Robot perception planning for industrial inspection

A perception planning method is proposed in this paper to automatically determine sensor positions for a robot in industrial inspection tasks. We consider that there exists a target model in the robot system and the vision sensor needs to move from one pose to another around the target to observe many geometrical features. Multiple 3D images are then taken from different vantage viewpoints and integrated together to analyze the manufacturing quality of the industrial part. Such a perception-planning problem involves determination of the least sensing operations, the optimal spatial distribution sensor placements, and a shortest path through these viewpoints. During the planning, object features are sampled as individual points to reduce the computational complexity. The optimal sensor distribution and robot execution path are determined by objective functions. Experiments are also carried out to demonstrate the proposed method.