Space Agencies Projects

The SHERPA project, selected by ESA Phi-Lab Spain, aims to boost climate resilience from orbit. Currently, Earth observation satellites send raw data to ground stations for analysis, causing delays of hours or even days during critical situations.

SHERPA transforms this paradigm by processing information using Edge AI directly on board the satellite. This allows images to be analyzed in space to generate instant alerts about natural disasters, such as wildfires or floods, reducing reaction times to mere minutes.

The technical key to achieving this is the XtratuM/NG (XNG) hypervisor. This technology creates secure virtual environments that strictly isolate demanding AI applications, ensuring they run safely without interfering with the satellite’s critical flight software. In this way, the platform acts as a “digital sherpa” that guides and accelerates the response of civil protection teams.

Reference project of ESA to try out and evaluate methods, technologies, and tools for space mission development using an Earth Observation satellite simulators as basis.

KOSMOS (Kit for Onboard Software in Modular Oriented Systems) is a generic flight software infrastructure based on reusable building blocks, developed by the French Space Agency (CNES). It standardizes essential spacecraft computing functions, allowing teams to focus on mission-specific goals rather than “reinventing the wheel” for projects like JUICE, SWOT, or the IDEFIX rover. To operate securely, this modular architecture relies on the XtratuM hypervisor by FENTISS. Acting as the core virtualization layer, XtratuM strictly isolates hardware resources. FENTISS provides this hypervisor alongside a support package that allows Linux to run as a Guest OS. This enables the simultaneous execution of standard Linux applications and highly critical real-time operating systems on the exact same onboard computer, guaranteeing complete isolation without ever compromising the space mission’s overall safety and reliability.

Avionics demonstrator developed at CNES using a single board computer around the Xilinx Zynq 7030 to demonstrate time and space partitioning implementing typical on-board applications (GNSS, TM/TC,…)