A Natural and Effective Calibration of the CyberGlove

Abstract

This work addresses the calibration of the CyberGlove, a device which provides information about the position and attitude in space of all the components of the user hand in a haptic system for the fruition of digital 3D contents by blind people. A realistic haptic interaction requires a reliable detection of user movements and a good calibration of the device to account for size and articulation capability of the specific hand. At the same time the application of interest calls for a calibration procedure that should be executable in a natural and straightforward way by every user. The two methods provided by the CyberGlove software are not suitable for these purposes: one is very simple and fast but produces quite imprecise results while the other is effective and accurate but involves a complex and cumbersome trial-and-error process not suited for general users. The proposed method aims to reach satisfactory results using a intuitive and simple approach. The user is asked to assume with his hand several pre-defined poses, each supplying the sensors output associated to known joints angles. Combining these data with the constraints derived by the anatomical structure of the hand it is possible to evaluate the values of the two parameters, gain and offset, that drive the digitalization of the sensor output. This process, in spite of its simplicity, has provided satisfactory results in several experimental sessions, enabling a realistic and reliable mirroring of user movements in the virtual space.