REBECCA: Develop efficient and secure Edge-AI Systems
Reconfigurable Heterogeneous Highly Parallel Processing Platform for safe and secure AI
The REBECCA Project officially kicks off in Chania, Greece, hosted by TSI in April 2023. There are 24 partners joining forces to democratize development of novel edge AI systems. Towards this aim, REBECCA will develop a purely European complete hardware and software stack around a RISC-V CPU.
REBECCA aims to provide significantly higher levels of performance, energy/power efficiency, safety, and security than existing systems. The project will develop a complete European hardware and software stack around a RISC-V CPU, utilizing state-of-the-art technologies and making significant scientific and technological advances in key domains, such as processing units, hardware accelerators, and AI libraries and frameworks.
The project will demonstrate its approach on four real-world use cases and two benchmarks from various domains such as automotive, avionics, industrial, robotics or healthcare domains. It will develop a novel chip comprising two tightly-coupled chipsets with features such as RISC-V multi-core, neuromorphic AI accelerator, and memory encryption.
The REBECCA platform will be complemented by a novel HW/SW design space exploration tool which will allow the development of highly efficient REBECCA-based systems. REBECCA will additionally provide the means for safety and security modelling and verification for the developed hardware and software from the very early design stages. The project represents a significant step forward in creating business and societal opportunities for a wide range of stakeholders.
The Role of SYSGO
SYSGO will port its hypervisor and real-time operating system PikeOS for RISC-V CPU architectures and will also develop device drivers (e.g., storage, I/O, TPC) and interfaces/extensions for integration of hardware configurations and AI accelerators based on safety and security analysis. PikeOS' separation kernel will also be extended to support orchestration with dynamic resource allocation, integration of kubernetes and AI-accelerator sharing.
More information at www.sysgo.com/pikeos
More information at the official REBECCA Project Website