Artificial Intelligence (AI) and space development by “New Space” companies are two technological trends of our time. Combining them and bringing artificial intelligence on-board of satellites and robotic space probes will remove limitations due to data transmission bandwidth and time delay and thereby enable completely new space mission capabilities.

In a preliminary project, we at Silicon Austria Labs - together with our academic project partner FH Hagenberg – work towards using the European Space Agencies’ recently-launched  “OPSSAT” satellite to demonstrate on-board neural-network-based processing of its live imaging data. A suitably-designed convolutional neural network (CNN) shall classify the satellite’s earth observation images in order to identify smoke plumes that could indicate forest fires. 

The on-board analysis could enable a fast response of the satellite, for instance, for in-depth follow up observations of the suspected site. Thus, the on-board identification of smoke plume detection could in the long run contribute to faster and more precise remote earlier-warning-systems for forest fires, which have a massive societal impact enhanced by climate crisis. Taking artificial intelligence to resource-constraint space platform, of course brings its own set of challenges to be overcome– and naturally their solutions will in turn benefit earth-bound embedded AI.