Artificial intelligence could detect clouds of smoke from space
Artificial intelligence (AI) and space development by "new space" companies are two technological trends of our time. Combining them and integrating artificial intelligence into satellites and robotic space probes removes limitations due to data transmission bandwidth and time delay, enabling entirely new capabilities for space missions. In a preliminary project, Silicon Austria Labs is working with its academic project partner FH Hagenberg to use the recently launched European Space Agency satellite "OPSSAT" to demonstrate live image processing based on an on-board neural network. An appropriately designed convolutional neural network (CNN) can classify the satellite's Earth observation imagery to identify smoke plumes that could indicate wildfires.
On-board analysis could enable rapid satellite response, for example, for in-depth follow-up observations of the suspected location. On-board smoke cloud identification could therefore contribute in the long term to faster and more accurate remote warning systems for wildfires, which, due to the climate crisis, have massive societal implications. Bringing artificial intelligence to a space platform with limited resources naturally presents a number of challenges that must be overcome - and of course, its solutions will in turn benefit Earth-based embedded AI.