Abstract — Hand and finger tracking has a major importance
in healthcare, for rehabilitation of hand function required due to
a neurological disorder, and in virtual environment applications,
like characters animation for on-line games or movies. Current
solutions consist mostly of motion tracking gloves with embedded
resistive bend sensors that most often suffer from signal drift,
sensor saturation, sensor displacement and complex calibration
procedures. More advanced solutions provide better tracking
stability, but at the expense of a higher cost.
The proposed solution aims to provide the required precision,
stability and feasibility through the combination of eleven inertial
measurements units (IMUs). Each unit captures the spatial
orientation of the attached body. To fully capture the hand
movement, each finger encompasses two units (at the proximal
and distal phalanges), plus one unit at the back of the hand.
The proposed glove was validated in two distinct steps: a)
evaluation of the sensors’ accuracy and stability over time; b)
evaluation of the bending trajectories during usual finger flexion
tasks based on the intra-class correlation coefficient (ICC).
Results revealed that the glove was sensitive mainly to
magnetic field distortions and sensors tuning. The inclusion of a
hard and soft iron correction algorithm and accelerometer and
gyro drift and temperature compensation methods provided
increased stability and precision. Finger trajectories evaluation
yielded high ICC values with an overall reliability within
application’s satisfying limits.
The developed low cost system provides a straightforward
calibration and usability, qualifying the device for hand and
finger tracking in healthcare and animation industries.