The project is being implemented under the ‘Research – Innovate’ Action of the ‘COMPETITIVENESS’ Program 2021–2027, with co-financing from Greece and the European Union. Code: ΕΚΠΑΡ01-0076945
The project is being implemented under the ‘Research – Innovate’ Action of the ‘COMPETITIVENESS’ Program 2021–2027, with co-financing from Greece and the European Union. Code: ΕΚΠΑΡ01-0076945
R&D Project · ongoing · 2025–2028
The Next Generation of Smart Fire Safety
Artificial intelligence models to support integrated, interconnected, and more reliable fire detection systems.
AI Models in Service of the Transition to the New Generation of Integrated and Smart Fire Detection Systems
36 months · Greece and EU · medium-sized enterprise
206
Employees
€20.7M
Turnover (2023)
36
Implementation Months
75%
Funding Intensity
The Project
SafeUP is a research and development project exploring the utilization of artificial intelligence in integrated fire detection systems. In collaboration with Olympia Electronics and Atlantis Engineering, methodologies, prototypes, and support tools are being developed.
It aims to reduce nuisance alarms, improve detection reliability, support preventive maintenance, and enable interconnection with smart buildings, BMS, and monitoring services.
By 2028
To establish Greece as a pioneer in the development and export of intelligent fire safety systems, protecting lives, property, and critical infrastructure.
Goals
Development and evaluation of hybrid AI models combining sensors, computer vision, and time series, aiming for high accuracy and reduction of nuisance alarms.
Predictive analytics for early detection of sensor failures and scheduled maintenance, reducing downtime.
Scalable design for integration into BMS, IoT platforms, facility manager applications, and first-responder systems.
Technology and Innovation
SafeUP does not replace the certified functions of existing fire detection systems; it explores ways to support, enrich, and optimize them.
Combination of smoke, temperature, and flame with optical detection and temporal models for high accuracy.
Decision in less than two seconds, even without cloud connection.
Integration into BMS, SCADA, IoT platforms, and building management applications.
Transparent decisions, essential for life-critical systems.
SAFEHUB
SAFEHUB is the proposed hub for data collection, local evaluation, and synchronization. It connects information sources, supports edge processing, and bridges dashboards, BMS, and cloud services.
Functionality will be finalized through specifications, prototyping, and pilot evaluation.
Detectors, panels, sensors, gateways, and future IoT sources.
Local preprocessing for low latency and operation with limited connectivity.
Synchronization with cloud analytics, dashboards, and building management.
Interactive Architecture
Select a stage for details. Filters focus on specific levels.
01
Smoke, heat, flame, gas, cameras, and IoT nodes.
02
Local processing for decision in less than 2 seconds.
03
Computer vision, time series, and anomaly detection.
04
Confidence score and false positive filtering.
05
Predictive maintenance and historical analysis.
06
BMS, dashboards, and notifications.
Sensor level.
EL
Collection of multimodal signals from certified detectors and optical sensors, with bandwidth-adapted sampling.
EN
Multimodal signals from certified detectors and visual sensors, with sampling adapted to available bandwidth.
Edge of network processing.
EL
On-device inference so response does not depend on the cloud, with privacy protection.
EN
On-device inference so response does not depend on the cloud, with improved privacy.
AI Core.
EL
Combination of vision models for smoke/flame with temporal sensor models and anomaly detection.
EN
Vision models for smoke/flame combined with temporal sensor models and anomaly detection.
Explainable AI.
EL
An explanation module that documents each alert — critical for life-critical systems.
EN
An explanation module that documents each alert — essential for life-safety systems.
Cloud level.
EL
Data aggregation for long-term analytics, building digital twin, and maintenance planning.
EN
Aggregated data for long-term analytics, a building digital twin and maintenance planning.
Action level.
EL
Interconnection with BMS, operator applications, first-responder services, and reporting APIs.
EN
Interfaces to BMS, operator apps, first-responder services and reporting APIs.
Test the Technology
Set indicative sensor values and see how a hybrid model weighs the decision. This tool is a demonstration, not a certified detector.
—
Waiting for analysis
Set values and run analysis.
Demonstration of weighted fusion logic. Does not replace certified fire detection equipment (EN 54).
Indicative directional comparison. Final performance will be documented after pilot evaluation.
Typical Approach
Nuisance Alarms
18–35%
Decision Time
45–90 s
Preventive Maintenance
Limited
The Next Generation of Smart Fire Safety
Nuisance Alarms
target: low
Decision Time
target: rapid evaluation
Preventive Maintenance
Target: predictive maintenance
The indicators are indicative targets / scenarios. Final quantitative results will emerge from the pilot evaluation.
Be Part of the Change
We are registering interest from organizations and technical partners who could participate in a future pilot evaluation, according to the project timeline.
Potential participation in an organized pilot test when prototypes allow.
Participation in scenario evaluation for reliability, maintenance, and operational insight.
Provision of real requirements for specification development.
Compliance and Safety-by-Design
SafeUP does not replace current certification requirements. It explores supporting, analytical, and explainability technologies that can be integrated in a safe and documented manner.
Consideration based on the fire detection framework, decision traceability, and reliability.
Recommendations must be evaluated by technicians, managers, and stakeholders.
AI is treated as supportive, not as uncontrolled automation.
Performance conclusions will result from laboratory and pilot evaluation.
Methodology
WP1
Coordination, financial management, reporting, and quality assurance.
WP2
Needs analysis, functional specifications, and design of SafeUP subsystems.
WP3
Use cases, safety protocols, and response procedures.
WP4
Datasets from real and simulated scenarios, annotation, and augmentation.
WP5
Training and optimization of hybrid detection and prediction models.
WP6
Prototypes, integration into Olympia Electronics products, and laboratory testing.
WP7
Installation in real environments and performance evaluation.
WP8
Publications, exhibitions, and preparation for commercialization.
Completion Date: 11/27/2028
Expected Results
80%+
goal of reducing nuisance alarms
Contribution to the UN Sustainable Development Goals
The Consortium
N. LAKASAS P. ARVANITIDIS S.A.
Olympia Electronics contributes industrial know-how, experience in certified fire detection and emergency lighting systems, as well as product, testing, and integration infrastructure.
206 employees · €20.7M turnover
Strategic R&D Partner
Atlantis Engineering supports data analysis methodologies, software, AI models, and digital monitoring tools, in collaboration with Olympia Electronics’ industrial know-how.
AI and Digital Twin
The consortium combines fire safety expertise, industrial experience, and AI technologies, aiming for evaluable and future-exploitable results.
Contact
For pilot installations, collaborations, or press inquiries, please fill out the form. Fields marked with an asterisk are required.
Data is used only to respond to your request. Not used for marketing and not sold.
lmaravelia.projects@olympia-electronics.com · +30 2353 051200 · Aiginio Pierias