Lidar-VIO Based Mapping of Elevator Shaft with UAV for Verification of Spatial Dimensions

Current methods utilized for the inspection of spatial dimensions of elevator shafts currently entail the need to take manual measurements at various levels of the shaft. Utilizing a UAV to collect spatial dimensions has potential to allow inspectors to complete their jobs in a safer manner and to boost the accuracy and consistency of inspection procedures. The study presented tests the feasibility of a designed UAV based system that hosts a miniaturized simultaneous localization and mapping (SLAM) scheme for usage of generating a point cloud of an elevator shaft environment for spatial dimension inspection. Secondary usages of the scheme are further demonstrated for cases where the scheme is employed while not mounted on a UAV. Average horizontal measurement accuracy for the system employed in flight is found to be 6.57 cm, and there is little difference in accuracy for mapping performed not in flight. Visual inertial odometry is investigated for pose estimation reliability in dark environments and environments with little structural information to process. Further sensor characterizations are performed to characterize properties of point clouds generated of a laboratory, a hallway, and an elevator shaft.