视觉抓取方案

MoveIt + Grasp Pose Generator - 适用于简单几何物体抓取 - MoveIt Grasps

anygrasp - 基于深度学习的抓取姿态估计方法，适用于复杂物体和多样化的抓取场景，由graspNet数据集训练得到 - anygrasp - anygrasp github - anygrasp配置

GGCNN - 输入Depth map，适用于2D平面抓取 - Closing the Loop for Robotic Grasping: A Real-time, Generative Grasp Synthesis Approach - GGCNN复现

RGI-NNet

VLM-Grasping

OADS - A Real-time Affordance-based Object Pose Estimation Approach for Robotic Grasp Pose Estimation

SuperQ-Grasp - 基于超二次曲面（superquadrics）的抓取姿态估计方法，适用于移动操作中的较大物体 - SuperQ-GRASP: Superquadrics-based Grasp Pose Estimation on Larger Objects for Mobile-Manipulation - SuperQ-Grasp Webpage