A novel surface segmentation approach for robotic manipulator-based maintenance operation planning

This paper presents a novel approach to segmenting a three-dimensional surface map by considering the task requirements and the movements of an industrial robot manipulator. Maintenance operations, such as abrasive blasting, that are performed by a field robot manipulator can be made more efficient by exploiting surface segmentation. The approach in this paper utilises an aggregate of multiple connectivity graphs, with graph edges defined by task constraints, and graph vertices that correspond to small, maintenance-specific target surfaces, known as Scale-Like Discs (SLDs). The task constraints for maintenance operations are based on the characteristics of neighbouring SLDs. The combined connectivity graphs are analysed to find clusters of vertices, thus segmenting the surface map into groups of related SLDs. Experiments conducted in three typical bridge maintenance environments have shown that the approach can reduce garnet usage by 10%–40% and reduce the manipulator joint movements by up to 35%.

► Overlapping target generation algorithm facilitates manipulator-based coverage tasks
► Considers task requirements and robot manipulator movements to segment a 3D map
► Improves autonomous abrasive blasting operations by exploiting surface segmentation
► Results show reductions up to: 40% in time; and 35% in manipulator joint movement.