Path planning on known 3D surfaces for industrial robotic applications

Path planning on known three-dimensional surfaces is a fundamental problem in industrial robotics, particularly in surface-treatment applications such as spray painting, polishing, inspection, and coating. In these scenarios, robotic trajectories must ensure complete surface traversal while respecting geometric constraints imposed by the object shape, tool and process requirements. This thesis investigates trajectory generation strategies for industrial robotic applications operating on known 3D surfaces. Two complementary planning paradigms are studied. The first part focuses on direct 3D surface planning, where trajectories are computed directly on the object geometry. Surface segmentation and geometry-aware strategies are employed to generate paths that adapt to local surface characteristics, making this approach well suited for general coverage. The second part addresses industrial scenarios in which the path must follow structured and repeatable motion patterns. In such cases, the surface planning problem is reformulated in a two-dimensional parametric domain. To enable systematic trajectory generation on complex surfaces, this thesis investigates and compares surface mapping strategies that transform three-dimensional geometries into planar representations, suitable for path design and analysis. Trajectories are designed in the parametric space, where pattern definition is more flexible, and subsequently projected back onto the 3D surface. Structured paths such as contour-parallel spirals are generated, and the approach is further extended to enable the mapping of user-defined planar patterns onto complex 3D surfaces. Overall, the thesis provides a geometry-aware planning framework for robotic trajectory generation on known 3D surfaces, supporting both general surface coverage and application-driven patterned paths. The proposed methods aim to improve the flexibility and suitability of robotic surface-treatment trajectories in industrial settings.