Object recognition in bin picking

垃圾箱拣选中的物体识别

• 批准号: 530916-2018
• 负责人: Greenspan, Michael
• 金额: $ 1.82万
• 依托单位: Queen's University
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2018
• 资助国家: 加拿大
• 起止时间: 2018-01-01 至 2019-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=652431
• 关键词: Object; recognition; bin; picking

The problem to be investigated is 3D computer vision-based Object Detection, Pose Estimation and Tracking**(OD/PE/Tr) of industrial parts for automated flexible assembly operations. In conventional automated**assembly, robotic devices manipulate parts which are highly structured and precisely aligned, using special**purpose fixtures and jigs. Flexible automation allows an increasing degree of uncertainty in the positions of the**parts to be manipulated, which can significantly reduce the cost and increase the value of the automation**process and equipment. Accommodating this increased uncertainty requires advanced capabilities, including**trajectory and grasp planning, collision avoidance, and computer vision methods of OD/PE/Tr.**The objective of this project is to initiate a research program to provide state-of-the-art 3D computer vision**methods and expertise to the partner company, Bluewrist Inc., a developer of innovative industrial automation**solutions and products in the areas of robotics and machine vision. The applicant, Prof. Michael Greenspan,**has a recognized research expertise in OD/PE/Tr in 3D range data.**This project will launch a meaningful research collaboration, which will increase Bluewrist's profitability**and allow them to hire more personnel to deliver, support and further develop their products. This will help**Canada with increased employment in high value technology development. There will also be a reputational**benefit to Canada, as Bluewrist further entrenches as an industry leader in innovative industrial automation**solutions.

要研究的问题是基于3D计算机视觉的对象检测，位姿估计和跟踪 **（OD/PE/Tr）的工业零件的自动化柔性装配操作。在传统的自动化装配中，机器人设备使用特殊用途的夹具和夹具来操纵高度结构化和精确对准的部件。灵活的自动化允许增加被操纵的零件位置的不确定性，这可以显著降低成本并增加自动化过程和设备的价值。消除这种增加的不确定性需要先进的能力，包括 ** 轨迹和抓持规划，避免碰撞，以及OD/PE/Tr的计算机视觉方法。该项目的目标是启动一项研究计划，为合作伙伴公司Bluewrist Inc.提供最先进的3D计算机视觉 ** 方法和专业知识，在机器人和机器视觉领域开发创新的工业自动化 ** 解决方案和产品。申请人Michael Greenspan教授 ** 在3D范围数据中的OD/PE/Tr方面具有公认的研究专长。**该项目将启动一项有意义的研究合作，这将提高Bluewrist的盈利能力 **，并使他们能够雇用更多的人员来交付、支持和进一步开发他们的产品。这将有助于 ** 加拿大在高价值技术开发方面增加就业。这也将给加拿大带来声誉上的好处，因为Bluewrist进一步巩固了其作为创新工业自动化解决方案行业领导者的地位。