Spline Path Based Autonomous Infrastructure Inspection Using an Unmanned Aerial Vehicle

This research demonstrates a novel idea to autonomously conduct infrastructure inspection using an Unmanned Aerial Vehicle (UAV) via simulation and a proof of concept. Currently, inspection is conducted either manually, which is time-consuming and exhausting, or by using the Orbit mode, which results in the UAV either missing to capture views partially/completely of an irregularly structured (e.g., concave hull) target or having improper visibility of the target in the frame. The presented approach executes a part of the infrastructure inspection autonomously to capture a video of the target on a spline path with the orientation of the UAV’s camera perpendicular to the course over ground and also the nadir views of the target. This allows infrastructure inspectors, first responders, insurance authorities to perform normal day-to-day or post-disaster inspections effectively, i.e. easily and with better and more target coverage. The performance of the presented algorithm was evaluated against orbit mode in simulation. From the simulated experiments, it was observed that using the presented algorithm the UAV‘s camera covered 6.2% (p < 0.0027) more of the target’s perimeter, with 10.27 ° (p < 4.95x10^-6) closer to perpendicular observation of the target and 18.49 m (p < 2.8x10^-7) closer to target than in orbit mode. UAV faces the target for about 10.9% (p < 0.0042) duration less using the presented algorithm compared to orbit mode, while traveling 38.79 m (p < 0.0102) less and 12.2 s (p < 0.0828) faster. It was evident that the presented algorithm was better in terms of perimeter coverage, average deviation in the angle of depression w.r.t target, and average observation distance, while Orbit was better in terms of visual stability and video smoothness. A proof of concept for the presented algorithm was implemented and tested in real-world experimental test-cases using a 3DR-Solo. The implementation can be further expanded with modifications as future work, and is expected to outperform orbit mode for autonomous infrastructure inspection in all terms, especially in the case of irregularly structured targets, making the inspection process more effective, i.e. easy and with better and more target coverage