The following article was featured in Issue #1 of the INESC INOV-Lab Newsletter, published on 30 September 2025. You can read this and other articles in full by downloading it from the Newsletter section of our website. We hope you enjoy your reading!
The OSiRIS project (On-board SAR Imaging with Reconfigurable AI Systems) draws inspiration from Osiris, the Egyptian god of fertility, agriculture, and renewal. It addresses two major goals of the UN 2030 Agenda for Sustainable Development: conservation of terrestrial and maritime ecosystems and pollution reduction. Achieving these requires efficient monitoring systems that can be rapidly deployed to regulate natural resource use, respond to disasters, and mitigate environmental risks.
Current aerospace monitoring systems, such as those using Synthetic Aperture Radar (SAR) onboard Sentinel satellites, provide high-resolution images independent of altitude, weather, or daylight, but their revisit periods limit on-demand monitoring. Deploying low-cost UAVs with on-board SAR processing is therefore of high interest. However, traditional algorithms like Backprojection demand extreme digital signal processing, conflicting with strict size, weight, and power (SWaP) constraints of airborne platforms.
OSiRIS explores a novel approach using artificial intelligence, namely Generative Adversarial Networks (GANs), to generate SAR images without classical DSP image generation algorithms. Moreover, instead of conventional FPGA implementations, the project targets new Xilinx VERSAL devices with embedded AI engines, enabling efficient deployment.
By replacing slow, resource-heavy units with smaller, faster ones, OSiRIS aims to pave the way for a new generation of on-board SAR processors, with potential for further development and industrialisation.