Onboard AI at Winter Demo
Onboard detection and operator-supervised navigation, demonstrated as part of the BAE Systems Bofors ALMA programme.
Explore the demonstrationFrom the field.
Into the detail.
Follow the field work, explore an engineering question, or get into the code. Reports, practical guides, and research from across Scaleout.
News & field workOut in the field.
Out in the field.
And in the conversation.
Field work, independent reporting, and developments from across Scaleout.
Selected stories
Small AI models help drones identify and prioritise targets
Reporting on Scaleout’s edge AI and drone work.
Resilience comes from designing for disconnection
A discussion of local capability and coordination when connectivity cannot be assumed.
NATO DIANA Innovator
Selected for NATO’s defence innovation programme. Exploring federated learning for demanding field conditions.
Scaleout wins the Security Award
Recognition for Scaleout’s work on edge AI infrastructure, presented by Sweden’s Minister of Defence, Pål Jonson.
Connecting synthetic data and edge AI
A collaboration around data and model improvement for tactical computer vision.
Hardware, software, and edge AI in harmony
A perspective on bringing the parts of an edge AI system together.
Fleet data as a physical security risk
Opinion on the implications of collecting and managing data across vehicle fleets.
More news & coverage28 items
Dagens IndustriNato lyfter Uppsalabolagets AI-tjänst till försvarsarenan
Dagens IndustriNy modell lär upp AI med digitala hembesök utan dataläckage
SVT NyheterScaleout utvecklar autonoma drönarsystem åt BAE i Karlskoga tillsammans med innovationsgrupp
Ultralytics NetworkScaleout cuts model updates from weeks to hours with Ultralytics YOLO
Oracle EcosystemOracle invites Scaleout to its defense ecosystem infrastructure
Oracle Cloud InsightsTactical edge cloud: Powering the Oracle defense ecosystem
Oracle Cloud InsightsDeploying secure federated learning architectures at the tactical edge
Uppsala UniversitetUppsala innovation in cyber security to help NATO DIANA accelerator
Uppsala UniversitetScaleout Systems: Advancing distributed machine learning security
Interspectral HubVinnova Project: Next-generation secure collaborative visualization
Partnership / ScaleoutScaleout and AI Verse Partner to Strengthen End-to-End AI Capability for Tactical Edge Operations
Partnership / ScaleoutAkkodis and Scaleout Accelerate Secure Edge AI
Company news / ScaleoutScaleout secures new investment round to accelerate Cloud-Edge AI
Partnership / ScaleoutScaleout and Flower partner on federated learning solutions
Articles & guides
Work through the ideas.
Technical explanations and examples for the people building, evaluating, and integrating AI systems.
100,000 geographically distributed clients
An evaluation targeting 100,000 simulated clients across ten global regions.
Why AI Misses What Matters in a Storm
Exploring how specialist models can help vision systems adapt to changing weather and operating conditions.
What did this update cost us elsewhere?
How to assess new learning alongside forgetting and performance on earlier tasks.
Federated object detection with YOLO
A worked example connecting object detection models to federated training.
Resilient Edge AI for ISR: Inside Our Swedish Air Force Demonstration
Local model operation and improvement through connectivity changes, demonstrated with the Swedish Air Force.
How do we test what model updates reveal?
Privacy auditing, reconstruction attacks, and the limits of what a test can establish.
More articles & guides27 articles
Federated learning10,000 asynchronous clients
Security · Technical articleFighting back against attacks in federated learning
Data selectionUseful observations at the edge
Implementation · Performance studyPython and C++ clients: a performance deep dive
Research
Questions we’re working on.
Research into how models learn, adapt, and operate across distributed systems. Explore the papers and the projects behind the work.
Read the findings.
Selected work on model adaptation, distributed learning, and privacy.
Quantifying Catastrophic Forgetting in IoT Intrusion Detection Systems
Continual-learning strategies for changing network-attack patterns.
Mixture-of-Experts Models in Vision: Routing, Optimization, and Generalization
A comparison of dense and mixture-of-experts image-classification models.
Practical Feasibility of Gradient Inversion Attacks in Federated Learning
Examining reconstruction attacks under defined access and threat assumptions.
Follow the work.
Research programmes spanning fleets, industrial systems, security, and space.
DREAM
Federated learning for autonomous-vehicle fleets, with a focus on robust training and efficient model updates.
Mixture-of-Experts for fleet intelligence
Specialised models and routing approaches for distributed learning.
TRUSTAM
Federated quality assurance for metal additive manufacturing.
Robust IoT security
Collaborative intrusion detection across IoT operators.
Satellite intelligence
On-orbit training and data filtering for dual-use satellite systems.
More about the research projects
Explore the aims and context behind the work.