DC4 - Hybrid Learning Control Strategies for Agent-based Intralogistic Material Flow Systems

Objectives

1. To research the development of hybrid learning-based control technologies that combine the adaptability of machine learning algorithms with the safety guarantees of classical control methods, maintaining strong controller performance while ensuring safety and reliability.
2. To develop technologies that refine simulations based on runtime data and usage of this information to improve the actual controller performance.

Expected Results

1. Hybrid Controller that learns effective intralogistic real time strategies that is safe and reliable during learning and run time.
2. Simulation of intralogistic material flow applications with safe and reliable learning components.

Planned Secondment(s)

S7: UBielefeld, Prof. Neumann, 2 months in M14-15
S8: UYork, Prof. Gerasimou, 2 months in M26-27

Required Skills

Essential

• Computer Science/Engineering degree
• Excellent programming skills (C++, Python)

Desirable

• Practical experiences with (multi-agent) path planning strategies, basic knowledge in ROS and simulation techniques
• Practical experience with Machine Learning algorithms and developing/testing of complex software systems
• Basic understanding of hardware / embedded software development