Researchers Develop Drones for Safer Confined-Space Inspections
Drones are increasingly being deployed to perform tasks that are either too dangerous or physically difficult for humans. This is especially true for inspections of infrastructure such as power lines, bridges, and tunnels. Given the success of drones in these applications, it is only logical that developers are exploring ways to adapt the technology for confined spaces like ventilation air ducts. These spaces are often inaccessible due to their narrow dimensions, making it difficult to carry out crucial maintenance needed for air quality and ventilation efficiency.
Traditionally, air duct inspections involve a technician removing access panels or vents, then visually checking the interior with a flashlight or pushing a small camera on a flexible rod through the ductwork. In larger systems, sections of duct may even be dismantled to gain visibility. This process is labor-intensive and can take hours to cover even a modest network of vents. Accuracy is limited because only straight runs or areas near openings can be seen, leaving much of the system unchecked. Costs rise quickly due to the specialized labor required, the downtime of HVAC systems, and the potential for missed issues like dirt buildup, leaks, or damage in remote sections.
Compact drones offer a more efficient solution, with the ability to navigate both horizontal and vertical duct sections that humans simply cannot access. However, operating a drone safely in such fragile and confined environments presents its challenges. Pilots must maintain precise control to avoid contact with the infrastructure and prevent accidental damage. Despite these difficulties, some companies and research teams have begun deploying specialized drones equipped with sensors and cameras to inspect air ducts, enabling faster, safer, and more thorough assessments than traditional methods allow.
Building on this momentum, a new research project led by Inria Senior Researcher Jean Baptiste Mouret and his PhD student Thomas Martin at the Centre Inria at the University of Lorraine in France is pushing the boundaries of what drones can achieve in confined spaces. Both researchers are part of the project team HUCEBOT (Human Centered Robotics). HUCEBOT is a joint initiative of Inria and four other French research institutions. Their study, titled Flying in Air Ducts, published in the journal npj Robotics, presents a novel method that allows small quadrotor drones to hover and safely fly within circular air ducts as narrow as 14 inches in diameter, marking a significant advancement in inspection technology.
The study first tackled the question of how airflow generated by a hovering drone inside a duct affects stability. Using a robotic arm outfitted with a force and torque sensor, the researchers measured aerodynamic forces at hundreds of positions inside a duct model. From this data, they produced the first detailed aerodynamic map of a drone hovering in a circular air duct, identifying unstable zones where recirculating air pushes the drone toward the walls as well as a safer position where those forces cancel out.
To help the drone maintain position in this recommended stable zone despite poor visibility and lack of visual cues, the team integrated small laser time-of-flight sensors with a neural network trained on motion capture data. As the team explained, the drone was equipped with “small lasers and artificial intelligence” so it could “stay in the position with the smallest turbulence.” This AI-based localization system enabled the drone to “hover precisely in areas with the least turbulence.”
The compact test drone measures around 7 inches in width and weighs approximately 4.6 ounces. The pilots successfully hovered the drone inside ducts only 14 inches in diameter. In flight tests, the drone managed to hover in place for over two minutes. In a separate trial, it traveled a 10-foot duct corridor, about 5 feet out and back, with lateral error kept under three-quarters of an inch.
The integration of aerodynamic mapping and AI-based positioning signals a leap in confined space drone capabilities. The authors note this achievement paves the way for “new and promising application domains for drones in industrial inspection and public safety.” Future work aims to produce more deployment-ready prototypes with payloads like cameras, thermal imaging systems, or gas sensors to enhance real-world inspection functionality.
As drones continue to prove their value in inspecting complex infrastructure, innovations like this study demonstrate how drone technology is evolving to tackle even the most challenging environments. By enabling safe, precise navigation inside narrow, fragile air ducts, these advances promise to make critical maintenance safer, faster, and more effective, unlocking new possibilities where human access is limited or impossible.
|