Diplomandenseminar

Die AG Robotik von der Universität Bremen bietet in Kooperation mit dem Robotics Innovation Center von der DFKI GmbH eine ganze Reihe von Seminaren und Vorträgen an. In dieser Reihe sind die Bachelor-, Master- und Diplomarbeiten enthalten. Die Termine werden hier regelmäßig aktualisiert. Externe können in vielen Fällen nach Rücksprache mit dem Sekretariat an den Vorträgen teilnehmen.

Vorträge

 -  13:00 Uhr
Exposé

The Impact of Simulation Design on the Simulation-to-Reality Gap in Reinforcement Learning for Underwater Robotics

von: Sören Töpper  (Universität Bremen)
When using reinforcement learning, simulations are essential, because the agent can learn without the risk of damaging the hardware in a real-world experiment. This is particularly evident in the domain of underwater robotics, where real-world experiments and failures are expensive. Following the tr…
 -  14:30 Uhr
Exposé

Task Scheduling in Multi-Task Reinforcement Learning via Task Similarity Metrics

von: Tom Dohls  (Universität Bremen)
Multi-Task Reinforcement Learning (MTRL) enhances sample efficiency by training a single agent on multiple tasks simultaneously. A key challenge in MTRL is negative transfer, where learning one task impedes progress on another, often due to a simplicity bias towards easier tasks. While current task …
 -  14:00 Uhr
Exposé

State Estimation for Bipedal Robot HyPer-1

von: Christoph Müller  (Universität Bremen)
This thesis presents the development of a real-time state estimation system for the bipedal research platform HyPer-1 using an Extended Kalman Filter (EKF). The system estimates the robot’s full state—including base pose, velocity, orientation, and foot positions—using only proprioceptive sensors su…
 -  13:00 Uhr
Kolloquium

Entwicklung und Regelung eines zweiarmigen Oberkörper-Klimmzugroboters

von: Ahmad Baker  (Universität Bremen)
In den letzten Jahren haben humanoide Roboter große Fortschritte in der dynamischen Regelung und der Integration drehmomentstarker Elektromotoren mit geringer Getriebeuntersetzung gemacht. Neue Entwicklungen erlauben ihnen komplexe Bewegungen, die eine hohe Leistungsdichte, Koordination und Präzisio…
 -  15:00 Uhr
Kolloquium

Evaluation of combined Few-short and continual learning approaches for Robotic perception

von: Saira Philip  (Universität Bremen)
State-of-the-art robotic learning frameworks struggle in typical real-world scenarios like dynamically changing or unseen environments that require robots to adapt quickly (few-shot and meta-learning). Furthermore, real-world settings require robots to continually learn without forgetting what they …
 -  13:00 Uhr
Kolloquium

Hybrid-Driven State Estimation for a Humanoid Robot

von: Lasse Hohmeyer  (Universität Bremen)
State estimation refers to all algorithms calculating the internal state of a robot. Since the true state is always unknown, it has to be estimated based on a series of noisy and inaccurate measurements. This is a crucial task in mobile robotics and a more accurate state estimation almost always lea…
 -  11:30 Uhr
Exposé

Using MDP similarity to increase generalization performance for Meta-RL agents

von: Sven Mandic  (Universität Bremen)
Meta-Reinforcement Learning (Meta-RL) addresses the limitations of standard reinforcement learning by enabling agents to quickly adapt to new, unseen tasks. Instead of learning a fixed policy, Meta-RL focuses on training agents to acquire adaptable learning strategies by practicing on a set of avail…
 -  14:00 Uhr
Exposé

Contact Implicit Control for the Vertical Hopper

von: Blanka Burchard  (Hochschule Bremen)
Legged robots bear significant potential for robotic applications that deal with diverse and challenging environments where wheeled robots may struggle to move effectively. Nonetheless, the promise of extended mobility comes at the cost of increased complexity due to the inherentlynonlinear dynamics…
 -  13:30 Uhr
Exposé

Footstep Planning for Bipedal Robot HyPer-1

von: Maximilian Jahn  (Universität Bremen (FB03))
This thesis presents a comparative study of two different footstep planning approaches for DFKI´s bipedal research platform HyPer-1: a heuristic method based on the principles outlined by Marc Raibert and a model-based method us-ing the 3D Linear Inverted Pendulum Model (3D LIPM). The goal is to ach…
 -  13:00 Uhr
Zwischenbericht

Graph-based Projected Entangled-Pair State with GPU Acceleration

von: Alexander Krug  (Universität Bremen)
Quantum algorithms promise to provide speedups or even solutions to previously unfeasible problems. As quantum computing matures, benchmarks are needed as a standardized method to evaluate and compare different quantum computing platforms. Hamiltonian evolution serves as a well studied problem setti…
© DFKI GmbH
zuletzt geändert am 06.09.2024