Unmanned Aerial Systems (UAS) have become essential assets for situational awareness and tactical support in modern defence operations. The current generation of UAS relies heavily on Global Navigation Satellite Systems (GNSS) for positioning and mission execution. However, in contested environments, adversaries try to take advantage of such vulnerability by employing electronic warfare tactics, including jamming and spoofing, to deny satellite access. Such "non-permissive environments" can render standard remotely piloted systems ineffective, limiting the operational capability of defence forces to locate targets or surveil areas of interest safely. There is a pressing need for systems that can operate autonomously and reliably without external navigation signals or continuous data links to a ground station.

The SWARMER project addresses this capability gap by developing a robust framework for autonomous drone swarms designed specifically for GNSS-denied environments. Unlike traditional systems that run on a single UAS or process data centralized on swarm, SWARMER emphasizes distributed and decentralized ways along with multi-sensor fusion (LiDAR, visual, thermal and inertial sensors) to be used in deep learning methods in order to achieve autonomous navigation, mission completion and decision-making. The goal is to deliver a scalable system where drones operate as a cohesive flock, utilizing decentralized strategies to share tasks and situational awareness. The main objectives of the SWARMER project include:

• To design and develop a robust autonomous navigation system that fuses data from heterogeneous sensors (Visual, LiDAR, Inertial) to operate without GNSS.
• To design and integrate high-performance Edge AI computing payloads capable of supporting heavy computational loads for navigation and AI processing within strict Size, Weight, and Power (SWaP) constraints.
• To develop decentralized swarming behaviors and Vehicle-to-Vehicle (V2V) communication protocols, allowing the swarm to self-organize, heal from dropouts, and allocate tasks without a central controller.
• To implement Cooperative Simultaneous Localization and Mapping (C-SLAM), enabling the swarm to collectively build a 3D map of the battlefield and localize themselves relative to one another.
• To employ advanced AI/ML algorithms for real-time object detection, threat identification, and dynamic path planning in cluttered environments.
• To design a comprehensive system architecture for a resilient UAS swarm capable of robust operation in GNSS-denied and electronically contested environments.
• To validate the integrated system through participation in four annual technological challenges, testing the swarm in realistic, increasing complex scenarios (indoor, outdoor, and hybrid) to ensure operational readiness.

MKLAB is responsible for designing and developing the core perception and localization engine of the SWARMER system, with a focus on cutting-edge computer vision and SLAM technologies. This involves creating innovative visual and LiDAR-based SLAM algorithms to enable precise localization and mapping in complex environments. MKLAB will also develop advanced computer vision tools for real-time object detection and avoidance, terrain analysis, and feature tracking, which are essential for autonomous navigation and environmental understanding. By exploiting state-of-the-art AI architectures, MKLAB aims to enhance the perception capabilities, providing the swarm with a reliable and accurate spatial awareness crucial for mission success in challenging operational scenarios.