The Horizon Europe project dAIEdge is part of the Network of Excellence initiative of the EU, with the main goal of capacity building between many european institutions. The topic of dAIEdge is „distributed, trustworthy, efficient and scalable AI at the Edge“, meaning to apply artificial intelligence in small remote systems – such as satellites.
dAIedge
Our demonstrator system is a multiprocessor system on a chip (MPSoC), combining an FPGA with a multi-purpose ARM CPU. While the training of the sea ice classification AI will be performed on our servers, the inference will be run on the FPGA faster and with strongly reduced energy consumption than on regular servers.
In the project, DLR leads the Space Use Case, where we will develop and implement a sea ice classifier from Synthetic Aperture Radar (SAR) data. We will build on our experience of the EO-ALERT project with satellite onboard processing. The development is performed on two ground-based demonstrator systems with an FPGA, similar to devices that could be used in actual satellites.
Processing remote sensing data on board of the acquiring satellite speeds up the data delivery, allowing users such as navigators faster access to the ice information.
