Remote sensing satellites and aerial vehicles are nowadays indispensable tools for Earth monitoring and for improving the predictability of Earth system processes. In particular, small satellite technology is characterized by lower development costs, low-power electronics, affordable launch opportunities and technology demonstration. A similar trend has been observed in Unmanned Aerial Vehicle (UAV) system development due to rapid industry expansion and the miniaturization trend of imagers and sensors. These trends have enabled agents to be equipped with advanced instruments such as multispectral and hyperspectral imaging systems. Using Hyperspectral imaging onboard a satellite or UAV, however, is challenging due to large information content, limited processing time, power and data downlink.
The project’s overall objective is to provide enhanced intelligence and autonomy to the aerial vehicles and small satellites in Earth monitoring applications using technologies such as artificial intelligence, embedded systems and optimized high-performance computing. ARIEL will contribute to fast decision making by advancing onboard processing on small satellites and aerial vehicles for persistent Earth observation. The planned outputs will contribute toward reaching the goals of the international conventions on sustainable development.
In this project, NTNU will collaborate with industry players who ensure that the research is driven by the industry's needs. The research on the implementation of algorithms for forest monitoring will be performed together with S[&]T, whereas algorithms for ship detection application will be explored together with VAKE. ARIEL will be supported by Xilinx to build solutions that will ensure the processing fulfils the real-time constraints, whereas Inventas will help in the development of a verification framework. The project will extensively use the hyperspectral data provided by HYPSO-1 CubeSat built at NTNU and launched in January 2022.
Remote sensing satellites and aerial vehicles are nowadays indispensable tools for Earth monitoring and for improving the predictability of Earth system processes. Due to limited processing and downlink capabilities, today’s satellite remote sensing data products are commonly generated on the ground. On the other side, autonomous and fast decision making have become among the principal demands in the development of satellite and aerial vehicle systems. To provide onboard generated data products and to enable the real-time acquisition, processing, and decision-making stages, there is a need for advanced onboard processing system architectures. The project’s overall objective is to provide enhanced intelligence and autonomy to the aerial vehicles and small satellites in Earth monitoring applications using technologies such as artificial intelligence, embedded systems and optimized high-performance computing. The intended work aims toward building the complex autonomous satellite and unmanned aerial vehicles (UAVs) systems as elements of a larger concert of agents. ARIEL will address the challenges and opportunities of artificial intelligence using optical imagery, and system development for intensive onboard data processing with a focus on new concepts, methods, and technologies that are revolutionizing small satellite and UAV capabilities in terms of real-time data analytics. Key research challenges are (i) the development of efficient methodology for remote sensing applications, (ii) system design incorporating developed methodology, and (iii) evaluation of the performance, benefits, and feasibility of the use cases. The project brings together a multi-disciplinary team of leading scientists and industry partners with expertise in remote sensing, machine learning, autonomy, high-performance and reconfigurable computing, small satellite systems, bio-geo-chemistry, and ecology.