一种三维视觉引导的机械臂自适应抓取方法及系统
By fusing global position and local geometric information into a 3D vision-guided robotic arm grasping method, calculating surface normals and principal curvatures, and generating a grasping compatibility matrix, this method utilizes attention mechanisms and pose regression networks to address the problem of insufficient integration of physical constraints in existing technologies, thereby achieving more accurate and feasible grasping pose evaluation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- LUOYANG INST OF SCI & TECH
- Filing Date
- 2026-05-09
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to integrate the physical constraints of robots into the feature extraction and relational reasoning processes of deep learning, resulting in inaccurate pose assessment when facing objects with complex curved surfaces or edge features, and an inability to achieve deep coupling between geometric perception and feasibility.
By acquiring the 3D point cloud data of the target object, extracting local geometric features, and fusing global position information with neighborhood geometric information, calculating surface normals and principal curvatures, generating a grasping compatibility matrix, and using an attention mechanism and pose regression network to output the optimal grasping pose.
This improved the accuracy and success rate of grasping pose, ensuring the physical feasibility and stability of the grasping solution.
Smart Images
Figure CN122143066B_ABST