Theses & Semester Projects

During your project, you will develop a mathematical model of the perfusion machine, which will be used to test new control algorithms in simulation. In a first phase, you will focus on the fluid dynamic component of the machine, modeling flows and pressures along the pipes. To gather the necessary data for the parameter identification, you will run your own experiments on the perfusion machine testbench and get the opportunity to collaborate with an entire team of dedicated and experienced researchers from USZ and the IDSC group. The second phase of the project is the development of sophisticated model-based controllers. Specifically, your goal is to develop a MIMO controller for the pressure and flow at the organ’s connections.

Derive a mathematical model and a corresponding MIMO controller for the perfusion machine

Marc-Philippe Neumann, mneumann@idsc.mavt.ethz.ch Dr. David Machacek, davidm@ethz.ch

This project is set to limited visibility by its publisher. To see the project description you need to log in at SiROP. Please follow these instructions:

• Click link "Open this project..." below.
• Log in to SiROP using your university login or create an account to see the details.

If your affiliation is not created automatically, please follow these instructions: http://bit.ly/sirop-affiliate

This project is set to limited visibility by its publisher. To see the project description you need to log in at SiROP. Please follow these instructions:

• Click link "Open this project..." below.
• Log in to SiROP using your university login or create an account to see the details.

If your affiliation is not created automatically, please follow these instructions: http://bit.ly/sirop-affiliate

Several Swiss cities have committed to electrifying their entire bus fleets over the next few decades. In this process, many important decisions need to be made that can have a significant impact on the efficiency and cost of electrification, such as the choice of charging infrastructure or the size and type of batteries. In the ongoing Swiss EBus Plus project, we have therefore developed and published a Python-based open source toolbox for the simultaneous optimization of charging station placement, battery size, and charging strategy. 1 The next phase of this research aims to enhance the toolbox for broader applications. A close collaboration with PostAuto is particularly valuable, given their diverse operational profiles in urban, rural, and alpine regions. In a first step, the compatibility with the available data must be ensured and tested across all Swiss regions. To allow broad application of the toolbox, additional charging strategies (induction, trolley, etc.) and energy sources (battery-assisted solar, fuel cells, etc.) will have to be implemented. Furthermore, the toolbox will be extended to incorporate further constraints, including regional limitations identified through accompanying research at other institutions. Lastly, computational efficiency needs to be improved to speed up the optimization process. The project spans 12 months, divided into a 6-month master’s thesis followed by a 6-month full-time scientific assistantship. To maintain continuity, both parts are ideally completed by the same person.

1. Literature Review & Data Compatibility – Review existing research on electric bus fleet optimization and ensure compatibility of the toolbox with available Swiss regional data.
2. Expansion of Charging Strategies & Energy Sources – Implement additional charging methods (e.g., induction, trolley) and integrate alternative energy sources (e.g., battery-assisted solar, fuel cells).
3. Optimization & Constraint Integration – Extend the toolbox to incorporate regional constraints and operational limitations identified through external research collaborations.
4. Computational Efficiency Improvement – Enhance the toolbox’s performance to accelerate optimization processes and enable real-time feasibility analysis.
5. Documentation & Validation of Results – Systematically document enhancements, validate results through case studies, and ensure usability for future applications.

Mohammad Moradi, ML H 42.1, moradim@ethz.ch

See the attached pdf

See the attached pdf

If you are interested, please contact Manish Prajapat, Amon Lahr and Johannes Köhler {manishp, amlahr, jkoehle} @ethz.ch.

Task:

1) Literature review on the co-design of mobility systems

2) Design a methodology framework for strategic interactions of mobility stakeholders, leveraging techniques from game theory, optimization, and policy making.

3) Conduct real-world case studies targeted to future Swiss scenarios.

Skills Required

Programming, game theory, mathematics, optimization, machine learning, simulation

References

[1] Zardini, G., Lanzetti, N., Belgioioso, G., Hartnik, C., Bolognani, S., Dörfler, F., & Frazzoli, E. (2023). Strategic Interactions in Multi-modal Mobility Systems: A Game-Theoretic Perspective.

[2] Zardini, G., Lanzetti, N., Guerrini, L., Frazzoli, E., & Dörfler, F. (2021). Game theory to study interactions between mobility stakeholders.

If you're interested in this project, do not hesitate to reach out!

Please send your resume/CV, transcript of records and a short description of your interests and skills via email to [minghe@ethz.ch].

More details about my research: https://mingjia-he.github.io/

In this semester thesis, our goal is to enable an F1Tenth car, an autonomous vehicle at 1:10 scale of a Formula 1 car, to race safely on a track that is perceived through RGB-D images captured by an onboard camera. The lane detection system must be lightweight to ensure real-time performance, as it provides a reference for the controller. Furthermore, the controller must ensure the car's safety on the track while effectively handling potential challenges caused by perception inaccuracies. The performance of the proposed method will be validated through hardware testing.

Our objective is to enable the F1Tenth car to autonomously race on an unknown racing track by perceiving the lane markings through an onboard camera. This thesis focuses primarily on the development of the control module that can be combined with the existing perception stack.

We aim to achieve this objective through the following tasks:

• Setting up the existing perception stack for centerline detection and potentially fine tuning it for the specific experiment setup
• Implementing the method that combines several consecutive centerline estimates into a road map
• Implementing a controller that uses the road map as a reference
• Testing the performance of the method on F1Tenth car

Qualifications:

• Experience with Python
• Experience with implementing model predictive control methods
• Basic knowledge of computer vision and image processing
• Interest in robotics and hardware applications

To apply for this semester thesis, please send your resume and transcript to Jelena Trisovic (tjelena@ethz.ch).

Dr. Johannes Köhler, jkoehle@ethz.ch

This project is set to limited visibility by its publisher. To see the project description you need to log in at SiROP. Please follow these instructions:

• Click link "Open this project..." below.
• Log in to SiROP using your university login or create an account to see the details.

If your affiliation is not created automatically, please follow these instructions: http://bit.ly/sirop-affiliate

This master thesis focuses on advancing energy management systems for hybrid ships through the application of Dynamic Programming (DP) and Model Predictive Control (MPC). A key aspect of this research involves integrating Long Short-Term Memory (LSTM) models into DP and MPC frameworks while providing enhanced predictive capabilities for system load forecasting, energy demand estimation, and operational decision-making. As part of an interdisciplinary team, you will contribute to the development of an optimization toolbox tailored to maritime energy management.

The following tasks must be tackled during the study: 1. Literature review on the state-of-the-art in maritime energy management. 2. Development of DP- and MPC-based Algorithms for optimizing hybrid ship energy usage. 3. Cost function definition for ship energy consumption, considering the real-world constraint. 4. Integration of LSTM models within DP and MPC frameworks for enhanced prediction and control. 5. Documentation of the results.

Markus Wenig, ML H 42.1, E-Mail: mwenig@ethz.ch Mohammad Moradi, ML H 42.1, E-Mail: moradim@ethz.ch

marischm@ethz.ch
ffluerenb@ethz.ch

Soft robots offer notable advantages over rigid robots in terms of flexibility and compliance, which facilitate safe and robust interactions with environments. These characteristics make them ideal for diverse applications, such as medical contexts and bionic robotics. However, due to the absence of precise models and the time-varying nature of their system dynamics, achieving optimal control performance on soft robots remains a significant challenge. With the rapid development of machine learning, we now have various learning-based methods to achieve precise pose control of soft robots [4]. Nevertheless, in practice, we have found that the (off-line) well-trained control policies often become ineffective upon deployment due to time-varying system dynamics, which lead us to rethink the traditional “sampling-training-deployment” paradigm in machine learning.

The rise of online learning can effectively enable models to adapt to changing system dynamics. Recent studies have shown that in online learning, even when using an approximate (or even linear) model, it is possible to ensure a sublinear convergence of the regret [2]. Additionally, online learning demonstrates high learning efficiency and stability in both simulations and complex real-world systems.

In this project, we will adopt gradient-based stochastic online learning to achieve precise pose control of a soft manipulator [1,3]. The cable-driven soft robot arm comprises three independent modules (about 0.2m for each module), and each module is actuated independently with three cables. More details of the robot setup, such as structural and mechanical description, can be found in our previous work [3]. We will first establish an approximate model of the system, then learn both feedforward and feedback controllers in an online manner, which enable the soft arm to perform a series of acrobatic demonstrations. Finally, we will compare our approach with some existing offline control algorithms to evaluate the performance in terms of learning efficiency, handling distribution shifts, and tracking accuracy.

Reference

[1] Chen, Z., Guan, Q., Hughes, J., Menciassi, A. and Stefanini, C., 2024. S2C2A: A Flexible Task Space Planning and Control Strategy for Modular Soft Robot Arms. arXiv preprint arXiv:2410.03483.

[2] Ma, H., Zeilinger, M. and Muehlebach, M., 2024. Stochastic Online Optimization for Cyber-Physical and Robotic Systems. arXiv preprint arXiv:2404.05318.

[3] Guan, Q., Stella, F., Della Santina, C., Leng, J. and Hughes, J., 2023. Trimmed Helicoids: An Architectured Soft Structure yielding Soft Robots with High Precision, Large Workspace, and Compliant Interactions. npj Robotics, 1(1), p. 4.

[4] Bern, J.M., Schnider, Y., Banzet, P., Kumar, N. and Coros, S., 2020. Soft Robot Control with A Learned Differentiable Model. International Conference on Soft Robotics, pp. 417-423.

Work packages

Stage 1 - Preparation

• Literature review on state-of-the-art work learning-based methods for soft robotics
• Identify a rough (differentiable) model for the soft manipulator, e.g. system identification in the frequency domain, data-driven method
• Setup the motion capture system for tracking the posture of the soft manipulator

Stage 2 - Combine online learning & soft arm

• Implement iterative learning control (ILC) for the soft manipulator to track a single fixed given reference trajectory
• Implement stochastic online learning method for tracking any reference trajectories (in the workspace of the soft manipulator)

Stage 3 - Analysis & demonstrations

• Quantify the control performance, and compare to a baseline (e.g. an offline trained controller)
• Perform some acrobatics using the soft manipulator for demonstration

Qinghua Guan, qinghua.guan@epfl.ch, EPFL

Hao Ma, haomah@ethz.ch, ETHZ

Cheng Pan, cheng.pan@epfl.ch, EPFL

To apply, please include a short motivation for this project, as well as a copy of your CV and transcripts.

Requirements: experience with a Python deep learning framework, understanding of 3D scene and camera geometry.

Please send us a CV and transcript.

Dr. Hermann Blum (blumh@ethz.ch) Dr. Zuria Bauer (zbauer@ethz.ch) Tifanny Portela (tportela@ethz.ch) Jelena Trisovic (tjelena@ethz.ch)

You will take the lead in selecting visualization methods and developing a user-friendly web-based application. Your project will be comprised of the following work packages:

• Literature research
• Data preparation, e.g., data transformation, aggregation
• Graphic user interface design and development: including storytelling framework, maps, layout, color scheme, and map interaction
• Interface evaluation (optional): conduct user studies to assess the usability

Pre-Requisites (or strong interests in)

• Programming in JavaScript and Python

References

• Storytelling with Maps on Paper and Screen. https://storymaps.arcgis.com/stories/8dc1ba3373fc434bbf17a46d1b23d88b
• Public Transit in ArcGIS. https://storymaps.arcgis.com/stories/b0042dc5d3b04fe29951df8c5210bfb1

Dr. Chenyu Zuo, zuoche@ethz.ch Mingjia He, minghe@ethz.ch

To apply for this project, please send your CV and your transcript of records to the supervisors.

This project is set to limited visibility by its publisher. To see the project description you need to log in at SiROP. Please follow these instructions:

• Click link "Open this project..." below.
• Log in to SiROP using your university login or create an account to see the details.

If your affiliation is not created automatically, please follow these instructions: http://bit.ly/sirop-affiliate

See attachment

Dr. Anna Scampicchio, ascampicc@ethz.ch

see attachment

see attachment

Dr. Anna Scampicchio, ascampicc@ethz.ch