Distributed AI Agent Architectures for 6G Robotic Control
| Reference number | |
| Coordinator | RISE Research Institutes of Sweden AB - RISE AB - Digitala System |
| Funding from Vinnova | SEK 99 946 |
| Project duration | February 2026 - September 2026 |
| Status | Ongoing |
| Venture | 6G - Competence supply |
| Call | 6G - Supervision of degree work |
Purpose and goal
The project aims to design and validate a scientifically novel multi-agent VLA architecture (Planner, Controller, Safety) for robotic control. The primary purpose is to create a ´collaborative mind´ where distributed AI agents cooperate. The goal is to develop the communication protocols for these agents, which is central to 6G Edge Computing and AI/ML-based optimisation in real time.
Expected effects and result
The main result is a novel VLA framework implemented on a physical robot. The effect will be increased knowledge about distributed AI for robotics and 6G systems. We expect new insights into performance, safety, and robustness compared to existing single-agent systems, strengthening Swedish industrial competence in autonomous systems.
Planned approach and implementation
Implementation involves the design and implementation of the multi-agent VLA framework. This is achieved by fine-tuning pre-trained models using Imitation and Reinforcement Learning. We will use open-source frameworks, specifically Hugging Face lerobot, to test the model on a physical robot and rigorously benchmark its performance and safety.