PAPER P16 · 2017
Autonomous Navigation and Mapping for Inspection of Penstocks and Tunnels With MAVs
Özaslan, T., Loianno, G., Keller, J., Taylor, C. J., Kumar, V., Wozencraft, J. M., Hood, T. — IEEE Robotics and Automation Letters 2(3):1740-1747
Abstract read Citations: openalex: 159 · crossref: 140 · semanticScholar: 142 (2026-10-22)
Develops estimation, control, navigation and mapping so a multirotor with onboard sensing can autonomously inspect dam penstocks and tunnels without GPS or external cameras. Penstocks are dark, featureless generalized cylinders, where range sensors alone are inadequate and cameras struggle in darkness. Fusing an IMU, lidar and cameras gives 6-DOF pose and velocity; this letter details the range-based part, which only assumes a generalized-cylinder model and tolerates shape changes. It was validated inside a penstock at Center Hill Dam, Tennessee.
Key findings
- 6-DOF pose and velocity estimated by fusing IMU, lidar and cameras
- Range-based estimator needs only a generalized-cylinder model and is robust to tunnel shape changes
- Authors report the first autonomous navigation and mapping in a penstock without external infrastructure (Center Hill Dam, TN)
Limitations
- Abstract only reviewed
- Estimator tailored to cylinder-like geometry; vision part described in earlier work
- Shared control (human in the loop) was part of the demonstration