Structural Inspection Planning for Mobile Robots

In this report we present a pipeline for static coverage planning of known objects, which is an important task in the field of mobile robot based inspection. We analyse the main components of the Structural Inspection Planner [1] and embed an improved implementation into a autonomous flight pipeline for UAVs. Triangle mesh models serve as input for an initial viewpoint sampling. Inspection quality and path length are optimized by formulating the viewpoint sampling as constraint QP. We thoroughly evaluate the ROS-based inspection pipeline on synthetic and real models using a Gazebo simulation. Our experimental evaluation shows that while an efficient inspection trajectory could be generated for most of the tested models, the result is very dependent on regular and well formed input models.