Article Details

Title · Authors · Abstract · Keywords · DOI · References

Research on Grasping Strategies for Symmetrical Objects with Template Imitation Learning

DING Cheng

(JAKA Robotics Co., Ltd., Shanghai 200241, China)

Section:Intelligent Manufacturing · 2025, Vol. 9, No. 11 · Pages:183-186 · 2096-4706(2025)11-0183-04 · CLC number:TP242.6 · Document code:A · DOI:10.19850/j.cnki.2096-4706.2025.11.034

Abstract

Traditional robot grasping involves pose estimation of objects and generation of grasping strategies. For symmetrical objects, it is necessary to specify the principal axis direction to eliminate pose uncertainty and determine the grasping direction. This research aims to enable non-professionals to achieve robot programming through a single demonstration to reduce development costs. The paper proposes recording the relative coordinate relationships between objects during the demonstration process as a template, and designing symmetry detection and template matching modules based on the rotation group to achieve optimal template matching. To verify the effectiveness of the proposed algorithm, the paper conducts grasping experiments on three categories of objects with different symmetries. The experimental results show that the proposed method is feasible and efficient.

Keywords

robot grasping; demonstration learning; symmetry detection; pose estimation; template matching

References

  1. [1] Chen Haiyong, Li Longteng, Chen Peng, etc 6D pose estimation deep learning network for complex scene point cloud data [J]. Journal of Electronics and Information Technology, 2022, 44 (5): 1591-1601
  2. [2] Wang Haonan, Liu Zhu, Zhang Yiyun, etc A Review of Deep Reinforcement Learning [J]. Frontiers of Information and Electronic Engineering, 2020, 21 (12): 1726-1744
  3. [3] He Huiteng, Zhou Yong, Hu Kaixiong, etc Multi action collaborative grasping strategy for robots based on deep reinforcement learning [J]. Computer Integrated Manufacturing Systems, 2024, 30 (5): 1789-1797
  4. [4] Jin Dongyin, Li Yue, Shao Zhenzhou, etc Design of Reinforcement Learning Reward Function for Robot Arm Trajectory Planning [J]. Computer Engineering and Applications, 2022, 58 (19): 302-308
  5. [5] Cui Ming, Gong Shengrong A review of research on visual guided adversarial imitation learning [J]. Computer Science and Exploration, 2023, 17 (9): 2075-2091
  6. [6] Huang Wenyu, Huang Shengjun Behavior cloning method based on demonstration active sampling [J]. Journal of Nanjing University of Aeronautics and Astronautics, 2021, 53 (5): 766-771
  7. [7] Song Li, Li Dazai, Xu Xin A review of research on inverse reinforcement learning algorithms, theory, and applications [J]. Journal of Automation, 2024, 50 (9): 1704-1723
  8. [8] Huang Yanlong, Xu De, Tan Min A review of imitation learning of robot motion trajectories [J]. Journal of Automation, 2022, 48 (2): 315-334
  9. [9] Zhang Hao, Chen Chengjun, Pan Yong Research on Visual based Industrial Robot Assembly Demonstration Teaching [J]. Machine Tool and Hydraulic, 2023, 51 (19): 25-31
  10. [10] WAKE N,ARAKAWA R,YANOKURA I,et al. A Learning from Observation Framework: One Shot Robot Teaching for Grasp Manipulation Release Household Operations [C]//2021 IEEE/SICE International Symposium on System Integration (SII).Iwaki:IEEE,2021:461-466.

About the authors

Ding Cheng (1990-), male, Han ethnicity, Shanghai native, visual expert, PhD, research direction: robot perception and control technology, robot application and development, human-computer interaction technology.