GTI5G和云化机器人白皮书（英文版）.pdf

1 GTI 5G and Cloud Robotics White Paper gtigroup 3 Confidentiality: This document may contain information that is confidential and access to this document is restricted to the persons listed in the Confidential Level. This document may not be used, disclosed or reproduced, in whole or in part, without the prior written authorization of GTI, and those so authorized may only use this document for the purpose consistent with the authorization. GTI disclaims any liability for the accuracy or completeness or timeliness of the information contained in this document. The information contained in this document may be subject to change without prior notice. Document History Date Meeting # Version # Revision Contents DD-MM-YYYY NA DD-MM-YYYY DD-MM-YYYY DD-MM-YYYY DD-MM-YYYY 4 5 Cloud Robotics: Trends, Technologies, Communications Abstract Cloud robots are controlled from a “brain” in the cloud. The brain, located in a data center, makes use of Artificial Intelligence and other advanced software technologies to deal with tasks that in traditional robots were undertaken by a local, on-board controller. Compared to local robots, cloud robots will generate new value chains, new technologies, new architectures, new experiences and new business models, this white paper will explore these aspects. 1. Introduction Cloud robotics is a relatively recent concept. Early work dates back to 2010, when the European Commissions RoboEarthiproject began. This aimed to establish a “World Wide Web for robots”. RoboEarth and later projects such as Rapyutaiiand Robohowiiiformalized the basic concept and technologies, and are still influencing cloud robotic research today. There are three core advantages of cloud robots compared to stand-alone robots: x Information sharing Many cloud robots can be controlled from one brain, and the brain can accumulate visual, verbal, and environmental data from all connected robots. Intelligence derived from this data can be used by all the robots controlled by the brain. As with other cloud services, information collected and processed on each robot will always be up-to-date and backed-up safely. Developers also benefit, as they can build reusable solutions for all cloud-connected robots. x Offloaded computation Some robot tasks require more computational power than a local controller can economically deliver. Offloading to the cloud data-intensive tasks such as voice and image recognition, voice generation, environmental mapping and motion planning will lower the hardware requirements and power consumption of robots, making them lighter, smaller, and cheaper. x Collaboration Cloud robots do not need to work alone. Using the cloud as a common medium, two robots can work together to carry an object too heavy for one, or a group of simple worker robots can work with a local map, provided by a leader robot with costly sensors. 6 Distributed version of AlphaGo exploited 40 search threads, 1202 CPUs and 176 GPUsx, no ordinary robot can install inside. But cloud robot can make use of it. 2. Applications for cloud robots Using cloud resources empowers robots and gives them new capabilities in many areas: x Intelligent visual processing: image classification, target detection, image segmentation, image description, character recognition. x Natural language processing: semantic understanding based on depth learning, accurate identification of user intent, multi-intention analysis, emotional analysis. Makes use of a powerful background knowledge base. x Facial recognition: face detection algorithm based on depth learning; In the real-time video stream to accurately detect the face; Any face mask and real-time detection under the viewing angle; To overcome: the side face, half obscured, blurred face; x Extension from current robot applications: outdoor map navigation, indoor positioning and navigation, typical product identification, universal item identification, environmental understanding, text reading, voice prompts. Figure 1: Large scale data collection with an array of robots (14 robots are sharing experiences of machine learning for grasping ) Source: research.googleblog/2016/03/deep-learning-for-robots-learning-from.html 7 The applications that will emerge for cloud robots are of many kinds; some are emerging now others are at an early stage of development. Logistics Amazon, Jingdong, S.F. Express and other companies have deployed logistics robot systems. The wheeled AGV (Automated Guided Vehicle) is the main type of logistics robot (though logistics companies are also trialling the use of aerial drones). By connecting to the cloud, AGVs can achieve unified scheduling (where all AGVs are working as a single system for maximum efficiency). In addition, AGVs can be equipped with machine vision systems, and video can be transmitted to cloud-based systems to handle a variety of situations on the road. Eventually this will result in AGVs coming out of controlled areas to take on more work, including in public places for delivery of parcels or food. Security and surveillance In public places, cloud robots can perform 24/7 security inspections, replacing security personnel. The cloud robot will collect video and still images and send them to the public safety cloud for real-time identification of suspicious people and activity. Such robots are already being used at Shenzhen airport in China. Guidance In public places such as enterprises, banks and hospitals, robots like Softbanks Pepper are being used to guide visitors. They are also being used to deliver retail services by companies including Nestle, Yamada Electric and Mizuho Bank. Cloud robots can make use of a vast knowledge database in the cloud, and communicate using natural language; they can even recognise and respond to peoples expressions using cloud-AI-based image analysis, to improve the use experience. Education, entertainment and companionship In recent years, the application of machine vision and artificial intelligence has resulted in the development of many robots for education and entertainment. Examples include Jibo, Asus Zenbo, and Softbank Nao. These robots have a humanoid appearance and the ability to use natural language. They can download content from the cloud to provide education and entertainment services. Personal assistance and care Providing personal assistance and care for the elderly is widely considered the “next big thing” in robotics. The power of the cloud makes care robots behave more like humans. They can carry out real-time monitoring of personal health, help people move about, and complete housework. An example of this type of robot is Softbanks Romeo. 3. Market trends Robots can be categorised as industrial robots or service robots, according to their use. Service robots can be further divided into professional services robots and personal home service robots. Professional service robots are used in the fields of medicine, construction, underwater engineering, logistics, defence and safety. Personal home service robots are used to undertake housework, provide companionship and personal assistance, and are also used in other fields. 226.2$bn According to market analyst company Tractica, the value of the global robot market will grow from $34.1 billion in 2016 to $226.2 billion in 2021, with a compound annual growth rate (CAGR) of 46% in value terms. Most of the growth will be in the market for non-industrial robotsIX. 2bn One of the major drivers of this market growth is the aging population. There are fewer working-age people to take care of the increasing Figure 2: Cloud-powered smart devices and communication robots Source: Softbank 8 numbers of the elderly. The UN has forecast that by 2050 21% of the global population will be over the age of 60 a total of over 2 billion people. Robots have a role to play here. In addition, industrial automation continues to develop at a rapid pace, with initiatives such as Industry 4.0 in Germany and Made in China 2025. Advances in technologies including Artificial Intelligence, the Internet of Things and wireless communications are making robots more capable. They can now identify their surroundings, calibrate their position, plan trajectories, and use natural interfaces to interact with humans. There have been increases in the capabilities of robots used in industry, agriculture, logistics and education. The rapid rise in the use of drones is also evidence of the increasing capabilities of robots. Cloud robots will soon become the norm Cloud-based AI and connectivity will shape the development of the robot market significantly in the next few years. These techniques have already begun to change the way that people interact: technology giants have developed AI-based systems that are becoming widely used. Examples include Google Cloud Speech API, Amazon Alexa, Baidu Duer, IBM Watson, Apple Siri and Microsoft Cortana. 12% According to Huawei GIV, by 2025 the use of mobile connectivity and artificial intelligence will result in robot penetration in the family of 12%; intelligent robots will change the face of all industries in the same way that the automotive industry was transformative in the 20thcentury. GTI cloud robotics working group research forecasts that by 2020 connected robots will account for 90% of all robots, and about 20 million new connections will be needed every year to support their day-to-day operations. GTI cloud robotics working group has examined the robotics market in detail. Its work suggests that by 2020 the proportion of connected robots globally will be 90%, and about 20 million new connections will be needed every year to support their day-to-day operations. Figures 3-7 show projections for sales of connected robots. Figure 5 Connected domestic robots (million) Source: GTI cloud robotics working group Figure 3: Connected robot sales 2016-2020 (million)Source: GTI cloud robotics working group Figure 4: Connected logistics system robots (thousand) Source: GTI cloud robotics working group 9 In the next few years, domestic robots and recreational robots will occupy most of the shipments of connected robots. With the increase in the capability of robots, the needs of individuals and families for service robots will continue to increase. The current public acceptance of robotic services, especially medical services, is not high. People are skeptical about whether robots can reach the levels of skill of human doctors. However, in the next few years, with robots abilities gradually improving, peoples acceptance of robotic medical services will increase. Research published by the Open Roboethics Initiative shows that the main expectation of home service robots is to complete housework to make life easier. In addition, education, inspection and security needs are relatively strong. Figure 6: Connected entertainment robots (million) Source: GTI cloud robotics working group Figure 7 Connected disabled care assistant robot (thousand) Source: GTI cloud robotics working group willing unwilling neither27 30 35 40 45 60 UK GERMANY NORWAY NETHERLANDS QATAR TURKEY Figure 8: Willingness to use AI and robots for healthcare Source: PwC 75% 38% 32% 32% 26% 19% 17% 11% 9% Other Fancy toy Pet replacementCompanion for famlily For coolness factor Education tool for childHome security Extension of electronic devices Household choresFigure 9: Willingness to have surgery performed by robot Source: PwC Figure 10: Reasons for purchase a home robot Source: Open Roboethics Initiative 10 4. The cloud robotics value chain The value chain of cloud robotics is shown in Figure 11. The robot platform provider delivers the robot which runs applications; these applications use intelligent services from the AI provider, making use of the mobile network to provide a “smart” user experience for end users. 4.1 Robot platform the technologies behind cloud robots The definition of robot may vary by context, but a general definition is “A mechanical system with three elements: controller, sensor, and effector/actuator”. Controller As the robot gains complexity and demands become more advanced, the controller part has also developed and todays robots are often controlled by OS or rich middleware, such as ROS (ros/), OpenRTM-aist, middleware compliant with Object Management Group (OMG) Robotic Technology Component (RTC) Specificationiv, and NAOqi (OS used in Softbanks Pepper). In cloud robots, the controller part is achieved by coordination of cloud and local systems. Sensors Robots use many different types of sensors relevant to their function. The most important types are: x Cameras and microphones Sophisticated cameras and microphones are required to sense the environment. For instance, Softbanks human-sized communication robot Peppervuses a 3D camera and two HD cameras (see Figure 12), and four directional microphones to detect where sounds are coming from and locate users position. 3D cameras are used to provide position detection and mapping (often referred to as SLAM (simultaneous location and mapping). Other 3D positioning sensors and technologies are also used, from inexpensive proximity sensing, sonar and photoelectric sensing to more accurate and costly techniques such as LiDAR that can be used to build up high resolution 3D pictures across a wide coverage area. Figure 12: Microphone array and top camera in Pepper robot Source: techon.nikkeibp.co.jp/article/COLUMN/20150623/424503/?P=2 Robot Platform Application Provider Mobile Network AI Provider End users Figure 11: Cloud robotics value chain Source: GTI cloud robotics working group