Applying a reinforcement learning framework in a testing environment

Type of thesis: Bachelorarbeit / location: Leipzig / Status of thesis: Open theses

The Reinforcement Learning (RL) research group at the TU Dresden (https://rl-dresden.de/) has developed a Deep Reinforcement Learning (DRL) framework for self-driving model cars. The DRL framework will be applied to a test environment in the Living Lab of ScaDS.AI (Center for Scalable Data Analytics and Artificial Intelligence). The task is to identify simulation-to-reality gaps in the transfer of the simulation model to reality and propose and proceed ways to close them. In addition, the DRL framework will be examined for strengths and weaknesses and, if necessary, concepts for overcoming the weaknesses will be suggested.

Counterpart

Dr. Thomas Burghardt

Leipzig University

Service and Transfer Center, Living Lab

TU
Universität
Max
Leibnitz-Institut
Helmholtz
Hemholtz
Institut
Fraunhofer-Institut
Fraunhofer-Institut
Max-Planck-Institut
Institute
Max-Plank-Institut