In space control centres, three operators are now enough where 20 were once required. This operational revolution is neither marketing hype nor technological fantasy: it demonstrates a rigorous approach to space artificial intelligence, where each algorithm addresses a precise operational need.
A discreet yet profound transformation is underway in the European space industry. The multiplication of launches and the growing complexity of constellations demand a radical rethink of ground operations. "This evolution has been gradual," explains Nicolas Frouvelle, Marketing and Communications Director for Space at CS (Sopra Steria Group). "It addresses a dual challenge: managing the increasing volume of data transmitted by satellites and automating increasingly complex operations to reduce costs."
Faced with this transformation, Sopra Steria has chosen a differentiating approach: prioritising genuine operational efficiency over announcements. "When discussing artificial intelligence in space, we must distinguish marketing from technical pragmatism," the expert notes. "Our added value lies in concrete applications: failure prediction, anomaly detection, operations optimisation." This rigour enables the deployment of solutions combining advanced algorithms, machine learning and, when operationally justified, neural networks.
Anticipation over reaction: the core of space automation
The distinction between predictive management and reactive monitoring structures the entire approach. "Predictive management provides a comprehensive view of the system's current and future state," explains Frouvelle. "Reactive monitoring requires rapid human intervention."
This difference addresses a fundamental space constraint. Unlike an aircraft where the pilot reacts instantly, a satellite requires information to transit through a ground station, reach the control centre, undergo decision-making, then descend back to the space vehicle. "These objects travel at several kilometres per second. Response time is incomparable with aeronautics," he emphasises. In this context, anticipation becomes an absolute operational necessity and artificial intelligence, correctly applied, a decisive tool.
GOSMIC: Sovereignty and modularity at heart
Against proprietary solutions from across the Atlantic, CS has developed GOSMIC, a ground segment that illustrates this pragmatic philosophy. "The objective is twofold: to offer a European alternative to existing systems and enable progressive integration," Frouvelle continues. The architecture relies on microservices enabling the integration of software components from multiple partners.
This modularity responds to commercial and technical logic. An operator can acquire only the control module if they already possess trajectory calculation tools, or vice versa. "We offer a complete product, but one that can be sold by components and complemented by other solutions," he says. This flexibility adapts to the specific needs of each space operator.
The operational gains demonstrate the effectiveness of this approach. "For a 20-satellite constellation, a conventional ground segment requires 15 to 20 operators. With GOSMIC, three operators maximum, suffice," he adds. This optimisation relies on task parallelisation and high-precision algorithms in orbital dynamics. "The numerical and analytical propagators we employ offer excellent trajectory precision," he continues.
Aiko: European innovation serving anomaly detection
The strategic partnership with Aiko, an Italian startup specialising in embedded AI, illustrates Sopra Steria's European cooperation strategy. "Their expertise focuses on anomaly detection in data exchanges with satellites," explains Frouvelle.
This collaboration extends beyond mere technological contribution. "Sopra Steria brings the solidity of a major European group and its operational expertise. Aiko brings us the agility of an innovative startup." This European cooperation logic structures the entire approach against American and Chinese actors, in a sector where technological sovereignty is becoming a major strategic issue.
The partnership also enables exploration of new markets. "This Franco-Italian collaboration opens commercial opportunities and strengthens our European positioning," the expert underlines.
Human-in-the-loop: a competitive advantage
For military space and sensitive operations, Sopra Steria has deliberately chosen to maintain human involvement in the decision-making loop. "Complete automation is neither desirable nor acceptable for operations with geopolitical stakes," states Frouvelle. "Satellite avoidance decisions involve considerations that exceed pure technique."
This requirement extends to on-orbit servicing, those complex satellite rendez-vous missions. "The precision required to approach space debris rotating in all directions still demands human expertise for critical decisions," he explains. A technical capability that few European actors master.
Far from being a limitation, this approach constitutes a differentiator against "black box" solutions. "Our clients understand the value of a transparent architecture where critical decisions remain under human control," observes Frouvelle.
A collaborative vision of space AI
For the future, Sopra Steria is banking on a collaborative approach to artificial intelligence. "AI will find its full utility in federative platforms where it orchestrates data and algorithms from multiple European actors," predicts Frouvelle.
For space surveillance particularly, this approach would optimise telescopes and sensors whilst producing more complete and precise results. "The complexity of space and the importance of the stakes impose controlled autonomisation," he concludes. "It will never be acceptable for a system to decide alone. Human intervention remains indispensable."
In this race towards space automation, Frouvelle and his teams are building technological sovereignty brick by brick, with pragmatism and rigour. An approach that could well prove more robust and sustainable than proprietary solutions.