arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1802.00520cs.RO
收藏

真实世界的多目标、多抓取检测

Real-world Multi-object, Multi-grasp Detection

Institute for Robotics and Intelligent Machines, Georgia Institute of Technology(佐治亚理工学院机器人与智能机器研究所)

Fu-Jen Chu, Ruinian Xu, Patricio A. Vela

摘要:

提出了一种深度学习架构来预测机器人操作的可抓取位置。它考虑的情况下,没有,一个或多个对象(S)被视为。通过将学习问题定义为具有零假设竞争而不是回归的分类,具有RGB-D图像输入的深度神经网络在Single Shot中预测单个对象或多个对象的多个抓取候选。该方法在Cornell数据集上的表现优于最先进的方法,分别在图像和对象分割上具有96.0%和96.1%的准确度。多目标数据集上的评估说明了该架构的泛化能力。抓取实验取得了96.0%的把握定位和88.0%的抓取成功率的测试集的家用物体。从图像到计划的实时过程不到0.25秒。

英文摘要:

A deep learning architecture is proposed to predict graspable locations for robotic manipulation. It considers situations where no, one, or multiple object(s) are seen. By defining the learning problem to be classification with null hypothesis competition instead of regression, the deep neural network with RGB-D image input predicts multiple grasp candidates for a single object or multiple objects, in a single shot. The method outperforms state-of-the-art approaches on the Cornell dataset with 96.0% and 96.1% accuracy on image-wise and object- wise splits, respectively. Evaluation on a multi-object dataset illustrates the generalization capability of the architecture. Grasping experiments achieve 96.0% grasp localization and 88.0% grasping success rates on a test set of household objects. The real-time process takes less than .25 s from image to plan.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑