AI Junior Research Group
Motivation
Robots are essential already for a long time in the German industry as well as increasingly also in the private sector. In this context, they must be increasingly dynamic and able to move at a speed that does not become dangerous for humans. At the same time, they should still perform their tasks precisely and quickly, for example for transportation in logistics. When moving fast and with a high center of gravity or when transporting high and heavy loads, the physical dynamics of the system play a major role for control engineering. It can be modeled manually, but is mostly highly abstracted from reality, which is known as the sim-to-real gap. Artificial Intelligence (AI) and Machine Learning (ML) approaches can be used to reduce this sim-to-real gap. The robot can learn a suitable model of itself, or it can take over the control entirely through Reinforcement Learning (RL).
Goals and Approach
The goal of this interdisciplinary AI junior research group is to use ML to optimize the dynamic locomotion of real robots of different locomotion modes in order to overcome the sim-to-real gap. Two ML approaches will be investigated, namely hybrid and guided RL. Hybrid learning, on the one hand, is not intended to replace classical methods, but to supplement them with ML in specific areas where they can actually improve performance. Guided RL, on the other hand, enables independent learning of complex control tasks for robots in the real world by integrating existing domain knowledge.
Innovations and Perspectives
Inspired by the research results of dynamics modeling from the field of humanoid robotics, this project will develop and investigate new ML approaches and platforms. By this, current challenges of AI research in robotics will be solved and can be transferred to different application domains of logistics to target practical benefits in various economic applications.
Team
|
|
|
|---|---|---|
|
Dr.-Ing. Oliver Urbann |
Julian Eßer |
Nicolas Bach |
Publications