Semester projects proposal – Autumn 2019

Integrated actuators – Prof. Y. Perriard

Note: Projects are intended for Microengineering, Electrical Engineering, Computer Science and Mechanical Engineering sections.

For information and registrations:

  • Prof. Y. Perriard, LAI, Rue de la Maladière 71B, CP 526, CH-2002 Neuchâtel 2, Office MC A4 298, phone number: 021.695.43.10, e-mail : yves.perriard@epfl.ch
  • P. Germano, LAI, Rue de la Maladière 71B, CP 526, CH-2002 Neuchâtel 2, Office MC A4 208, phone number: 021.695.42.33, e-mail : paolo.germano@epfl.ch

A fixed rate of CHF 600.- for transportation and meal expenses will be granted to each student doing its semester project at LAI on Microcity site in Neuchâtel.

  1. Position Sensor Systems for Magnetic Levitation / P. Peralta

At LAI, a bearingless motor is being developed with the support of a Swiss company. In this system, the rotating part of the motor needs to be magnetically levitated in a continuous fashion, by means of electromagnetic forces.

For successful motor levitation, the rotor position has to be continuously measured. The implemented sensor scheme needs to provide fast bandwidth, present linear characteristics, and be robust against geometric changes, manufacturing imperfections, and electromagnetically immune.

The task will require an evaluation of off-the-market available sensors along their advantages and their limitations. This will imply the construction of a test bench, and ultimately, a weighted evaluation of the tested schemes.

  1. Magnetic Bearings with Simplified Power Electronics / P. Peralta

At LAI, a bearingless motor is being developed with the support of a Swiss company. The integration  of a magnetic bearing will however increase control complexity and the power electronics of the system.

In this work, two topologies which enable simple control schemes will be presented. These will then be analyzed through computational simulations. Ultimately, their performances will be compared to individuate the best variant.

  1. Active access card and non-blocking access gates / Mohaghegh / A. Boegli

PROJECT ASSIGNED

The framework of this work is a project financed by InnoSuisse on behalf of a company based in the Neuchâtel region. The aim is to develop a new approach to perform the access control to ski resorts. Based on an active access card, the system under development  will permit to position customers on the ski resort and give a non-blocking access to facilities. In this work, we propose to explore methods to position the card with good accuracy. This information will then be used to determine whether or not the person has the required authorization to access the facilities.

  1. Development of an automated Linear Impact generator / C. Hernandez

The EPFL – LAI is investigating novel haptic technologies for rendering rich vibrotactile feedback in digital musical interfaces. The main objective is to develop an interactive surface that is able to render a multi-touch vibrotactile stimulus using piezoelectric actuators and wave focusing strategies.

The goal of the project, is to develop an automated system that can generate repeatable impacts in different positions within the boundaries of a surface. Furthermore, the impact amplitude needs to be controlled and modified. The system will be tested over thin plates of different materials, to obtain a dataset of impacts that are recorder using piezoelectric transducers. This dataset will be used to train a deep learning model that will detect the position of the impact.

Our project will allow you to get hands-on experience with the development of a linear actuator or similar device, and to integrate the developed system with a 3-axis CNC table.

  1. AI-enhanced LoRa Based Indoor Localization System / A. Boegli

GPS-based location systems suffer from accuracy deterioration and are almost unavailable in indoor environments. Building upon the ranging capabilities of newer LoRa ICs, this project aims at developing and deploy a location system able to enhance the position accuracy reached by existing LoRa-based location systems relying on TDOA or ranging mechanisms. The use of learning techniques is expected to improve accuracy. In this project, we propose to first play with such LoRa systems before starting to evaluate the use of machine learning/deep learning techniques for fingerprints based positioning.