Saftey-Aware shared Human-Robot Autonomy with controlled LLM Planning and Simulation based Evaluation
- Masterarbeit -
Description:
This thesis will explore safety aware shared human robot autonomy supported by large language model (LLM) based planning in a simulated environment. The general idea is to study how varying levels of autonomy and different planning constraints may affect robot performance, safety, and the need for human supervision. A simulation based setup will be used to examine how an LLM can support or generate action sequences for robot manipulation tasks and how safety interventions can be modeled and evaluated.
The work is planned to include experiments comparing alternative control or decision strategies to observe their influence on task reliability and adaptability. Quantitative indicators such as task success, efficiency, and the occurrence of safety interventions could be used to analyze performance differences. The expected outcome is to identify measurable relationships between autonomy, safety, and transparency, and to contribute to the understanding of how human aware AI systems can balance user involvement with autonomous decision making.
Research Question:
How do saftey supervision and autonomy modes influence the success, efficiency and robustness of LLM-driven robotic manipulation in simulation?
Anforderungen/Kenntnisse:
Python-Programmierung, Simulation erstellen (mit PyBulltet), Integration von LLM in Agenten
Bearbeitung:
Abdelmageed Kamel Abdelmageed Mohamed
Betreuung:
Prof. Dr. rer.nat. Nele Rußwinkel
Institut für Informationssysteme
Ratzeburger Allee 160 ( Gebäude 64 - 2. OG)
23562 Lübeck
Telefon: 0451 / 3101 5700
