Closing the Loop: AI-Driven Vision Integration for Next-Generation Unmanned Systems

Purpose and goal

The project aims to demonstrate an integrated pipeline from operational requirements to deployed AI perception on unmanned drone systems. The consortium develops and validates a workflow that converts real-world operational data into synthetic training datasets, deploys optimized AI models on edge hardware, and integrates the solution on drone platforms in a relevant operational environment, advancing the system from TRL 6 to TRL 7.

Expected effects and result

Expected results include a validated prototype demonstrating AI-driven perception deployed on unmanned drone platforms in a relevant operational environment. The project is expected to strengthen Swedish capabilities in autonomous aerospace systems, establish long-term international collaboration with operational end users, and create a scalable pipeline for rapid adaptation of AI perception models from operational data to edge deployment.

Planned approach and implementation

The project is implemented through collaboration between Swedish industry partners and the Ukrainian end-user partner SD Group. SCAILAB develops synthetic training datasets and AI perception models, Synclair Vision deploys optimized models on the MOSS edge platform, and Defilade coordinates operational integration and end-user feedback. The solution is integrated and validated on drone hardware through iterative testing in a relevant operational environment.