UAV tunnel inspection

Tunnel inspection is critical for maintaining infrastructure safety and reliability but poses significant challenges. Traditional methods expose human workers to hazardous conditions and are time-consuming. This project leverages unmanned aerial vehicles (UAVs) to autonomously inspect tunnels, offering a safer, more efficient alternative.

The primary objective is to enable UAVs to autonomously navigate dynamic tunnel environments, even under poor lighting conditions, and perform detailed inspections to detect defects or deficiencies in tunnel walls. The project addresses challenges such as unpredictable structural changes, limited GPS availability, and low visibility by integrating advanced navigation algorithms and 3D reconstruction technologies. Simulations demonstrated that the UAV achieves high-precision 3D modeling of tunnel walls, with a maximum error of only 5 cm compared to ground truth data. This accuracy ensures reliable detection of structural issues, which is critical for proactive maintenance and public safety.

Key Contributions and Outcomes:

• Social Value: Enhances worker safety by eliminating the need for human presence in hazardous environments and reduces inspection time, thereby optimizing resource use.
• Effectiveness: Achieved significant results in simulation, demonstrating the feasibility of the solution.
• Role/Contribution: My role included designing and implementing the navigation algorithms, conducting 3D reconstruction simulations, and validating the UAV’s accuracy against ground truth data.

Conclusion and Future Work: This project demonstrates the potential of UAVs to transform tunnel inspections by improving safety, accuracy, and efficiency. Future work includes real-world testing to refine algorithms and enhance adaptability to even more complex tunnel environments.

More details can be found here