Emergent Systems in Production
Abstract
Modern production environments increasingly rely on heterogeneous robot fleets — autonomous mobile robots (AMR), collaborative robot arms and stationary processing stations — that jointly handle manufacturing and logistics tasks. Today, coordination of these systems is predominantly centralised via supervisory computers or Manufacturing Execution Systems. However, centralised coordination is vulnerable to disruptions, machine failures or short-notice order changes: if the central node fails, the entire process chain comes to a halt.
This M-APR project investigates whether and under what conditions decentralised, emergent coordination strategies — where agents autonomously take on roles and distribute tasks based on local information — outperform centralised control in production scenarios. The focus is on the systematic comparison of centralised and decentralised approaches under controlled variations of adversarial conditions (communication failures, machine failures, dynamic order changes).
Project Description
This 3-semester M-APR project (starting WiSe 2026/2027) compares centralised and decentralised coordination strategies for heterogeneous robot fleets in simulated and real production scenarios. The scientific contribution lies in the empirical characterisation of the transition between both coordination paradigms under varied adversarial conditions.
Structure over Three Semesters
Semester 1 — Foundations and Simulation
- Literature review on decentralised coordination and multi-agent reinforcement learning
- Building the ROS 2-based simulation environment (at least two agent types: mobile and stationary)
- Implementation of a centralised baseline and a simple decentralised approach
Semester 2 — Comparative Study and Learning-Based Coordination
- Implementation of learning-based decentralised coordination (MARL)
- Systematic comparison under varied adversarial conditions (communication failure, machine failure, dynamic order changes)
- Statistical evaluation and identification of transition points between coordination regimes
Semester 3 — Validation and Publication
- Transfer of selected scenarios to real hardware at TTZ
- Evaluation of sim-to-real transferability
- Completion of master’s thesis and preparation of a conference publication (MAPR conference)
Requirements
- Bachelor’s degree in electrical engineering, computer science, mechatronics, robotics or equivalent
- Programming in C++ and/or Python
- Basic knowledge of machine learning / reinforcement learning
- Experience with ROS 2 and simulation (Gazebo or similar) is an advantage
- Interest in multi-agent systems and production robotics
This topic can also be completed as a project or master’s thesis subject to agreement.
Supervision
| Role | Name | |
|---|---|---|
| Supervisor | Prof. Dr. Christian Pfitzner | christian.pfitzner@th-nuernberg.de |
| Co-Supervisor | Waldemar Haag | waldemar.haag@th-nuernberg.de |
Location: TTZ Nürnberger Land, Lauf an der Pegnitz
Further information: TTZ Nürnberger Land · Mobile Robotics Lab
Prof. Dr. Christian Pfitzner
Professor of Mobile Robotics
Professor of Mobile Robotics at TH Nürnberg Georg Simon Ohm, Faculty efi. Director of the TTZ Nürnberger Land. Research interests in autonomous navigation, drone technology and sensor fusion.