Automated Guided Vehicles (AGVs) are an established means of automating internal transport processes, particularly in production and warehouse logistics. In the process, large volumes of process and sensor data are continuously generated at various system levels. However, the potential of analyzing this data is often not fully utilized to optimize operational workflows or to support long-term planning decisions based on data. At the same time, the heterogeneity of the vehicles and systems in use complicates a standardized analysis and the generalization of efficiency potentials.
The aim of the KIDaFTS project is to develop an AI-based analysis and recommendation system that significantly enhances the performance of automated guided vehicle systems and other third-party logistics systems through intelligent data processing. To this end, existing process and diagnostic data will be used to automatically derive operationally relevant key performance indicators and prepare them for operational as well as tactical-strategic decision-making. At the core is an expert system that uses machine learning and explainable AI (e.g., through the use of large language models) to generate comprehensible action recommendations for users in intralogistics.
The project is divided into six key work packages:
- Systematic analysis of existing key performance indicator (KPI) systems for intralogistics and decision support methods in the context of automated guided vehicle systems (AGVs).
- Development of a generic data model for the structured collection, storage, and analysis of relevant process and machine data.
- Collection and annotation of real operational data to create a training dataset for AI algorithms.
- Training and configuration of an AI-based decision support system that derives meaningful recommendations from KPIs, domain knowledge, and condition data.
- Development of a user-oriented interface as a technical demonstrator, presenting information in a target group–appropriate and comprehensible manner.
- Evaluation of the developed prototype in a practical test environment with multiple AGVs and realistic transport processes.