Magnetic Localization for an Electromagnetic-Based Haptic Interface

Abstract

In this letter, we develop a magnetic localization system for an electromagnetic-based haptic interface (EHI). Haptic interaction is achieved using a controlled magnetic force applied via an EHI on a magnetic dipole attached to a wearable finger splint. The position of the magnetic dipole is estimated using two identical arrays of three-dimensional magnetic field sensors in order to eliminate the magnetic field generated by the EHI. The measurements of these arrays are used to estimate the position of the magnetic dipole by an artificial neural network. This network maps the field readings to the position of the magnetic dipole. The proposed system is experimentally validated under four cases of the magnetic field generated by the EHI. These cases are likely to be encountered during the haptic rendering of virtual shapes. In the absence of the magnetic field, the mean absolute position error (MAE) is $0.80\pm 0.30$ mm ( $n=125$ ). Static and sinusoidal magnetic fields are applied, and the MAEs are $1.26\pm 0.43$ mm ( $n=125$ ) and $0.91\pm 0.33$ mm ( $n=125$ ), respectively. A random time-varying magnetic field is applied, and the MAE is $0.86\pm 0.33$ mm ( $n=125$ ). Our statistical analysis shows that the repeatability of the magnetic localization is acceptable regardless of the field generated by the EHI, at $\alpha$ = 0.05 and 95 $\%$ confidence level.