The team focused on quadrotor drones, which are small enough to traverse narrow ducts. However, the confined geometry causes their airflow to recirculate and destabilize the drone mid-flight. These unpredictable aerodynamic forces severely limit flight stability and accuracy, which are critical in inspection missions where collisions must be avoided.
To address this, the researchers created a comprehensive map of aerodynamic forces acting within circular air ducts. Using a robotic arm and a force/torque sensor, they collected measurements at hundreds of positions inside the duct. This allowed them to pinpoint dangerous zones where airflow pushes drones toward the walls and identify safer areas where turbulent forces balance out.
To maintain stable flight in these safe zones - especially in ducts that lack light or visual features - the team developed a localization system that combines laser sensors with a neural network trained on motion capture data. This AI-based system enables the drone to hover precisely in areas with the least turbulence, significantly reducing the risk of collision.
This innovation paves the way for broader industrial and safety applications. Future prototypes will incorporate payloads such as cameras, thermal imagers, or gas sensors for enhanced functionality in inspection tasks.
The project was spearheaded by Inria Senior Researcher Jean-Baptiste Mouret and PhD student Thomas Martin of the HUCEBOT team, a collaboration between CNRS, Inria, and Universite de Lorraine. The study also involved Aix-Marseille Universite and multiple research laboratories under CNRS and Universite de Lorraine, showcasing the depth of interdisciplinary cooperation among French scientific institutions.
Research Report:Flying in air ducts
Related Links
Centre Inria de l'Universite de Lorraine
UAV News - Suppliers and Technology
| Subscribe Free To Our Daily Newsletters |
| Subscribe Free To Our Daily Newsletters |
