Pinned Repositories
BallAndBeamController
This project considers a robot on which there is a beam with a ball. The robot is given various reference positions in sequence, which it is to move to without the ball falling off the beam. For this purpose, the angle of the beam and the position of the robot can be actively controlled by a motor. A controller was designed that reacts correctly to reference position changes without violating the constraints, despite disturbances.
FootballTracker
Learning a ball detector and tracker for videos of football games using Computer Vision techniques and Convolutional Neural Networks
InversePendulumController
In this project a controller for an inverse pendulum in the unstable standing position is designed. Flatness-based feedforward control allows the pendulum to be laterally displaced for arbitrary carriage positions. The control concepts are checked and validated in Simulink.
LearningToSimulateMaterialFlow
In this project a Machine Learning model is trained for learning a simulator for Material-Flow systems, where the Machine Learning model is based on Graph Neural Networks. The data for training the model is generated using an accurate physicbased simulator.
KlausHerburger's Repositories
KlausHerburger/BallAndBeamController
This project considers a robot on which there is a beam with a ball. The robot is given various reference positions in sequence, which it is to move to without the ball falling off the beam. For this purpose, the angle of the beam and the position of the robot can be actively controlled by a motor. A controller was designed that reacts correctly to reference position changes without violating the constraints, despite disturbances.
KlausHerburger/FootballTracker
Learning a ball detector and tracker for videos of football games using Computer Vision techniques and Convolutional Neural Networks
KlausHerburger/LearningToSimulateMaterialFlow
In this project a Machine Learning model is trained for learning a simulator for Material-Flow systems, where the Machine Learning model is based on Graph Neural Networks. The data for training the model is generated using an accurate physicbased simulator.
KlausHerburger/InversePendulumController
In this project a controller for an inverse pendulum in the unstable standing position is designed. Flatness-based feedforward control allows the pendulum to be laterally displaced for arbitrary carriage positions. The control concepts are checked and validated in Simulink.