Abstract：Aiming at the unstructured grasping environment in real life, an autonomous grasping method based on the basic objects composition of irregular objects is proposed. The key to autonomous grasping of robots lies not only in the recognition of object types, but also in the good grasping of object's shapes (such as the composition of shape primitives). The irregular complex object is simplified as the composition of some simple objects. The 3D data points of the grasped object are segmented into the main body and the branch part based on the mesh segmentation using feature point and core extraction (MFC) algorithm, and the parts are fitted to a sphere, an ellipsoid, a cylinder or a parallelepiped according to an optimal fitting algorithm. Based on the simplified result, the grasping posture is constrained, and then the grasping training of the simplified object is completed to obtain the optimal grasping frame, thereby realizing the autonomous grasping of the unknown object. Finally, the method proposed realizes the grasping operation with an accuracy of 93.3% on the Baxter robot. The experimental results show that the method can be applied to unknown irregular objects of different shapes and postures with strong robustness.
 Saxena A, Driemeyer J, Ng A Y. Robotic grasping of novel objects using vision[J]. International Journal of Robotics Research, 2008, 27(2):157-173.
 Hsiao K, Chitta S, Ciocarlie M, et al. Contact-reactive grasping of objects with partial shape information[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. Piscataway, USA:IEEE, 2010:1228-1235.
 喻群超,尚伟伟,张驰.基于三级卷积神经网络的物体抓取检测[J].机器人,2018,40(5):762-768.Yu C Q, Shang W W, Zhang C. Object grasp detection based on three-level convolution neural network[J]. Robot, 2018, 40(5):762-768.
 Le Q V, Kamm D, Kara A F, et al. Learning to grasp objects with multiple contact points[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2010:5062-5069.
 Lenz I, Lee H, Saxena A. Deep learning for detecting roboticgrasps[J]. International Journal of Robotics Research, 2014, 34(4-5):705-724.
 LeCun Y, Bengio Y, Hinton G. Deep learning[J]. Nature, 2015, 521(7553):436-444.
 Suzuki K. Overview of deep learning in medical imaging[J]. Radiological Physics and Technology, 2017, 10(3):257-273.
 Huebner K, Ruthotto S, Kragic D. Minimum volume bounding box decomposition for shape approximation in robot grasping[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2008:1628-1633.
 George P L, Borouchaki H. Delaunay triangulation and meshing:Application to finite elements[M]. London, UK:Kogan Page, 1998.
 Katz S, Leifman G, Tal A. Mesh segmentation using feature point and core extraction[J]. Visual Computer, 2005, 21(8-10):649-658.
 Kurant M, Markopoulou A, Thiran P. On the bias of BFS[C]//22nd International Teletraffic Congress. 2010. DOI:10.1109/ITC.2010.5608727.
 Au O K C, Zheng Y Y, Chen M L, et al. Mesh segmentation with concavity-aware fields[J]. IEEE Transactions on Visualization and Computer Graphics, 2012, 18(7):1125-1134.
 Katz S, Tal A. Hierarchical mesh decomposition using fuzzy clustering and cuts[J]. ACM Transactions on Graphics, 2003, 22(3):954-961.
 Goldfeather J, Interrante V. A novel cubic-order algorithm for approximating principal direction vectors[J]. ACM Transac-tions on Graphics, 2004, 23(1):45-63.
 Fang Z G, Jiang J X, Xu J, et al. Efficient collision detection using bounding volume hierarchies of OBB-AABBs and its application[C]//International Conference on Computer Design & Applications. Piscataway, USA:IEEE, 2010:V5242-V5246.
 El-Khoury S, Sahbani A. A new strategy combining empirical and analytical approaches for grasping unknown 3D objects[J]. Robotics and Autonomous Systems, 2010, 58(5):497-507.
 Lian Z, Godil A, Bustos B, et al. SHREC'11 track:Shape retrieval on non-rigid 3D watertight meshes[DB/OL]. (2011-04-01)[2018-12-20]. https://www.itl.nist.gov/iad/vug/sharp/contest/2011/NonRigid/data.html.
 Cherkassky V, Ma Y Q. Practical selection of SVM parameters and noise estimation for SVM regression[J]. Neural Networks, 2004, 17(1):113-126.
 Johns E, Leutenegger S, Davison A J. Deep learning a graspfunction for grasping under gripper pose uncertainty[C]//IEEE/RSJ International Conference on Intelligent Robots and Sys-tems. Piscataway, USA:IEEE, 2016:4461-4468.
 Rusinkiewicz S, Levoy M. Efficient variants of the ICP algori-thm[C]//3rd International Conference on 3-D Digital Imaging and Modeling. Piscataway, USA:IEEE, 2001:145-152.
 Torr P H S, Zisserman A. MLESAC:A new robust estimator with application to estimating image geometry[J]. Computer Vision and Image Understanding, 2000, 78(1):138-156.