Hey there, I'm Julian Stiefel! 👋
Based in Zurich, Switzerland, I'm a Robotics Engineer with over four years of experience in developing medical software and spearheading technical initiatives at a medtech startup. My journey includes earning a master's degree in Mechanical Engineering from ETH Zurich, where I focused on Robotics. My expertise revolves around projects focused on robotic perception of the environment, SLAM, and kinematic control, where I thrive in crafting intricate algorithms primarily using C++ and Python.
Beyond robotics and software development, my interests span a wide spectrum including finance, law, literature, triathlon, photography, economy, and craftsmanship, not to mention relishing outdoor adventures. To maintain balance, I dedicate time to volunteering as a firefighter and engaging with diverse communities, always eager to broaden my skill set and refine my abilities.
When I'm not coding, you'll find me immersed in hands-on projects such as building bicycles, crafting furniture or decorations in my workshop, tinkering with microcontrollers for fun projects, and restoring classic cars. These endeavors not only expand my horizons but also provide a refreshing counterbalance to my daily work routine.
cad-percept - Precise robot localization in architectural 3D plans
In this project, my contributions centered on implementing cutting-edge algorithms for precise robot localization (SLAM) within imperfect building models, as well as detecting discrepancies between real-time data and mesh models. Leveraging 3D LiDAR point cloud data from a mobile robotic platform, our pipeline significantly surpassed traditional ICP-based alignment methods, consistently reducing localization errors by a minimum of 30%.
I extend my heartfelt gratitude to my accomplished supervisors, @gawela and @hermannsblum, whose invaluable guidance and expertise were instrumental in shaping this endeavor.
Remote magnetic navigation enables precision telesurgery
At Nanoflex Robotics AG, I spearheaded the implementation and upkeep of the software stack and algorithms crucial for steering magnetic devices in compliance with ISO 13485 standards, particularly for ischemic stroke treatment. A recent highlight involved orchestrating a field study aimed at enabling remote system control by surgeons from anywhere globally. This endeavor entailed integrating essential functionalities for remote control, as well as facilitating data and video transmission within our pre-clinical setup.
The culmination of our efforts materialized in a successful technology validation during a collaborative study with the Mayo Clinic in Phoenix, spanning a distance exceeding 9000 km. Notably, we achieved a remarkable Round-Trip-Time of under 200 ms, demonstrating the robustness and efficiency of our technology in real-world scenarios.
Control of magnetic fields using Mixed Reality
Inspired by the launch of HoloLens 2, we embarked on a groundbreaking venture to visualize magnetic fields and manipulate our magnetic field generator through Mixed Reality. Leveraging Unity and MRTK, I implemented an application that seamlessly integrated with our ROS framework and seamlessly operated with the HoloLens 2.
Real-time logging using InfluxDB and Grafana
In industrial settings, the accumulation of sensor data is commonplace. However, for remote infrastructure of critical importance, it's imperative to monitor this data in real-time to swiftly address potential issues before they escalate into damaging scenarios. In this particular project, I took the helm in developing a C++ hardware data logger tailored to this need. The collected data is seamlessly transmitted to the real-time, time-series database InfluxDB for storage, and subsequently visualized through a Grafana dashboard for intuitive monitoring and analysis.
GUI development using Qt and Qml
While not primarily a user interface designer, I find pleasure in crafting GUIs from time to time. Over time, I've designed and developed numerous interfaces for tasks ranging from data collection and visualization to device control. My preferred toolkit for this endeavor is Qt and Qml, enabling me to efficiently separate design from backend functionality. In one notable instance showcased below, I leveraged Snappy Ubuntu Core to release the entire software package as a snap for seamless deployment on embedded systems.
Komoot BLE navigation
This exciting project aimed to develop a cost-effective turn-by-turn navigation device utilizing an ESP32 and Komoot's BLE Connect API. Stemming from the scarcity and high cost of cycling navigation devices in 2018, this endeavor sought to provide a budget-friendly alternative for cyclists.
Kinematic control of a 5-DoF intravitreal injection robot
At Ophthorobotics AG, I played a pivotal role in the development of software and electronics essential for the kinematic control of their innovative five-degree-of-freedom eye injection robot. This involved crafting various device driver libraries, such as those for Thorlabs and Maxon, to ensure seamless integration and operation. Leveraging inverse kinematics computations, I orchestrated precise control of the end-effector, while also designing and implementing a user-friendly Qt GUI to streamline user interaction.