Robots with eyes and brains reshaping manufacturing, says Solomon Technology chairman@DIGITIMES
Industrial robots have been employed at car factories for many years, but robots are beginning to have a bigger influence on people's daily lives, as they now have eyes (enabled by 3D vision) and brains (artificial intelligence, AI) to have recognition capability and become smart and thuse are fundamentally changing manufacturing industries, said Johnny Chen, chairman of Solomon Technology.
工业机器人已经在汽车工厂工作多年，但机器人开始对人们的日常生活产生更大的影响，因为他们现在有眼睛（由3D视觉启用）和大脑（人工智能，AI）具有识别能力，变得聪明，这些从根本上改变了制造业。所罗门科技公司（Solomon Technology）董事长Johnny Chen说。
Aging populations and rising labor cost are spurring manufacturing industries to upgrade to smart manufacturing, letting robots do highly repeated, low value-added, labor-consuming and dangerous jobs. However, the major challenge to boosting competitiveness is breakthrough in eye (3D vision), brain (AI), hand (robotic arm) and foot (automated guided vehicle, AGV) technologies as well as integration and application of these technologies, Chen analyzed.
人口老龄化和劳动力成本上升促使制造业升级为智能制造业，让机器人做高度重复、低附加值、耗费劳动力和危险的工作。然而，提升竞争力的主要挑战是眼睛（3D视觉），大脑（AI），手（机器人手臂）和脚（自动导引车辆，AGV）技术的突破，以及这些技术的集成和应用，Chen分析。
For industrial automation in the past, robots at most factories were used to undertake monotonous duties because they lacked the intelligence and sensing ability that human workers have, Chen noted. Along with development of AI and 3D vision technologies, robots are becoming smarter and more capable, Chen indicated. Significantly different from traditional industrial robots, smart and vision-enabled robots can be used as real assistants in manufacturing industries of semiconductors, precision devices, biomedical products and electronics to undertake inspection/detection, scanning, size measurement, recognition, picking and placement of components and assembly, Chen said.
对于过去的工业自动化，大多数工厂的机器人被用来承担单调的职责，因为他们缺乏人类工人的智力和感知能力，Chen指出。随着人工智能和3D视觉技术的发展，机器人变得更加智能和强大。与传统工业机器人明显不同的是，智能和视觉功能的机器人可以用作半导体，精密设备，生物医学产品和电子产品制造行业的真正助手，以进行检查/检测、扫描、尺寸测量、识别，拾取和放置元件。
It is difficult to use AOI (automated optical inspection) to detect defects of irregular shapes (such as scratches, stains and fissures), but machine vision based on deep learning can solve the problem, Chen noted, adding that Solomon has developed AI-based machine vision software Solvision for the purpose.
很难用AOI（自动光学检测）来检测不规则形状（如划痕，污渍和裂缝）的缺陷，但基于深度学习的机器视觉可以解决这个问题，Chen指出，并补充说所罗门已经开发出基于AI的机器视觉软件Solvision就是为了这个目的。
For machine vision to recognize defects, deep learning does the job without lengthy and complicated programming and the machines only need to learn several samples with marks representing types of defects. In case there are many types of defects and it is unable to know all types of defects, software based on unsupervised learning technology can be used to detect such defects. The technology lets the software learn only what golden samples are, allowing it recognize the difference between a golden sample and an item with some defect rather than teaching the software to know every defect.
为了使机器视觉能够识别缺陷，深度学习可以完成工作而无需冗长和复杂的编程，并且机器只需要学习几个带有表示缺陷类型的标记的样品。在存在许多类型的缺陷并且不能知道所有类型的缺陷的情况下，可以使用基于无监督学习技术的软件来检测这些缺陷。该技术使软件只能学习黄金样本，使其能够识别黄金样本和具有某些缺陷的项目之间的差异，而不是教导软件了解每个缺陷。
It is necessary for 3D vision software used in robots to be easy to operate and program and equipped with personalized user interfaces as well as applicable to 6-axis robots or SCARAs (selective compliance assembly robot arms) developed by main international vendors, Chen said. Denmark-based Universal Robots, a developer of collaborative robots, has adopted such 3D vision software.
Chen说，机器人中使用的3D视觉软件必须易于操作和编程，并配备个性化的用户界面，以及适用于主要国际供应商开发的6轴机器人或SCARA（选择性合规装配机器人手臂）。总部位于丹麦的协作机器人开发商Universal Robots采用了这种3D视觉软件。
Smart and vision-enabled robots are eventually aimed to collaborate with human workers and independently work at factories as well, Chen noted. Modularized IMS (intelligent manufacturing systems) feature flexible and seamless integration of AI solutions, 3D vision solutions, robotic arms, IoT (Internet of Things) networks and AGVs through users' choosing modules from IMS based on their needs to realize smart manufacturing.
Chen指出，智能和具有视觉功能的机器人最终的目标是与人类工作人员合作，并在工厂独立工作。模块化IMS（智能制造系统）通过用户根据需求选择模块，将人工智能解决方案、3D视觉解决方案、机器人手臂、物联网（IoT）网络和AGV的灵活无缝集成在一起，实现智能制造。
网友评论