Convolutional neural networks for robust angular measurement with xMR sensor arrays

The contactless and wear-free measurement of angles, rotational speeds and currents is of prime importance for many industrial applications with several different measurement setups and concepts in use. In the automotive area the deployment of magnetic field sensors based on Hall- or magnetoresistive-(MR) effects increases steadily. Especially MR-based sensors play a dominant role in angular position sensing applications because of their high accuracy and robustness. Nevertheless, upcoming vehicle concepts lead to increasing requirements for such sensors, which cannot be fulfilled by the applied concepts completely. However, in many application fields artificial neural networks have proven the capability to enhance existing solutions. Therefore, in this paper the potential use of neural network computing is examined with regard to stray field immunity for xMR angular position sensors.