Visual servo assembly method for component docking

This paper adopts a spatial discretization model and grid model in traditional path planning methods to discretize the posture mechanism's workspace into a gridded array of positions with various posture angles. A collision detection model using octree segmentation is then applied to detect collisions and document the first safe posture encountered. This approach establishes an enhanced path planning method that separates collision detection and path planning, specifically designed for large component docking. It boosts posture adjustment efficiency while maintaining safe docking operations. Additionally, a visual servo posture control flow is proposed for large component docking based on a traditional servo control system model, introducing evaluation metrics and performance indicators. A visual-guided posture adjustment system is developed for ship sections and validated through experiments. The system architecture, hardware selection, and establishment are discussed in detail. Hand-eye calibration of the 3D camera is performed to unify the docking system's internal coordinate system. Path planning, posture adjustment, and visual-guided docking experiments are then conducted to verify the system's reliability under various poses.