Month: April 2023

A number of in-space software applications has been already trialled on the unique ESA satellite, called OPS-SAT, launched in 2019. Here is short list of some of the most notable ones:

• ContainerWatch. To monitor ocean surface for floating intermodal containers. To reduce the computational resources required to an extent that the application can run as a background task.
• Mobile Satellite Imager. To explore issues in real time satellite imaging from a mobile field operative. To develop a mobile phone application, server infrastructure necessary to queue requests and allow real time control of attitude and camera.
• Autonomous Optical Navigation for Faint Object Detection, Identification & Tracking. Autonomous detection of faint, far planet (asteroid missions) and faint, close object (e.g. GEO) (debris removal missions, MSR)
• Cloud detection on-board. Take a picture. Then run a detection algorithm and see if we can preprocess the image on-board. Output an image with clouds highlighted.
• AI in Space. Proof-of-concept of the use of artificial intelligence machine learning algorithms for performing satellite autonomous tasking. Execution of autonomous satellite tasking for taking images on preselected ground targets under specific conditions (attitude, illumination)
• Onboard real-time aggregated surface analysis. Autonomously identifying and cataloguing regions of interest on distant astronomical bodies. Reduce transmission volume required to identify regions of interest on distant bodies, identifying possible landing sites for an asteroid lander via computer vision methods.
• On-board autonomy or formation flying testing. Develop a twin-sat formation flying simulation in EuroSim on Linux. Put it on OPS-SAT, so that it can ‘communicate’ with the simulated sat.
• Development and test of satellite to satellite (S2S) communication subsystem for small satellite onboard software update. The idea is to use the commercial space based communication systems for small satellite onboard software update.

Introduction

Earth Observation (EO) satellites have been transforming various industries, from agriculture to urban planning, by providing valuable insights and data. Traditionally, these satellites send raw imagery data to Earth, where it is processed and analyzed. However, new approaches are emerging that take advantage of in-space processing and data fusion techniques to provide novel and efficient solutions for businesses. In this blog post, we will explore a few of innovative methods and discuss the potential benefits they offer.

At present, most EO satellite data is processed on Earth, with only minimal processing occurring on the satellite itself. For example, the Landsat program, a series of Earth-observing satellites operated by NASA and the US Geological Survey, captures images of the Earth’s surface and sends raw data to ground stations for further processing and analysis. This traditional approach is adequate for many applications, but it can be limiting when it comes to handling more complex data types like hyperspectral imagery or synthetic aperture radar (SAR) data.

Recent advancements in EO satellite technology have made it possible to perform more pre-processing directly in space, offering substantial benefits for various industries. By incorporating machine learning algorithms on satellites, specific business insights, regions of interest, or spectral bands can be extracted, allowing for more efficient and targeted data transmission. PhiSat-1, also known as FSSCat A, is a CubeSat launched by ESA in 2020 as part of the FSSCat mission, which aims to provide data for the Copernicus Earth observation program. PhiSat-1 is equipped with a hyperspectral imaging sensor and an artificial intelligence (AI) processing unit, which enables the satellite to perform onboard data analysis and filtering. The AI processing unit, powered by the Intel Movidius Myriad 2 Vision Processing Unit (VPU), is specifically designed to perform machine learning tasks directly on the satellite, such as detecting clouds in the captured images. By identifying and discarding images with significant cloud cover, PhiSat-1 can significantly reduce the amount of data that needs to be transmitted to the ground, thus saving valuable bandwidth and improving the overall efficiency of the mission. A follow-up mission, PhiSat-2, to be equipped with multispectral imager is already in the works.

Another innovative concept being explored is data fusion, which involves combining data from different types of satellite sensors. Currently, this process is primarily performed on the ground, but performing it in space could offer several advantages. By equipping satellites with the necessary hardware and software, data fusion can be done more efficiently and quickly, providing real-time insights. This could be particularly useful in emergency situations, where rapid access to accurate information is critical.

For example, consider a natural disaster scenario like a forest fire. By fusing data from optical, short- and longe-range IR sensors on a single satellite, responders could quickly assess the extent of the fire, identify hotspots, and even predict its future behavior. This information would be invaluable for coordinating firefighting efforts and mitigating damage to communities and ecosystems.

In conclusion, the future of EO satellite technology is bright, with new approaches such as in-space processing and data fusion offering novel and efficient solutions for businesses across various industries. By harnessing the power of machine learning algorithms and advanced sensor technology, these methods have the potential to revolutionize the way we access and utilize Earth observation data, leading to innovative applications and valuable insights.