Argotec is an aerospace engineering company based in Turin whose research, innovation and product development activities use engineering and computing for space applications and renewable energy systems. Particular attention is also given to the field of Human Space Flight and Operations as well as the development of space exploration systems. Argotec is also concerned with the production of nanosatellites and engineering solutions[buzzword] that can be adopted on the International Space Station (ISS), which also have immediate applications on Earth.
This paper presents the results of an R&D activity, funded by the ARTES Advance Technology programme of the European Space Agency, aimed at integrating RF, thermal and mechanical functions in a monolithic K/Ka-band antenna-feed assembly through Additive Manufacturing (AM) for GEO High Throughput Satellites (HTS). After a trade-off analysis between different configurations, an AM-oriented version of an asymmetric antenna-feed architecture was selected and realized by the laser-based powder bed fusion of metals (PBF-LB/M) process. The prototype was successfully tested for Radio-Frequency (RF), thermal and mechanical performances.
Earth Observation (EO) applications are facing an increasing demand of computational power at the edge, including Artificial Intelligence (AI), to reduce the downlink bandwidth and provide faster services to the end users. However, radiation-induced faults pose a challenge to the deployment of Neural Networks (NNs) on spacecrafts and one main concern is about the reliability of such applications on Commercial Off The Shelf (COTS) devices. This work presents a fault detection tool that identifies faults by maximizing Neural Network (NN) activations through an optimized set of images. Genetic algorithms are used to find the set with the minimum size while ensuring high activation. Finally, a statistical fault injection process is applied to validate the effectiveness of the images in detecting radiation-induced faults.