Accurate tracking of the hands and fingers allows users to employ natural gestures in various interactive applications. Hand tracking also supports health applications, such as monitoring face-touching, a common vector for infectious disease. However, for both types of applications, the utility of hand tracking is often limited by the impracticality of bulky tethered systems (e.g., instrumented gloves) or inherent limitations (e.g., Line of Sight or privacy concerns with vision-based systems). These limitations have severely restricted the adoption of hand tracking in real-world applications. We present MagX, a fully untethered on-body hand tracking system utilizing passive magnets and a novel magnetic sensing platform. Since passive magnets require no maintenance, they can be worn on the hands indefinitely, and only the sensor board needs recharging, akin to a smartwatch.

Overview

We optimized the layout the magnetometer array and the passive magnets for daily usage. MagX can perform all compute locally and only requires 0.38W total (220mW on the sensor platform plus 159mW on the computing unit) to perform real-time tracking, offering "all day" fully untethered operation on a typical smartwatch-sized battery. Since passive magnets require no maintenance, they can be worn on the hands indefinitely, and only the sensor board needs recharging, akin to a smartwatch.

Performance

We used MagX to conduct a series of experiments, finding a wearable sensing array can achieve millimeter-accurate 5 DoF tracking of two magnets independently. For example, at 11 cm distance, a 6cm × 6cm sensing array can achieve positional and orientational errors of 0.76 cm and 0.11 rad. At 21 cm distance, the tracking errors are 2.65 cm and 0.41 rad. MagX can leverage larger sensor arrays for improved long-distance tracking, e.g., a 9.8cm × 9.8cm array can achieve 2.62 cm and 0.55 rad tracking performance on two magnets at 27 cm distance.

Applications

The robust performance can facilitate ubiquitous adoption of magnetic tracking in various applications. For example, face touching detection for healthcare application, remote control for AR/VR and Metaverse.