GATSBI: An Online GTSP-Based Algorithm for Targeted Surface Bridge Inspection

We study the problem of visual surface inspecting of a bridge for defects using an Unmanned Aerial Vehicle (UAV). The geometric model of the bridge is unknown beforehand. We equipped the UAV with a 3D LiDAR and RGB camera to build a semantic map of the environment. Our planner, termed GATSBI, plans in an online fashion a path to inspect all points on the surface of the bridge. The input to GATSBI consists of a 3D occupancy grid map created online with LiDAR scans. Occupied voxels corresponding to the bridge in this grid map are semantically segmented out. Inspecting a bridge voxel requires the UAV to take images from a desired viewing angle and distance. We then create a Generalized Traveling Salesperson Problem (GTSP) instance to cluster candidate viewpoints for inspecting the bridge voxels and use an off-the-shelf GTSP solver to find the optimal path for the given instance. As the algorithm sees more parts of the environment over time, it replans the path. We evaluate the performance of our algorithm through high-fidelity simulations conducted in AirSim. We compare the performance of this algorithm with a frontier exploration algorithm. Our evaluation reveals that targeting the inspection to only the segmented bridge voxels and planning carefully using a GTSP solver leads to more efficient and thorough inspection compared to the baseline algorithm.