At Ubotica we are excited to be a partner on the MESEO (Multi-mission Efficient and Secure High-capacity End-to-end Earth Observation) project, a transformative initiative supported by Horizon Europe. Our contribution of cutting-edge AI-driven solutions is key to MESEO’s advancement of Earth Observation (EO) technologies to address the needs of the future.

Pioneering On-Board Processing

A key element within the MESEO project is the use and deployment of on-board edge AI technologies to enable efficient and effective upstream data processing. On-satellite edge compute, which can support the rapid and dynamic deployment of AI models in-flight, has already become a ‘must-have’ on contemporary satellites. Such compute expands the satellite system capabilities and enables enhanced operational efficiency.

Within MESEO, Ubotica is supporting the development of the following key edge AI technologies:

• AI-Driven Image Triaging: This technology involves using AI to analyse and prioritise EO data in real-time. By performing initial data processing on the satellite itself, we can assign a ‘value’ or score to each image. This score can then be used to determine if and how to further process that image. For example, it can be used to prioritise data downlink to ensure that only the most relevant and valuable information is transmitted back to Earth. This approach significantly reduces the volume of data sent, optimising bandwidth and improving overall efficiency.

• Onboard AI Processing of close-to-raw EO Data: Instead of applying AI to fully fledged EO data products generated on-board, and therefore incurring the time and power penalties of data product generation prior to AI interrogation, we are building the MESEO system to be able to extract valuable insights from close-to-raw data. This involves building workflows that process data close to the sensor, deploying AI algorithms that are specifically trained to operate on near-raw data. This reduces the product generation effort by dynamically truncating the product generation workflow for images of low value (e.g., cloudy images).

• Enhanced System Usability and Power Efficiency: By integrating Ubotica’s SPACE:AI advanced edge processing capabilities, we aim to optimise power efficiency and system usability through flight-proven technology. Efficiency is driven by deploying a custom low-power AI acceleration board, and system usability is addressed by the capabilities of the processing solution to execute multiple AI tasks concurrently. A lightweight and low-code software framework ensures that MESEO system users will be easily able to customise and update the on-board processing..

Solving the EO Data Problem

Ubotica’s involvement in MESEO is not just about contributing technology; it’s about transforming how data is processed and utilised in space. Earth Observation currently has a huge data problem – ever increasing amounts of data are being captured from EO sensors as they advance in capabilities and resolution, and this increase is far outstripping the ability to downlink and process this data on the ground in a timely manner. By enabling a holistic system with on-board AI at its core, Ubotica and the MESEO project are addressing this data problem by performing intelligent data reduction and processing directly on-board, at the data source. This AI-driven approach is set to revolutionise EO systems, making them more efficient, more responsive, and more effective.