Automated guidance vehicles (AGV) are increasingly being utilized in industry due to labor shortages and the demand for sustainable and future-oriented processes. Studies show that AGV, particularly in indoor settings (intralogistics), are well researched and considered a key technology for automation and efficiency improvement. However, controlled environmental conditions are required for smooth operation. The process of loading and unloading trucks is predominantly manual, typically performed with forklifts. Currently, no mature solution exists on the market for fully automated truck loading and unloading using AGV. The handling of non-standardized load geometries (e.g., paper rolls) and the operation of AGV in outdoor environments pose significant challenges for a fully automated loading and unloading process. Furthermore, the interaction between humans and AGV in mixed traffic zones is considered a complex and insufficiently researched area. Studies indicate that the coexistence of autonomous vehicles and humans presents numerous safety risks, as human behaviour is often difficult to predict. The objective of AUTARK is to develop and validate a fully automated, outdoor-capable AGV demonstrator (TRL 4) for truck loading and unloading in mixed traffic zones. This system should be suitable for application in industrial environments under harsh and dynamic outdoor conditions. Machine vision will be employed to develop algorithms capable of real-time identification and data processing of dynamic environmental conditions (e.g., truck loading bays), with the goal of increasing truck loading and unloading efficiency by 30%. The identification and classification of human behaviour will enhance safety in mixed traffic zones. This is expected to reduce accident risk by 10% and material damage by 30% through intelligent control of AGV actuators. AUTARK thus contributes to SDGs 8, 9, 12, and 13. To achieve the project's goals, a holistic analysis of the current processes will be conducted at the outset, involving stakeholders to identify technical and procedural requirements. The automated loading and unloading process necessitates the integration of AGV hardware, software, and control systems. Therefore, target processes will be derived, subsystem interactions defined, and a communication concept developed, which will be documented in a requirements specification. Machine vision algorithms for precise object recognition and analysis of human activities will be developed and tested both in simulations and in the laboratory environment of TU Wien’s Pilot Factory Industry 4.0. A proof-of-concept demonstrator (TRL 4) will be developed and validated to test and verify the functionality of hardware and software under simulated and realistic conditions. In doing so, AUTARK addresses the research of key digital technologies with the aim of developing innovative solutions for the flexibilization and automation of processes in production-related environments. This will contribute to scientific advancements with an application-oriented focus on the digital transformation of manufacturing companies.