Reinforcement Learning for an Autonomous Robot
In this projects a robot should optimize its actions using reinforcement learning, for example in a handing-over task. Starting from a simple implementation of a reinforcement learning problem, this procedure should be enhanced with recent algorithms from the literature.
Project type:
Bachelor Thesis
Project overview: We develop an autonomous household robot that adapts to its environment. To facilitate learning in the program, we use the programming language RoLL (Robot Learning Language), which allows the robot to learn every time it performs a certain task. RoLL is not restricted to a certain learning algorithm. In this project, RoLL should be used with reinforcement learning algorithms to optimize some behavior.
Task description: - Getting an overview of reinforcement learning techniques for autonomous robots;
- Implementing and testing a simple reinforcement learning algorithm in RoLL on a simple example;
- Enhancing the learning with techniques described in the literature to include preexisting knowledge
Prerequisites: - Some knowledge about machine learning, especially reinforcement learning;
- Lisp programming skills are a plus, but not required
Professor: Prof. Beetz, Michael, PhD
Supervisor: Dr. Kirsch, Alexandra
State: Running
Student: Patrick Meyer
