Autonomous Model-Based Inspection Planning for 3D Visual Coverage Tasks

The visual inspection planning of 3D objects consists of the problem of finding suitable viewpoints. While complete visual coverage of the object surface is required, not too many viewpoints should be needed. In this paper a view-planning algorithm for a camera-equipped Mobile Industrial Robot Arm (MIRA) is proposed. The MIRA is mounted on a mobile platform, which can move freely on the ground around the object being inspected. The proposed model-based approach finds the required platform positions and the corresponding viewpoints to obtain a full visual coverage. Random sampling and a potential-field method are used to generate viewpoint and platform position candidates. The viewpoint and platform position planning both are formulated as a Set Covering Problem (SCP) afterwards. The effectiveness of selected algorithms to solve the SCP for this application is compared through a number of computational tests. The overall applicability of the proposed approach is demonstrated in a simulated environment.