Présentations des mémoires de master

This project aims to design an very-low-power data logger for an ASIC. In a first design step a prototype VHDL implementation on an FPGA techology is foreseen.
The final ASIC is intended to replace an existing discrete solution, which is under development in a peer project to improve orthodontic patient compliance using multi-brackets and clear aligners. The primary functions of the data logger are to periodically measure forces from orthodontic elastics, store the data on the device, and transmit the stored data upon request using wireless NFC technology. The final orthodontic microsensor device must be highly compact, necessitating a very low power implementation du to the limited battery capacity. The most crucial goal is to develop a hardware algorithm that supports a very-low-power implementation with ASIC technology. Central to the VHDL design is the development of a hierarchical state machine framework that efficiently manages the I2C slaves: the real-time clock, pressure sensor, and RFID tag. This project results in a working prototype using FPGA technology.

Profile Electrical Engineering

Bio-signals play a crucial role in the understanding and diagnosis of various physiological processes, such as cardiac rhythm, breathing, oxygen and temperature regulation. So far, our knowledge is primarily based on bio-signals acquired using distributed extra- or intra-corporeal sensors and mutually exclusive signal processing techniques to extract corresponding vital signs. However, continued miniaturization of transducers in combination with advancements in flexible printed circuit board technologies have enabled fabrication of smart catheters integrating multiple sensing modalities on a single device. Hence, bio-signals simultaneously acquired from several vital organs, locations, and modalities at a suitable orifice (such as the esophagus) are now possible, paving the way for a more comprehensive understanding of their nature and synergizing their properties for improved signal processing. To facilitate multi-modal signal acquisition, our institutes are seeking a versatile interface that can acquire bio-signals with a novel, liquid-crystal polymer (LCP)-based catheter (see Figure) integrating multiple multiplexed sensors (e.g. electrodes for bio-potential/impedance, pressure, oximeter
and temperature sensors).

Profile Electrical Engineering

Digital payments that are comprehensive, privacy compliant and secure are a game changer when it comes to replacing physical money. The GNU Taler payment system provides a modern, privacy-preserving, tax-friendly and secure method for making digital payments.
The aim of this project is to produce a first proof of concept for a one-way offline payment using OTP. The context here is the GNU Taler payment system. First, the OTP scheme needs to be investigated for embedded hardware implementation. Second, an identification of the key requirements such as computing power, power consumption, interface, time generation and so on. Thirdly, to assess the requirements for different technologies – e.g. specific microcontrollers. Finally, an initial prototype to serve as a proof of concept for such a system. The prototype should be a demonstration that the system is technically feasible. This will be followed by the steps of integration, optimization and industrialization.

Profile Electrical Engineering

Motion capture (MoCap) technology is used to determine the three-dimensional movement of a body in space. Infrared (IR)-based MoCap systems represent the current standard for high tracking accuracy and sampling rate. Although MoCap technology is mature, the cost of commercial systems is prohibitively high, and suffers from vendor lock-in. There is a lack of affordable, open-source, high-precision, and low-latency solutions that use IR for MoCap. Most openly developed systems use machine learning and omit any trackers to gather motion data, which can come at the cost of tracking accuracy, flexibility of tracking subjects, robustness, and sampling rate. Field-programmable gate arrays (FPGAs) have the potential to massively speed up existing software algorithms by implementing them in hardware. The hardware architecture for an optical marker detector proposed in [1] has been extended and ported to an open-source FPGA platform. The system has been extended with a data transmission unit, which transmits the detected markers to a system capable of plotting their two-dimensional position.

Profile Electrical Engineering

The treatment of surfaces is an important issue in a wide range of industries. In the watch industry, these processes not only represent a bottleneck in production, they are also a relevant cost driver.
New technologies such as plasma electrolytic polishing (PEP) therefore have great industrial potential. A proof of concept of PEP technologies is being carried out as part of this study. An application is being tested on both the PEP bath and the PEP jet. During the test series, three materials were tested that are frequently used in the watch industry: Brass (CuZn38Pb2), carbon steel (1.1274) and stainless martensitic steel (1.4112). It is not only crucial whether the PEP treatment works technically, but also what concrete potential exists. To achieve this, the process must meet various quality and process requirements. Low roughness, high gloss and the possibility of carrying out the treatment zonally and not over the entire surface are the most important aspects here. Furthermore, a cost calculation was carried out and compared with the current processes.

Profile Mechanical Engineering

This research project aims to enhance the efficiency and quality of image and video recordings of blood cells (RBC and WBC) captured with a microscope and processed via a Jetson device. During this process, a bitstream is generated, whose encoding can be influenced. The project explores the codecs H.264 and H.265, supplemented with various levels and bitrates, to achieve optimal compression while maintaining consistent image quality. As part of this work, a tool is being developed to test and evaluate different settings for image and video compression. This tool provides qualitative measurements, allowing the identification of the optimal configuration. This configuration is then embedded into the bitstream to compress the images without requiring further verification.
The main goal of the project is to develop this tool, which identifies the optimal configuration for compressing blood cell images. Both image quality and storage space requirements are optimized in the process. Through iterative evaluation and software adjustments, the tool is ensured to meet the practical needs. The project enables efficient and high-quality compression of blood cell images, improving the processing and storage of this medical data.

Profile Business Engineering

The new Swiss Robotics Competence Center (S3C) bridges research and industry to boost robotic applications in Swiss manufacturing. The S3C‘s main service shall be testing on Cobotic Base Cells (CBCs), validating industrial process with cobotic solutions. In this work, the S3C testing processes and the generated documentation have been designed and validated.

Profile Mechatronics & Automation

In industrial automation, the development of user-friendly interfaces for robot programming is pivotal. This work introduces an innovative, no-code interface designed to make robot programming and reconfiguration accessible to workers without extensive technical skills, particularly benefiting small and medium-sized enterprises with flexible production needs. The implementation is demonstrated in a lab-automation cobotic cell at the S3C.

Profile Mechatronics & Automation

Today, many assembly and verification process of machines are still manual. Due to the difficulty in finding qualified labor and high costs, In this thesis, the integration of automatic or partially automatic processes in the manual assembly using cobots were studied for Rollomatic SA. A use case was
implemented into a cobotic cell. Recommendations are made for future integration.

Profile Mechatronics & Automation

Visual inspection of machined parts requires a high level of concentration but is repetitive and monotonous. Current state-of-the-art solutions to automate this task use anomaly detection, but they only classify parts as defective, not the defect type. A semi-supervised method to predict the type of defect was developed with Criva AG on a Cendres+Métaux SA use case.

Profile Mechanical Engineering

Today, there exists an urgent need for innovative agricultural practices, including robotics, to increase productivity and reduce operational costs. The Swiss startup Arbro is developing an innovative robotic system. Apple picking is a good candidate as apples are resilient, can be grown in dense flat plantations, and make up a significant share of the fruit market. In this project, a synthetic apple branch was built that mimics the fruit stems' dynamics during picking. An end-effector was designed, build, and evaluated using this lab setup for Arbro.

Profile Mechanical Engineering