CIRCE pioneers a smart traffic light system for Spain powered by NVIDIA

CIRCE Technology Center has developed SAFETRA, a pioneering smart traffic light system that combines artificial intelligence and computer vision to optimize traffic management in real time utilizing NVIDIA Metropolis vision AI platform.

The first pilot will take place in Zaragoza, Spain as a part of the city’s Smart Mobility Strategy.  The project will be officially presented at the Smart City Expo World Congress 2025, to be held from November 4 to 6 in Barcelona.

The initiative promotes the use of technology for safer and more sustainable mobility, while strengthening Zaragoza’s position — recognized by the European Commission as one of the 100 Climate-Neutral and Smart Cities — as a center of innovation and a technology hub in energy and digitalization.

From traditional traffic lights to artificial intelligence

Conventional traffic lights, with fixed timing or manual activation, often cause unnecessary stops, increase congestion, and can lead to unsafe situations for vulnerable pedestrians.

In response to this challenge, CIRCE has developed SAFETRA, a system that uses computer vision and deep learning models to dynamically adapt to the real-world needs of pedestrians and vehicles.

With local real-time processing, the system detects and analyzes the presence of people, and the intention to cross. Based on those inputs, the system automatically adjusts the signal phases according to demand - ensuring a safe and accessible crossing.

One of its main innovations is the ability to resume vehicle flow as soon as no pedestrians are detected within the area of influence, improving traffic flow, reducing waiting times, and minimizing the risk of accidents.

The system can operate autonomously or be integrated into existing urban networks, enabling progressive deployment in complex city environments.

Technical architecture powered by NVIDIA

To ensure real-time performance and energy efficiency, SAFETRA runs on an NVIDIA Jetson device, capable of processing video data 24 hours a day with low power consumption.
 The system also integrates NVIDIA DeepStream SDK, which orchestrates the entire video workflow — from capture to inference — simplifying deployment and improving scalability.

Beyond the technologies already deployed, CIRCE is exploring the integration of video analytics AI agents with cutting-edge AI models, such as NVIDIA Blueprint for video search and summarization (VSS) and Cosmos Reason, a vision-language model (VLM) capable of being prompted with natural language to understand the visual world. These advances expand the system’s contextual understanding of urban traffic, allowing it to interpret complex situations such as crowd formations or atypical pedestrian and vehicle behaviors.

CIRCE’s SAFETRA solution incorporates robust, scalable, and cutting-edge technologies, accelerated by NVIDIA technology, that reinforces CIRCE’s position as a benchmark in applying artificial intelligence to urban mobility.

A commitment to sustainable urban mobility

The implementation of SAFETRA will optimize traffic management, reduce emissions, and enhance road safety.

Aligned with the European Green Deal and the European Strategy for Sustainable and Smart Mobility, the initiative represents a decisive step toward safer, more efficient, and more desirable to live in cities.

“With SAFETRA, we are taking a qualitative leap toward safer and more efficient mobility. The collaboration with the City of Zaragoza demonstrates the potential of technology when it is placed at the service of society”, said Andrés Llombart, General Director of CIRCE.

 Natalia Chueca, Mayor of Zaragoza, emphasized: “This pilot project, like others already implemented in the city, reaffirms Zaragoza’s commitment as a city open to technological innovation that improves quality of life. The true leap forward happens when we bring innovation from the laboratory to the streets, turning them into safe testing spaces”.