Optical Earth Observation satellites strive to find a balance between spatial and temporal resolution, i.e., ground sampling distance and revisit period. Given a mission profile, using very similar components but different configuration can allow the platform to acquire large-swath observations, offering very high revisit frequencies at the expense of lower spatial resolution or vice versa. We propose an innovative solution where we consider platforms equipped with two imaging systems, a camera acquiring large swath images and a second camera acquiring high resolution images over limited-swath regions. The challenge in this case lies in the selection of the region to image. The extremely tight timing constraints prohibit more traditional approaches which rely on high latency space-to-ground communications. As a result, this challenge can only be met by introducing AI systems on-board the satellite, following an edge-computing paradigm. The proposed concept offers novel sensing capabilities and demonstrates the merits of advanced autonomy. The paradigm is applicable to numerous applications including rapid response and environmental awareness and can also be extended to other domains like deep-space platform autonomy.
