Specialized localization interface technologies for industrial robotic inspection

用于工业机器人检测的专业定位接口技术

• 批准号: 499663-2016
• 负责人: Young, James
• 金额: $ 5.89万
• 依托单位: University of Manitoba
• 依托单位国家: 加拿大
• 项目类别: Collaborative Research and Development Grants
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=618848
• 关键词: Specialized; localization; interface; technologies; industrial

Industrial inspection robot use is booming, e.g., in oil and gas wells, for inspections from a safe, remote location. Although robot technology is maturing, human-machine interfaces are limited, often having only raw video feeds and simple joysticks to remotely control a robot. Improved interface technologies are needed to support operator awareness of remote robots within environments, to improve operation performance. Through this project we will invent and prototype next-generation robotic interfaces, providing operators with intuitive access to cutting edge robot localization technologies and algorithms. While robot localization - the knowledge of where a robot is within an environment - is an established research area, there is little work on how this technology can be used by non-expert robot operators. E.g., maps can be difficult to interpret, crucial details for real-world use is often omitted (e.g., uncertainty, error) and operators cannot provide input at run time (e.g., to mark an error). Further, industrial partner Inuktun Services Ltd. has indicated that existing packages do not work with their environments, e.g., robots in pipe networks or attached to boat hulls, where global features, common edges such as walls, corners, and doors, are missing. Our team expertise includes human-robot interaction and interface design (Dr. Young), human perception (Dr. Bruce), deep learning (Dr. Wang), and computer vision (Drs. Bruce, Wang) expertise, which will provide HQP with multi-disciplinary training in employable areas including robotics, computer vision, and interface design. We will prototype and evaluate novel localization interfaces and algorithms for practical industrial inspection robots. We will research interfaces to help operators interpret output from, and provide input to, localization software, and novel algorithms that provide sufficient uncertainty data, receive operator input, and work in our inspection scenario. As the Canadian robotics industry continues to emerge on the international scale, this project positions Inuktun Services Ltd. to be a leader in establishing Canada as an important player in next-generation robotic control interfaces.

工业检测机器人的使用正在蓬勃发展，例如，在石油和天然气井中，从安全、远程的位置进行检测。尽管机器人技术日趋成熟，但人机界面仍然有限，通常只有原始的视频输入和简单的操纵杆来远程控制机器人。需要改进接口技术来支持操作员在环境中对远程机器人的感知，以提高操作性能。通过这个项目，我们将发明和原型下一代机器人接口，为操作员提供直观的访问尖端机器人定位技术和算法。虽然机器人定位——了解机器人在环境中的位置——是一个成熟的研究领域，但关于如何让非专业的机器人操作员使用这项技术的工作却很少。例如，地图可能难以解释，实际使用的关键细节通常被忽略（例如，不确定性，错误），操作员无法在运行时提供输入（例如，标记错误）。此外，工业合作伙伴Inuktun Services有限公司表示，现有的包装不适合他们的环境，例如管道网络中的机器人或附着在船体上的机器人，这些环境缺少全局特征，如墙壁、角落和门等共同边缘。我们的团队专长包括人机交互和界面设计（Dr. Young）、人类感知（Dr. Bruce）、深度学习（Dr. Wang）和计算机视觉（Dr. Bruce）。这将为HQP提供包括机器人、计算机视觉和界面设计在内的就业领域的多学科培训。我们将原型和评估新的定位接口和算法为实际的工业检测机器人。我们将研究接口，以帮助操作人员解释来自本地化软件的输出并提供输入，以及提供足够不确定性数据、接收操作人员输入并在我们的检查场景中工作的新算法。随着加拿大机器人行业不断在国际范围内崭露头角，该项目将使Inuktun服务有限公司成为建立加拿大作为下一代机器人控制界面重要参与者的领导者。