Telespazio UK, as the prime contractor to ESA for the project, is working with a number of strong and expert sub-contractors comprising the Valencian Research Institute for Artificial Intelligence, University of Edinburgh, University of Leicester, Rothamsted Research and the Science and Technology Facilities Council.
The EU Green Deal sets out the direction that Europe will be taking to predict and manage environmental disasters. Action is now being taken to begin developing the next generation of decision support systems to provide a means to understand the consequences of human activity and enable decision making. Very high precision digital models of the Earth, or Digital Twins of the Earth (DTEs), are being developed, which aim to provide a digital modelling platform to visualise, simulate, monitor and forecast the impacts of activity on the planet, and support sustainability through guided decision-making.
The aim of the AI4DTE project is to map and begin the journey towards effective use of AI technologies in future DTEs. In addition to the use of cutting-edge AI software, the wider aspects of architecture and governance required to properly harness the technology for enterprise scale operational systems and services, are also being considered. Challenges of IT infrastructure, integration, interoperability, scalability, data quality, availability and uncertainty, as well as privacy, security, trust and user expectations, are being studied.
Sponsored by ESA Φ-lab, which aims to become “the reference” for the transformational innovation and a key influencer in the Earth Observation ecosystem, the 15-month AI4DTE project has now passed its midway point.
So far, the Telespazio UK-led team has undertaken a period of stakeholder identification and consultation, performed a deep technology analysis, and captured a baseline of current and future requirements of the system. Now, the team is designing and developing a growing set of AI capabilities, by building an extensible software stack and IDE – the ‘AI-Pipeline’ – and is preparing to demonstrate its value through three example DTE Use Cases (UCs), for Slope Instability, Crop Yields and Road Pavement Condition.
The demonstrator system is comprised of three main components: the IDE, which is a collection of development-related services; a Workspace, which provides a place to store and track data, experiments and trained models, and to view and test changes before they’re published to the exploitation environment; and an Execution Environment, which is designed to allow consumer users to find and exploit data, models and decision-support tools.
The three demonstration UCs have been developed to showcase different AI features and functions of the software stack and IDE at the Data, Model and Decision Support Layers, where the Data Layer considers pre-processing, post-processing, augmentation, data mining and data discovery; the Model Layer considers modelling approaches and solutions; and the Decision Support Layer considers how users can taking full advantage of the available data.
In the final phase of the project and working with the target stakeholders, the effectiveness of the demonstrator with be evaluated through a number of demonstration events, and a roadmap for a future operational system will be developed.
Mark Hewer, CEO at Telespazio UK, said: “The AI4DTE project presents a great opportunity for us to be part of ESA Φ-lab’s mission, which is to accelerate the future of Earth Observation by means of transformational innovations. We are proud to be leading such a strong team and look forward to showcasing our AI-Pipeline demonstrator system, which will help guide the way for AI usage in the wider EU Destination Earth initiative, potentially making a big difference in the fight to create a sustainable future.”
Rochelle Schneider, from ESA Φ-lab, added: “DTEs are set to be a vital tool for society, giving scientists and policy makers greatly enhanced capabilities for monitoring and predicting the behaviour of our planet. The AI4DTE project is a crucial element in the path to building comprehensive, AI-driven Earth models, and we look forward to seeing the outcome of Telespazio’s efforts.”