SINTRA demonstrates progress towards smarter, integrated critical infrastructure security
On 5 October, the Dutch consortium of the ITEA project SINTRA – Security of Critical Infrastructure by Multi-Modal Dynamic Sensing and AI – organised a symposium at Eindhoven University of Technology (TU/e). Representatives from Dutch ports, infrastructure organisations and the police were invited to see how the project’s multi-modal sensing and AI technologies are progressing towards a new approach to protecting critical infrastructure.
Critical infrastructure sites such as ports and logistics hubs need to monitor large areas and substantial amounts of sensor and camera data. The challenge is not only detecting events, but identifying which ones genuinely require attention. SINTRA addresses this by combining different sensing technologies with AI to create stronger situational awareness and recognise complex or unusual behaviour.
From detection to relevant alerts
During the symposium, three demonstrations showed how different sensor inputs can be combined with AI to identify potentially suspicious activities. AI models have, for example, been trained to recognise sequences of behaviour around trucks and port areas, such as exchanging bags, returning to a vehicle in an unusual way or removing and placing objects.
By combining information from multiple sensors, the SINTRA system can identify patterns that may be difficult to recognise from individual camera feeds alone. Advanced reasoning using Vision-Language Models (VLMs) then adds context to detected events and helps determine which ones deserve further investigation.
The results demonstrated a substantial reduction in the number of alerts requiring human attention. In one scenario, more than 200 initial alarms were reduced to 22 relevant events. In a harbour use case, around 1,000 daily alerts were narrowed down to 22 after detection, labelling and advanced reasoning. Detection accuracy in the current demonstrations is approximately 85%.
This could allow security personnel to spend less time monitoring large numbers of screens and alerts and more time assessing situations that may require action.
Towards integrated critical infrastructure security
The demonstrations also showed SINTRA's broader ambition of connecting different stages of physical security into an integrated chain; sensors detect unusual behaviour, AI analysis the situation and associated objects or substances, and the resulting information can support an appropriate response, including intervention by drone, security services or the police.
Potential benefits include earlier detection of organised crime, fewer false alarms, improved employee safety and better decision support. Bringing relevant information together in one dashboard could also provide security teams with a clearer overview of what is happening across a site.
Stakeholders at the symposium responded positively to this innovation, with one port representative describing such an integrated solution as the type of security capability the sector would ultimately like to achieve.
Promising results with further integration ahead
The symposium also highlighted the steps needed to move from a research prototype to operational deployment. Integration with existing infrastructure requires considerable effort and depends on factors such as available training data, the specific activities to be detected and the computational cost of advanced VLMs. Communication between sensors, environmental conditions, vessel recognition and regulatory requirements also remain important considerations.
Nevertheless, several individual SINTRA technologies are already available or approaching the market. With further investment and site-specific integration, the project results offer a promising basis for more intelligent and integrated security of critical infrastructure.
SINTRA symposium 2026 impression
The Dutch SINTRA consortium consists of TU/e, Avular, IQSIGHT, MantiSpectra, Omines, Port of Moerdijk, Safecity, ViNotion and Sorama.
