FetchFly.AI: Intelligent Drones for the Future of Intralogistics
Overview
How can artificial intelligence and modern aerial robots revolutionize material supply in production? As part of the collaborative project “FetchFly.AI”, the Chair of Artificial Intelligence and Formal Methods (AIFM) is researching exactly this question, bringing self-learning navigation systems for intralogistics into the third dimension.
The Challenge: Dynamics in Modern Manufacturing
The manufacturing industry is undergoing a transformation: production processes with high product variability at fluctuating production volumes are rapidly gaining traction. Particularly, small and medium-sized enterprises need to design their logistics to be fast and adaptive within dynamic, unstructured environments. Rigid conveyor technology and traditional logistics solutions cannot accommodate such flexible production conditions.
The Solution: Airspace as a Flexible Transport Route
The objective of the FetchFly.AI project is to develop a self-learning aerial robotic system for autonomous material supply. By using drones in production halls, the third dimension of space is unlocked for transport within logistics facilities. This eliminates the need for extensive infrastructure modifications, allowing the factory of the future to be flexibly adapted to changing requirements.
Our Contribution at the AI-FM Chair: Autonomy through AI
As AIFM, we contribute our core competencies to the project to ensure that the aerial robots are not only capable of flight, but also intelligent and safe.
The research focus of our team lies in the development of navigation methods based on sequential decision making. To ensure reliable movement in complex and dynamic production environments we rely on:
- Safe Reinforcement Learning: The aerial robots learn to make optimal decisions for safe route planning in dynamical environments in industrial settings using techniques from safe reinforcement learning.
- SLAM Technologies (Simultaneous Localization and Mapping): For the precise and dynamic real-time mapping of the environment.
The algorithms and navigation models developed in this project are tested and validated under realistic conditions at the flight laboratory of the ZESS (Research Center for the Engineering of Smart Product-Service Systems), among other facilities.
Project Partners
FetchFly.AI is a prime example of successful technology transfer between research and industry. We are pleased to implement this project together with our partners:
- Chair of Production Systems (LPS), Ruhr University Bochum
- Multikopter.de
- w3logistics AG
Project Details & Funding
The FetchFly.AI project is funded under the ERDF/JTF Programme NRW 2021-2027 (funding reference IN-ML-3-013a) and is co-financed by the European Union.
