VISER - Visual Inspection of airspace for air traffic safety and societal SEcuRity

Important results from the project

The project was able to demonstrate the overarching goal, that a low-cost vision-based sensor network can be used to detect drone intrusions in protected airspace such as airports and power plants. The project successfully demonstrated that all steps in distributed image acquisition, object detection, and edge computing are feasible. It also showed that AI-based drone classification using trained deep-learning models is a promising approach going forward.

Expected long term effects

The project hopes that Saab or another actor will continue to build on the experiences gained in this project to strengthen protection against drones. An important lesson from the project is that distributed, cost-effective camera systems with adequate processing capacity represent a promising way forward, especially as developments in technology and AI continue to advance rapidly.

Approach and implementation

The project was organized into two parallel workstreams, led by RISE and Saab respectively. Collaboration between RISE and Saab worked very well, with coordination meetings held every two weeks. During the project, RISE lost its two main developers, which caused a temporary loss of knowledge and resulted in delays before new team members could regain momentum. The Saab team remained stable throughout. We were also able to perform multiple field tests during the project.