Jonas Bohn

O

Open Source AI-enabled Smart Inhaler for Asthmatic Patients

Jun 2024 - Jan 2025 (8 months)

This project resulted as part of the Master’s Thesis at the ADAMMA - Core for AI & Digital Biomarker Research at ETH Zurich. The project aimed to develop a novel open-source smart inhaler compatible with the three most common asthma inhaler types (MDI, Turbohaler and Diskus). The smart inhaler uses passive sensors such as accelerometers to infer inhaler usage using machine learning. The project was done in collaboration with doctors and patients from the University Children’s Hospital in Basel, where the smart inhaler was tested in a clinical study. The goal of the project was to develop a tool that can help patients with asthma to better manage their disease and to provide doctors with more accurate data on the patient’s inhaler usage. Gra

C

Clear as Day: Low-Power Object Detection for Challenging Conditions

Oct 2023 - Jan 2024 (4 months)

This work explores sensor fusion for object detection in challenging light conditions using the flexx2 3D camera. The focus is on integrating infrared and depth data to enhance object detection performance on resource-constrained and low-power devices, particularly in robotics and autonomous systems where efficiency and accuracy in object detection are crucial under varied environmental conditions. As part of this work, a novel dataset for object detection combining infrared and depth data is introduced, employing the Faster Objects, More Objects (FOMO) model for sensor fusion. The thesis showcases the feasibility of using sensor fusion and FOMO for fast and low-power object detection on constrained devices. Grade: 5.75/6

A

Automated vessel detection for fetal surgery

Aug 2020 - Feb 2021 (7 months)

As part of my bachelor thesis, I developed an automated blood vessel detection algorithm using deep learning techniques. All algorithms have been implemented in Python using Tensorflow and Keras with a strong focus on semantic segmentation of fetoscopic images and videos. Grade: 6/6

S

Study on a robotic arm for sampling lunar regolith

Mar 2020 - Aug 2020 (6 months)

Within this project, I developed a robotics operations concept for an ISRU (In-Source Resource Utilization) mission to the moon in collaboration with Airbus. Within this mission, the robotic system's goal is to sample regolith from the lunar surface and perform various manipulation tasks with a reaction chamber placed on the lander. The robotics operation concept included the analysis and sequencing of the operation modes, which the system has to fulfill, and the first-order design of a robotic arm and various end-effectors. The overall solution had to be developed under stringent constraints such as reducing weight while maintaining a high level of safety and robustness. Grade: 6/6

About

Detail

Timeline

Résumé