Research On Fundamental Algorithms Of Inspection Planning For Area Scanning Robots

Measurement planning for three-dimensional measurement systems is oneof the key problems in the research of automatic inspection of complex surfaces.In this paper, the basic algorithms of measurement planning for area scanningrobots is investigated to lay the foundation for the implementation of automaticinspection of complex surfaces using area scanning robots. The main researchtopic of measurement planning is to compute reasonable viewpoints accordingto the CAD model of the workpiece under test to ensure that the complete pointcloud data of the workpiece are collected. The point cloud will be used toanalyze form error of the workpiece.In this paper, three basic algorithms of measurement planning for areascanning robots, i.e. minimum bounding box algorithm, viewpoint planningalgorithm and visibility analysis algorithm for the surface patch in the field ofview, are investigated. The oriented bounding box for the triangle mesh model iscomputed in this paper. The minimum bounding box of a point set in threedimensions can be efficiently approximated using the convex hull of the pointset. Based on the relationship between convex polyhedron and the Gaussiansphere of a point set, the “Vernier caliper” algorithm is employed to compute theminimum bounding box of according to the moving paths of three orthogonalpairs of “calipers” on the Gaussian sphere.A viewpoint planning algorithm based on the minimum bounding box of thesurface is investigated in this paper. In order to compute reasonable scanningviewpoints, the model is decomposed recursively using the minimum boundingbox of the CAD model until the system parameters of the scanning probe issatisfied. The pose and orientation of each viewpoint is roughly determinedusing the minimum bounding box of each surface patch.Finally, Visibility Maps of a surface is computed using the dualrelationship between Gaussian maps and Visibility maps. Visibility analysis ofthe surface is explored using the Visibility Map to facilitate the determination ofposes and orientations of final viewpoints.All algorithms are implemented in C++language, under the MFCframework, and the final results are displayed using OpenGL APIs.