On 3 September 2020 at 03:51 CEST, a Vega launch vehicle carried two FSSCat nanosatellites into orbit, one of which houses the brand new HyperScout 2 instrument. This advanced remote sensing system has two unique aspects. Firstly, it is equipped with both a spectral channel operating at visible wavelengths and a multispectral channel operating in the thermal infrared. Secondly, it incorporates an ultra-low-power artificial intelligence accelerator to perform processing tasks that were previously only possible on the ground.
The Small Spacecraft Mission Service Proof of Concept was launched into low Earth orbit this morning from the Guiana Space Center in Kourou on board an Arianespace Vega launcher. Together with 51 other small satellites, the rocket carried two 6U nanosatellites for the Federated Satellite System tandem mission (FSSCat) of the Nano-Satellite and Payload Laboratory (UPC NanoSat Lab) at the Technical University of Catalonia-BarcelonaTech (Universitat Politècnica de Catalunya).
Poised to monitor sea ice and soil moisture in support of the Copernicus Land and Marine Environment Services, one of the nanosatellites is carrying the first instance of the HyperScout 2 instrument, developed and built by Dutch company cosine and its partners.
Extending FSSCat with a new version of cosine’s HyperScout instrument was made possible by support from the Earth Observation Directorate of the European Space Agency (ESA) and the Netherlands Space Office. For the first time, the already powerful onboard processing capabilities of HyperScout have been extended with multiband thermal infrared imaging, as well as the Myriad 2 Vision Processing Unit (VPU) from Intel. This allows artificial intelligence to run on the spacecraft much more efficiently than before, enabling in-orbit analyses that previously had to be executed on Earth.
cosine led the developments for the first artificial intelligence in-orbit experiment, Ф-sat-1, developed as part of ESA’s Ф-lab program, which is intended to drive space technology innovations using small satellites. The Ф-sat-1 experiment on the FSSCat mission demonstrates these combined capabilities for the first time, on board a small satellite.
For this demonstration, the Ф-sat-1 project consortium developed and trained a neural network to detect clouds in Earth observation data. Onboard differentiation between clouded and unobscured areas of the observations will significantly reduce the amount of data that needs to be sent to Earth. The University of Pisa (IT) developed the cloud detection algorithm, Sinergise (SI) developed the training dataset, and Ubotica (IE) provided the VPU and software testing. cosine extended HyperScout’s spectral range with three thermal infrared bands, modified the electronics and software for the integration of the Myriad 2 VPU, and developed a new method for the onboard processing of the spectral data, which has resulted in the new HyperScout 2 instrument.
After commissioning, HyperScout 2 will support FSSCat applications related to soil moisture, ice extent, ice thickness, as well as the detection of melting ponds over ice, and will demonstrate the Ф-sat-1 capabilities in orbit.
“The HyperScout 2 data is going to be paired with the first HyperScout, which has been in orbit since early 2018. This will allow experimental programs to investigate the use of artificial intelligence on distributed platforms and on imagery acquired with shorter revisit times and which contains additional spectral information. This greatly increases the number and variety of scientific, operational and commercial applications that can be implemented using HyperScout,” explains Marco Esposito, Business Unit Manager Remote Sensing at cosine.
Additional HyperScout instruments are scheduled for launch in 2020 and 2021. The development of these new systems has been made possible by cosine’s 20 years of experience in designing advanced miniaturized space instrumentation and is based on strong collaborations with the partners involved.
The launch of HyperScout 2 showcases the intelligent solutions devised by cosine and its partners to overcome many of the volume, mass and power constraints resulting from the use of small satellite platforms. cosine makes these technologies accessible not only for space applications, but also for terrestrial ones, such as land surveillance, agriculture, and air and water quality monitoring.
The FSSCat consortium consists of DEIMOS Engenharia (PT), Universitat Politècnica de Catalunya (ES), Golbriak Space (ET), Tyvak International (IT) and cosine remote rensing (NL). The Ф-sat-1 consortium comprises cosine Remote Sensing (NL), University of Pisa (IT), Sinergise (SI) and Ubotica (IE).