Suggested topics for future Master’s thesis projects at Algoryx
Are you interested in doing a Master’s thesis at Algoryx? On this page, you can review a list of potential master’s thesis projects.
If you don’t find a suggested project idea that fits your profile, feel free to contact us – we have many more ideas waiting to be investigated.
THESIS PROPOSAL
Symmetry Preserving Transformers For Physical Foundation Models
Description: Develop and validate a transformer and tokenizers that reflect the symmetries of physics which can potentially improve the performance of generative artificial intelligence.
THESIS PROPOSAL
Physics Informed Inference In Deep Networks
Description: Develop and validate a method that imposes physics in the inference of trained deep neural networks and thus selects physics-preserving trajectories.
THESIS PROPOSAL
OpenPLX Object Server And Protocols
Description: Develop a web- and cloud-based object storage for OpenPLX models that can be use both in editors and for 3D visualization.
THESIS PROPOSAL
Differentiable Physics Engine
Description: Develop methods and software for a differentiable physics engine, which is an essential component for machine learning controllers and physical AI.
THESIS PROPOSAL
Automatic System Identification
Description: Develop AI methods for automatic system identification of real-world models and parameters into the OpenPLX format.
THESIS PROPOSAL
Physics Informed Foundation Model For Machine Modeling
Description: Train a large language model so that it based on prompting can generate and edit OpenPLX models of high quality.
THESIS PROPOSAL
Human-AI Interface For Autonomous Machines
Description: Fine-tune a large language model so that it can serve as an interface between an autonomous machine and a human operator. Implement this in 3D-simulation,
THESIS PROPOSAL
Physics Authoring In The Cloud
You are: Interested in learning and researching new technologies and concepts. You should be interested in web technologies and possibly gaming physics. You might be a student of Computer Science or Engineering Physics for example.
You will be doing: Advanced web development work, creating a 3D physics editor in the cloud using a Domain Specific Language (DSL) C++ library. This probably means compiling the C++ library to Web Assembly and interfacing with it from a Javascript web framework like React or Vue using 3D and physics libraries.
This is important for: Lowering the threshold for describing physics using a Domain Specific Language. Enabling non-coding simulation experts to model physics and robotics using a graphical UI.
Your work could lead to: A future career in e.g robotics, graphics, gaming, or a future position at Algoryx which takes the result to the market.
THESIS PROPOSAL
Simulation Validation Framework
You are: A student of Engineering Physics, or equivalent. You love solving problems, and have been studying numerical methods. Maybe you have taken a few extra programming courses?
You will be doing: Development of a framework for parameterization and validation of simulation models. The goal is to make simulated robots and real-life robots move identically, by parameterizing simulation models based on training data from the real-life robots.
This is important for: Closing the simulation-to-reality gap. The results from simulated experiments generate what we call “synthetic data”. If the simulation models used move and interact with the surroundings identical to the real world, we get high-quality synthetic data.
Your work could lead to: A standardized method for validating simulation models that can close the reality gap.
This would be a great tool for solving general automation problems and enabling robots to do more general things.
THESIS PROPOSAL
Domain Specific Modeling
You are: A Master’s student in Robotics and Control, or maybe a UX Design student with programming skills.
You will be doing: Development/research of models and numerical algorithms for modelling a conveyor belt that can be applied to the constraint model of AGX Dynamics. You will implement a prototype to test and validate the models.
This is important for: Lowering the threshold for advanced robot simulations. Enabling non-simulation experts to access advanced simulation features.
Your work could lead to: A new standard for robotics! The current standards are far from complete. A widely used format is URDF, which is not capable of covering everything needed for a full robot or environment description.
THESIS PROPOSAL
Constraint-Driven Conveyor Belt
You are: a student of Engineering Physics. You love solving problems and have been studying numerical methods. Maybe you have taken some extra programming courses?
You will be doing: Development and evaluation of robotics software, simulation models and tools for robot simulation. Interviews and research for designing a future proof domain specific language for modeling of robots.
This is important for: Virtual commissioning and digital twins of factories.
Your work could lead to: A job at Algoryx working on developing simulation models to simulate realistic physical behaviors for the industry around the world.