During Warsaw It Days, Michał Affek, an Embedded Machine Learning Researcher, discussed the topic of data processing on remote edge devices in his presentation “Satellite onboard data processing.” The topic revolves around data processing on remote edge devices. Namely, challenges related to such low-resource computing with the use of NVIDIA’s Jetson platform will be presented. A brief introduction to machine learning algorithms, followed by an in-depth explanation of performance optimization methods for their use with limited resources in a harsh environment.
Key points included:
- types of data generated by satellites (broadly)
- edge computational devices suitable for space projects
- methods for neural network optimizations
Satellite data collection project
Michał is building his PhD research with us regarding prioritizing satellite-collected data based on potentially interesting and critical areas. The tools he uses to build the solution involve algorithm optimization techniques (such as neural networks). However, he is also developing his own tools and methods for identifying the relevance of an image and assigning it the appropriate priority. There is also room for new, smaller algorithms based on prioritizing incoming data and allocating importance to a picture (because of restricted resources).
