A Hall-Sensor-Based Localization Method With Six Degrees of Freedom Using Unscented Kalman Filter

Magnetic sensors are widely used in automotive and industrial applications to measure linear or angular movements. Although the measurement of the magnetic field vector offers the opportunity to estimate all mechanical degrees of freedom, previous approaches could not be applied in these environments. This paper presents an improved method for sensing a magnetic source's position and orientation with six degrees of freedom. To solve the underlying inverse magnetostatic problem, an Unscented Kalman Filter in combination with an analytical model of the magnetic source's field is used. The performance of the method is demonstrated on the basis of a novel multi-axis input device. It comprises a cuboid magnet and a CMOS (Complementary metal-oxide-semiconductor) Hall-Sensor array with up to 36 elements. The localization algorithm and the measurement model are implemented in an efficient way to achieve real-time capability and sampling rates up to 80Hz on embedded hardware. This outperforms known methods significantly and allows for a wide application of multi-degree of freedom sensors. Moreover, an absolute position and orientation accuracy of 71 μm and 1.4° are achieved. The work describes the basis for advanced input devices but can also be transferred to other kinds of magnetic localization problems.