David Swapp, Alan Marshall, Anthony Steed
Abstract
The Dexmo glove is a haptic exoskeleton that provides \nkinesthetic feedback in virtual reality. Unlike many other gloves \nbased on string–pulleys, the Dexmo uses a free-hinged link-bar \nto transfer forces from a crank to the fingertips. It also uses \nan admittance-based controller parameterized by position, as \nopposed to an impedance-based controller parameterized by force. \nWhen setting the controller’s target position, developers must use \nits native angular coordinate system. The Dexmo has a number of \nuninstrumented degrees of freedom. Mature forward models can \nreliably predict the hand pose, even with these unknowns. When it \ncomes to computing angular controller parameters from a target \npose in Cartesian space however, things become more difficult. \nComplex models that provide attractive visuals from a small \nnumber of sensors can be non-trivial or even impossible to invert. \nIn this letter, we suggest side-stepping this issue. We sample the \nforward model in order to build a lookup table. This is embedded \nin three-dimensional space as a curve, on which traditional queries \nagainst world geometry can be performed. Controller parameters \nare stored as attributes of the sample points. To compute the driver \nparameters for a target position, the application constrains the \nposition to the geometry, and interpolates them. This technique is \ngeneralizable, stable, simple, and fast. We validate our approach \nby implementing it in Unity 2017.3 and integrating it with a Dexmo \nglove.
BibTeX
@inproceedings{ral2019_positionbasedcon,
title = {Position-Based Control of Under-Constrained Haptics: A System for the Dexmo Glove},
author = {Sebastian Friston and Elias Griffith and David Swapp and Alan Marshall and Anthony Steed},
booktitle = {RA-L 2019},
year = {2019}
}