< Back to previous page
Project
O&O project Artificial Neural Networks for Hall-Effect Magnetic Position Sensors (AENEAS). (VLAOO22)
The goal of the project is to develop novel Artificial Neural Networks for Hall-Effect Magnetic Position
Sensors (AENEAS) and draw up the roadmap towards a new AI-based solution of smart Hall-effect sensors
developed by Melexis. The proposed algorithms will estimate the position for the cases of 2D and 3D
movements using shallow neural networks. Any electrical system generates magnetic fields which disturbs
the output signal, therefore, the proposed architectures will use its highly non-linear response function to
reject several types of perturbation. A magnetic field simulator will be developed to have access to a large
amount of data so that we can train robust neural network models that estimate an accurate magnet
position for 2D and 3D movements. At the same time the project will develop a hardware platform so that
the trained neural network models will be validated in real-life scenarios. The project aims to provide the
fundamental research to build AI-solutions into magnetic sensors. The final goal, after additional sensor
product development work at Melexis, is to introduce robust and more sensitive magnetic sensors in the
rotary encoder market as well as in the automotive market.
Sensors (AENEAS) and draw up the roadmap towards a new AI-based solution of smart Hall-effect sensors
developed by Melexis. The proposed algorithms will estimate the position for the cases of 2D and 3D
movements using shallow neural networks. Any electrical system generates magnetic fields which disturbs
the output signal, therefore, the proposed architectures will use its highly non-linear response function to
reject several types of perturbation. A magnetic field simulator will be developed to have access to a large
amount of data so that we can train robust neural network models that estimate an accurate magnet
position for 2D and 3D movements. At the same time the project will develop a hardware platform so that
the trained neural network models will be validated in real-life scenarios. The project aims to provide the
fundamental research to build AI-solutions into magnetic sensors. The final goal, after additional sensor
product development work at Melexis, is to introduce robust and more sensitive magnetic sensors in the
rotary encoder market as well as in the automotive market.
Date:1 May 2021 → Today
Keywords:Neural Networks, Hall-effect sensors
Disciplines:Sensors, biosensors and smart sensors not elsewhere classified
Project type:Collaboration project