LINA targets safety bottleneck holding back AI-powered drones
Switzerland’s LINA platform aims to close the safety gap slowing AI-powered drone deployment.
Unmanned aerial systems are moving into logistics, mobility, public safety, agriculture and disaster response, but most still depend on humans for critical decisions. AI is improving flight control, perception and airspace management, yet aviation rules demand evidence that adaptive systems can operate safely before they are allowed into regulated airspace.
LINA, the Shared Large-Scale Infrastructure for the Development and Safe Testing of Autonomous Systems, is presented as an integrated Swiss platform for experimentation, validation and regulator-aware development. The academic-industry effort brings together testing facilities, processes and expertise designed to evaluate autonomous systems under realistic and safety-critical conditions across technology readiness levels.
The need is driven by a mismatch between AI development and aviation certification. Drone algorithms can advance in months, while certification practices and regulatory adaptation often move over years, leaving a gap between laboratory performance and operational readiness.
Key technologies include reinforcement learning, model predictive control, computer vision and large language or foundation models. Each can help UAVs navigate, plan, sense obstacles and make decisions, but each also raises hard questions about reliability, explainability, accountability and proof of safe behavior.
LINA’s central role is to turn testing into usable evidence. If such infrastructure can support repeatable validation and regulatory trust, AI-enabled drones will have a clearer path from research prototypes to real-world operations, including higher-autonomy and beyond-visual-line-of-sight missions.