Today's oceans contain tons of waste, most of which can be found underwater. While a lot of effort has been put into the collection of maritime surface waste, to this day only few attempts were made at collecting litter on the sea floor. Additionally, most submarine measures require human divers, complicating such procedures even further.
As a Team of researchers at TUM and in collaboration with partners across Europe, for the SeaClear project we are working on strategies to automate the collection of underwater waste. For this purpose, we develop concepts for autonomous robots that can support us in this task.
Resulting are a number of interesting research questions in a diverse field of topics:
Hardware: The collection of waste requires the development of appropriate mechanical grippers.
Robotics and Control: Modelling and control of underwater robots and the cooperation with humans call for novel theoretical concepts.
Artificial Intelligence: Bad vision underwater in addition to many unknown disturbances opens up several interesting applications for machine learning.
Today, robots are commonly used in industrial settings for a variety of tasks. However, in our everyday lives, besides simple wheeled systems, advanced robots have not yet found wide usage. One of the main reasons for this fact is that robotic grasping is currently rather underdeveloped. Additionally, most methods for grasping are restricted to parallel-jaw grippers which are insufficient for more complex tasks.
Consequently, we are researching robotic grasping with dexterous (hand-like) grippers in various scenarios. In this pursuit, we are using both modern learning-based methods as well as classical approaches from optimization and optimal control.
Robotics in Maximal Coordinates
Typically, robotic systems are described in minimal (also called generalized) coordinates. Here, each coordinate represents a single degree of freedom of the underlying structure (for example the angle of a pendulum). The advantage of this parameterization lies in the small number of variables and the avoidance of constraints.
However, for modern robots carrying out complicated tasks minimal coordinates are not always ideal. Instead, it can be beneficial to use maximal coordinates basically resulting in a decoupled description of the system which can then be put together with additional constraints. This type of representation offers a number of numerical and control theoretic advantages. At the same time there are quite a few open questions still to be answered.
I am always looking for motivated students showing interest in my research. Feel free to reach out if you are interested in working on a thesis in my field of research, even if no open topics are listed below.
Please contact me by sending your transcript of records and resume (if available) so I can choose a topic matching your skills and background. Additionally, please let me know when you plan to start working on the thesis.
Robotic Grasping and Object Manipulation
(Model-Predictive) Control for Robotic Systems in Maximal Coordinates
Trajectory Optimization and Linear-Quadratic Regulation for a Walking Robot
Inverse Dynamics and Computed Torque Control for Robotic Systems in Maximal Coordinates
Parallelization of Computations for Robotics Algorithms
Cooperative Control for Robotic Systems
Kalman Filter in Maximal Coordinates
Design, Construction, and Control of Real Hardware Systems (e.g., Acrobot, Cartpole, ...)
J. Brüdigam; M. Schuck; A. Capone; S. Sosnowski; S. Hirche: Structure-Preserving Learning Using Gaussian Processes and Variational Integrators. 4th Annual Conference on Learning for Dynamics and Control, 2022 more…BibTeX
Full text (mediaTUM)
J. Brüdigam, J. Janeva, S. Sosnowski, and S. Hirche: Linear-Time Contact and Friction Dynamics in Maximal Coordinates using Variational Integrators. 2022 IEEE International Conference on Robotics and Automation (ICRA) (submitted), 2021 more…BibTeX
Full text (mediaTUM)
J. Brüdigam and Z. Manchester: Linear-Time Variational Integrators in Maximal Coordinates. Workshop on the Algorithmic Foundations of Robotics (WAFR), 2020 more…BibTeX
Full text (mediaTUM)