In the ADAFI project, we propose a new strategy for co-development of digital twins and intelligent adaptive control systems in networked multi-robot systems. The versatility of ICT infrastructures is exploited for construction and simulation of digital twins in the context of intelligent control system design for robots. Multi-fidelity digital twins are built based on the input-output data and/or the underlying physics of the robots. A control system chooses the appropriate fidelity level of a digital twin based on witnessed control and runtime computing performance. The proposed framework utilizes distributed computing methods in a hierarchical cloud-fog-edge context, and models digital twins based on different sources of data in the network. This provides high-performance in-place adaptation and simulation of digital twins. The proposed idea will be applied in industrial applications with special attention on TUAS eRallycross race car power train and drone flight control problems.

2021

2023