LAST Autonomous Surveillance Tower is a dual-use autonomous ISR and early-warning platform that delivers persistent situational awareness for wildfire prevention, civil protection and remote-area monitoring. The tower runs edge AI on board to detect, classify and geolocate smoke and fire signatures in real time, fusing electro-optical and thermal imagery so that hazards are identified at the sensor rather than at a back-end node. Geo-referenced alerts are pushed directly into incident command, compressing the Observe-Orient-Decide-Act cycle for first responders.
The system is built for 24/7 unattended operation and maintains a continuously updated common operational picture. A communications and fusion node ingests feeds from UAVs and additional sensors and distributes prioritised detections, tracks and imagery over satellite and terrestrial links. Configurations span civilian, marine and multi-domain use, with sensor mosaics of optical, RF, thermal and radar matched to the operational theatre, reducing reliance on human spotters and patrols.
Operational traction: A LAST tower station has been handed over to the Hellenic Ministry of Climate Crisis and Civil Protection to equip the Fire Service for autonomous fire detection in mountainous terrain.
Interoperability and dual use: Outputs (video, metadata, alerts and tracks) are designed for ingestion into civil C2 and EOC platforms. Where defence-grade interoperability is required, they can be interfaced via gateways aligned with NATO ISR data exchange practices (for example STANREC 4869 / SAPIENT for sensor tasking and STANAG 4609 / 4676 for motion imagery and track products), subject to end-user integration choices.
Capability gaps and needs addressed (DIREKTION / IFAFRI):
Real-time detection, monitoring and analysis of threats and hazards
Remote acquisition of information
Integration of information into incident command operations
Actionable intelligence for responder decision-making
Contact
The objective of this proposal is to field the LAST Autonomous Surveillance Tower as a near-to-deployment early-warning and situational-awareness capability for civil protection and fire services operating in remote, hard-to-access areas. LAST is offered as an operational tool, not a research prototype: it detects emerging wildfire indicators early, maintains persistent surveillance over priority zones, and delivers decision-ready alerts that strengthen incident command from preparedness through initial attack.
It does this by combining electro-optical and thermal sensing with edge AI to autonomously detect, classify and geolocate smoke and fire signatures, and to publish geo-referenced alerts and an evolving operational picture for command centres. The tower is built for austere, unattended operation, with autonomous power and resilient communications backhaul ensuring that prioritised detections and imagery keep flowing even where access is constrained and human lookouts or patrols cannot provide continuous coverage. Outputs are designed for direct integration into existing EOC and incident command workflows, supporting a shared common operational picture and information exchange across agencies.
The proposal is highly relevant to Disaster and Resilience societal gaps. Wildfire growth is rapid, information is typically incomplete in the early stages, and responder safety depends on timely, trusted situational awareness. By enabling real-time hazard detection, remote acquisition of information, integration into incident command workflows and actionable intelligence for dispatch and coordination, LAST supports faster decisions and safer resource allocation. Its modular deployment concept scales from single sites to networked regional coverage as needs expand. LAST is aligned with the Awards' focus on mature, demonstrable solutions, including operational deployment in Greece and a handover to the Hellenic Ministry of Climate Crisis and Civil Protection to equip the Fire Service.
LAST's innovation lies in turning persistent, unattended sensing into decision-grade early warning at scale. Rather than relying on continuous human monitoring or heavyweight active-surveillance infrastructure, LAST couples electro-optical and thermal sensing with on-tower edge computing to automate detection, tracking, classification and geolocation. A key differentiator is the ability to assign geographic coordinates to image information and classify events of interest without active ranging sensors such as radar or laser rangefinders, a passive approach well suited to wide-area forest protection and other disaster-risk-prone environments where active emitters are impractical or undesirable. The system presents a unified picture that fuses feeds from multiple towers and sensors into a single, intuitive, mission-oriented map view, reducing operator workload while converting raw imagery into geo-referenced alerts and an evolving common operational picture.
Innovation is also delivered through architecture and deployment. LAST is modular and can fuse inputs from UAVs and other sensors through an associated communications device. Resilient GSM and satellite connectivity supports continuity of service in degraded conditions, while solar-based power with battery storage enables 24/7 operation with minimal logistics and rapid installation without heavy machinery. Outputs ingest cleanly into existing emergency management and C2 systems, shortening the time from detection to dispatch.
LAST resolves a critical operational gap faced by first responders: the need for continuous, trusted situational awareness over wide high-risk areas with austere terrain and limited communication coverage, where minutes determine whether an incident escalates into a major disaster. In wildfire operations, early detection is routinely constrained by limited human lookout coverage, finite patrol capacity, delayed reporting and incomplete visibility in complex terrain. These constraints compress decision time, increase uncertainty for incident command and expose crews to greater risk.
By providing persistent, unattended monitoring and automated early warning, LAST shortens the time from ignition to detection and from detection to decision. It continuously surveys designated zones and produces geo-referenced alerts and supporting imagery that integrate into command workflows, giving incident command a clearer and timelier operational picture while the incident is still containable. This enables earlier tasking of resources, better prioritisation of competing alerts and safer planning of access routes and staging areas based on where the hazard is actually developing. In practice, it reduces reliance on chance visual sightings and decreases the workload on personnel who would otherwise need to watch multiple camera feeds or conduct repeated patrols.
Coverage is scalable. A single site can protect a high-value area, while networked towers provide layered regional surveillance during peak risk periods, with autonomous power and resilient backhaul keeping information flowing through infrastructure disruption.
Delian's team combines the autonomy and delivery expertise required for a safe, robust, real-time situational-awareness system. The core team brings depth in AI, machine perception, sensor fusion, calibration and state estimation. Its members come from organisations including Palantir, Meta, Uber and Apple, and from globally esteemed universities. The factory in Athens is staffed with experienced engineers and machinists, enabling rapid manufacturing, prototyping and field deployment.
Delian's team combines the autonomy and delivery expertise required for a safe, robust, real-time situational-awareness system. The core team brings depth in AI, machine perception, sensor fusion, calibration and state estimation. Its members come from organisations including Palantir, Meta, Uber and Apple, and from globally esteemed universities. The factory in Athens is staffed with experienced engineers and machinists, enabling rapid manufacturing, prototyping and field deployment.