Collaborative Handheld 3D Sensors

建议摘要-Laurendeau 17782*该研究计划致力于解决计算机视觉中的一个有前途的主题，并包括探索人类操作员如何利用协作式手持3D传感器组进行质量控制。虽然过去人们已经对安装在移动机器人上的协作传感器问题进行了深入的研究，但人类开发的协作3D传感器的话题却很少受到关注。随着快速准确的手持式3D传感器的出现，这个话题正变得高度相关。这些传感器现在被广泛用于一般的质量控制和检查。*申请人开发的3D视觉专业知识表明，人类操作员直观地知道在哪里扫描3D数据，但在决定应该在哪里收集新数据或在对象上是否收集了足够的“可靠数据”方面效率要低得多。因此，迫切需要研究3D传感器如何在数据收集过程中为人类操作员提供指导。*该计划的目标是：*1.研究人类操作员如何在质量控制中有效地使用手持3D扫描仪组来对大型人造物体进行测量。3D传感器协作问题解决了3D传感器组收集数据的优化方法的制定，并探索了传感器定位、视图对齐、3D数据分割为几何基元和特征(边)以及对象部件与CAD模型之间的对齐等子问题；*2.建立支持上述协作方法的插件软件框架；*3.将人-传感器-接口结合到优化方法中，帮助人类操作员使用手持3D扫描仪收集足够可靠的数据；*4.培训3D视觉和协作研究方面的高素质人员；*5.在期刊和会议上传播研究成果，并将研究成果转移给潜在的行业合作伙伴。*根据ARC咨询小组的数据，2012-2016年间，3D扫描市场预计将以8.8%的年增长率增长。手持式3D传感器在这个市场上的份额越来越大。然而，这些传感器仍然很难被非视觉专家使用。该研究计划建议研究新的优化方法和人机接口，以指导人类用户在他们的数据收集任务中取得更好的结果。*一个主要成果将是一个范例的形成、实施和验证，该范例将优化由人类用户操作的手持3D传感器组之间的协作，这个问题几乎没有受到关注，但随着这些传感器在工业中的广泛使用，这一问题变得非常重要。*预计该计划中将要开发的优化框架和补充技术将导致更好地使用手持3D传感器，减少扫描时间和后处理时间，并提高质量控制的效率。这与3D扫描行业高度相关，加拿大在该行业处于世界领先地位。高素质人才的培养也将有助于该行业保持其领先地位。目前3D视觉专家短缺，那些将参与这项研究的人将找到令人兴奋的工作。

Summary of proposal - Laurendeau 17782*The research program addresses a promising topic in computer vision and consists in exploring how groups of collaborative handheld 3D sensors can be exploited by human operators for Quality Control. Although the problem of collaborative sensors mounted on mobile robots has been investigated in depth in the past, the topic of collaborative 3D sensors exploited by humans has received little attention. This topic is becoming highly relevant with the advent of fast and accurate handheld 3D sensors that are now widely used in quality control and inspection in general.*The expertise in 3D vision developed by the applicant has demonstrated that human operators intuitively know "where" to scan the 3D data but are much less efficient in deciding where new data should be collected or whether or not enough "reliable data" has been collected on the objects. There is thus an urgent need for investigating how 3D sensors could provide guidance to human operators during the data collection process.*The objectives of the program are to:*1. study how groups of handheld 3D scanners can be used efficiently by human operators in Quality Control to perform measurements on large man-made objects. The 3D sensors collaboration problem addresses the formulation of an optimization approach for data collection by groups of 3D sensors and also explores sub-problems such as sensor positioning, view alignment, segmentation of 3D data into geometric primitives and features (edges) as well as alignment between object parts and a CAD model;*2. build a plug-in software framework supporting the above collaboration approach;*3. combine Human-Sensor-Interfaces to the optimization approach to help human operators to collect sufficient reliable data using handheld 3D scanners;*4. train highly qualified personnel in 3D vision and collaborative research;*5. disseminate research results in journals and conferences and transfer results to potential industrial partners.*According to the ARC Advisory Group, the 3D scanning market is expected to grow at an 8.8 percent annual rate for 2012-2016. Handheld 3D sensors are a growing part of this market. However, these sensors still remain difficult to use by non-vision specialists. The research program proposes to investigate new optimization approaches and human-sensor interfaces to guide human users in achieving better results in their data collection tasks.*A major outcome will be the formulation, implementation and validation of a paradigm which will optimize the collaboration of groups of handheld 3D sensors operated by human users, a problem that has received little attention but that is of major importance with the widespread use of these sensors in industry.*It is anticipated that the optimization framework and complementary technology to be developed in the program will lead to better ways of using handheld 3D sensors, reduce scanning time and post-processing time, and increase efficiency in quality control. This is highly relevant to the 3D scanning industry, a sector in which Canada is a world leader. The training of highly qualified personnel will also help this industry to maintain its leadership position. There is currently a shortage of specialists in 3D vision and those who will participate in this research will find stimulating jobs.