The AI-DISCO project (Edge-Cloud AI for Distributed Sensing and Computing) addresses the challenges of data throughput, energy consumption, scalability, resilience, and security by researching an edge/cloud platform (AI-DISCO platform). The AI-DISCO platform will feature an innovative, reconfigurable edge node design that supports the integration of RISC-V processor cores, DSP processing units, and parameterizable deep learning accelerators specifically optimized for edge applications. The edge node will be based on neuromorphic AI chips, utilizing innovative technologies such as memristive devices and mixed-signal architectures for in-memory computing. The development of the architectures and system topologies is based on, and supported by, neurobiological insights and computational models. The resulting edge node hardware will be easily adaptable to the diverse requirements of applications and offer the potential for significant energy optimization in AI edge computing. At the cloud level, the development of the AI-DISCO platform is characterized by a consistent implementation of the 3C (Connected Collaborative Computing) architecture.
Partners
- Leibniz-Institut für innovative Mikroelektronik (IHP)
- Brandenburgische Technische Universität Cottbus-Senftenberg (BTU)
- Fraunhofer Institut für Photonische Mikrosysteme (IPMS)
- Leibniz-Institut für Neurobiologie (LIN)
- Perinet GmbH
- ENERTRAG SE
- ABB AG
Prof. Dr.-Ing. Hans Dieter Schotten
