首页|Research from University of Tehran Yields New Data on Robotics (Multi-Modal Robu st Geometry Primitive Shape Scene Abstraction for Grasp Detection)
Research from University of Tehran Yields New Data on Robotics (Multi-Modal Robu st Geometry Primitive Shape Scene Abstraction for Grasp Detection)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Fresh data on robotics are presented i n a new report. According to news reporting originating from Tehran, Iran, by Ne wsRx correspondents, research stated, "Scene understanding is essential for a wi de range of robotic tasks, such as grasping. Simplifying the scene into predefin ed forms makes the robot perform the robotic task more properly, especially in a n unknown environment." The news editors obtained a quote from the research from University of Tehran: " This paper proposes a combination of simulation-based and real-world datasets fo r domain adaptation purposes and grasping in practical settings. In order to com pensate for the weakness of depth images in previous studies reported in the lit erature for clearly representing boundaries, the RGB image has also been fed as input in RGB and RGB-D input modalities. The implemented architecture is based o n the Mask R-CNN network with a backbone of ResNet101. By using RGB and RGB-D im ages as input, the proposed approach has thus improved the segmentation Dice sco re over primitive shape abstraction by 3.73% and 6.19% , respectively. Moreover, in order to improve and evaluate the robustness of the model to occlusion and a variety of primitive shapes and colors that may occur in the scene, different versions of simulation-based datasets are generated usin g the Coppeliasim simulator. Additionally, a real-world primitive shape abstract ion dataset is created to make the model more robust in more complex objects and real-world experiments. To further generalize the model to apply to a wider ran ge of objects, new primitive shapes, such as cones, and both filled and hollow t ypes of each primitive shape, are considered. Subsequently, the point clouds of the segmented parts are generated, and the ICP algorithm is used to derive the 6 -DOF grasp parameters using reference primitive shapes and their predefined gras ps. Simulation experiments result in a 95% grasp success rate usin g the Coppeliamsim simulation environment on unseen objects. A Delta parallel ro bot and a 2-fingered fabricated gripper are used for practical experiments."
University of TehranTehranIranAsiaEmerging TechnologiesMachine LearningRobotRoboticsRobots
NETL
NSTL
