Applied Reinforcement Learning

What it is about

You will learn practical methods of reinforcement learning (RL) for sequential decision problems and how to apply them. In the end you will be able to model typical RL scenarios, implement common algorithms, and solve and assess simple robotics tasks (e.g. on the e-Puck) with RL.

What you will be able to do

• describe classical RL scenarios
• explain the basic principles of common RL methods
• model real engineering problems with RL
• practically compare the performance of different RL algorithms
• select and justify suitable RL algorithms
• implement RL algorithms and apply them to e-Puck robots

What the module consists of

• Frontalunterricht (zweitägige/blöcke vor Semesterbeginn)Introduction to fundamentals and concepts
• wöchentliche Tutorien (2 Std./Woche)Discussion and support for group projects
• ProjektarbeitDevelopment, implementation, and demonstration of RL solutions

Teaching method

• Frontalunterricht (Tafel und Folien)Conveying of fundamentals and concepts
• Gruppen- und Einzel­diskussionenDeepening and application of definitions and examples
• Projektbetreuung in TutorienSupport for practical implementation and problem solving