Shulou(Shulou.com)06/02 Report--

The combination of AI and 5G with robotics is giving rise to new categories of consumer electronics. AI solves the problem of machine understanding of the world and human-computer interaction. 5G expands the active boundary of the robot and provides more computing power and storage space for the robot (cloud cooperative robot).

According to the classification method of IFR, robots can be divided into two categories: industrial robots and service robots. The traditional industrial robot is mainly used in automobile and spare parts, electronic manufacturing, metal and machinery, food processing and so on. Its main feature is to complete the prescribed action along the prescribed path according to the predetermined procedure.

How does AI/5G empower robots?

The application of AI artificial intelligence gives birth to the service robot. Service robots cover a wide range of areas, including medical, logistics, agriculture, commercial, civilian and other aspects. Through the use of AI technology, through data acquisition, analysis, calculation, service robots can learn human behavior, understand human intentions, and cooperate with human beings.

Main categories and representative products of service robots

According to IFR's estimates, the global robot market reached US $23.2 billion in 2017, of which the industrial robot market reached US $14.7 billion, with 85 units per 10,000 people. The rest is the service robot market, which is worth 8.5 billion US dollars.

By 2030, the robot market is expected to reach $102.8 billion, about 20 per cent of the smartphone market. Service robots will reach $56.1 billion, maintaining a compound annual growth rate of 16%, which is faster than the compound annual growth rate of industrial robots.

Split of robot market scale

1. The machine vision hardware can collect the surrounding environment information.

At present, the commonly used vision sensors are: camera, ToF lens and lidar technology.

Machine vision cameras. The purpose of the machine vision camera is to transmit the image projected to the sensor through the lens to the machine equipment that can be stored, analyzed and / or displayed. You can use a simple terminal to display images, such as using a computer system to display, store, and analyze images.

Lidar technology. Lidar is a kind of scanning sensor using non-contact laser ranging technology. Its working principle is similar to that of ordinary radar systems. It detects targets by emitting laser beams, and collects reflected beams to form point clouds and obtain data. These data can be generated into accurate three-dimensional images after photoelectric processing. By using this technology, high-precision physical space environment information can be obtained accurately, and the ranging accuracy can reach centimeter level.

ToF camera technology. TOF is the abbreviation of time-of-flight (Time of Flight) technology, that is, the sensor emits modulated near-infrared light and reflects when it meets an object. By calculating the time difference or phase difference between light emission and reflection, the sensor converts the distance of the scene to produce depth information. In addition, combined with traditional camera shooting, the three-dimensional outline of the object can be presented in the form of topographic maps with different colors representing different distances.

2. AI vision technology algorithm helps the robot to identify the surrounding environment.

Visual technology includes: face technology, object detection, visual question and answer, image description, visual embedded technology and so on.

Face technology: face detection can quickly detect faces and return to the face frame position, accurately identify a variety of face attributes; face comparison by extracting face features, calculate the similarity of two faces and give the similarity percentage; face search is to find similar faces in a designated face database; given a photo, compare with N personal faces in the designated face database to find the most similar face or faces. According to the degree of matching between the face to be recognized and the face in the existing face database, the user information and matching degree are returned, that is, 1VR N face retrieval.

Object detection: object detection technology based on deep learning and large-scale image training, which can accurately identify the comprehensive information such as object category, position, confidence and so on. Visual question and answer (VQA) system can take pictures and questions as input and produce a human language as output. Image description: need to be able to capture the semantic information of the image and generate human-readable sentences. Visual embedded technology: including human body detection and tracking, scene recognition and so on.

3. SLAM technology gives the robot the ability to plan its movement better.

SLAM, whose full name is Simultaneous Localization and Mapping, is called simultaneous location and mapping in Chinese. In SLAM theory, the first problem is called Localization, the second is called Mapping, and the third is the subsequent path planning. Through the mapping of machine vision, the robot can locate and draw the map of the location environment at the same time through complex algorithms. Through SLAM technology, it can effectively solve the problem of unreasonable planning, path planning can not cover all areas, resulting in general cleaning effect.

When there is no SLAM at all, because there is no map and no path planning, the sweeping robot will turn back in a random direction every time it encounters obstacles, unable to cover every area. When there is a SLAM, it can be covered to any area. In addition, the floor-sweeping robot is also equipped with cameras to identify shoes, socks, animal droppings and other items to achieve intelligent avoidance.

4. Ultra-wideband positioning technology based on ToF machine vision.

In robots, based on ToF technology, it can be used for high-precision ranging and positioning. At present, ultra-wideband positioning technology is commonly used.

UWB (Ultra Wideband) is a wireless communication technology, which can be used for high-precision ranging and positioning. UWB sensor streamlining devices can be divided into two types: tag and base station. Its basic working mode is to use TOF (Time of flight) to carry out wireless ranging, and calculate the position quickly and accurately according to the measured value.

5. AI natural language processing is an important technology of human-computer interaction.

90% of the means of human access to information rely on vision, but 90% of the way they express themselves depends on language. Language is the most natural way of human-computer interaction. However, it is very difficult for natural language to deal with NLP, and there are differences in grammar, semantics and culture, as well as non-standard languages such as dialects. With the maturity of NLP, the voice interaction between human and machine becomes more and more convenient, which will promote the robot to be more "intelligent".

The array microphone and loudspeaker technology of robot has been relatively mature. With the rapid development of intelligent speaker and voice assistant in recent years, microphone array and miniature loudspeaker are widely used. In the Iron Man companion robot, the voice interaction with the user depends on the microphone array and loudspeaker, this kind of companion robot is like a moving "smart speaker", expanding the boundary shape.

At present, dialogue robot can be divided into general conversation robot and professional dialogue robot. The development of natural language processing technology will enhance the interactive experience between robots and human beings, and make robots appear more "intelligent".

6. AI deep learning algorithm helps robots evolve to self-consciousness.

Hardware:

With the development of AI chip technology, robots have higher computing power. Due to the development of Moore's Law, the number of transistors per unit area is increasing, which promotes the miniaturization of chips and the improvement of AI computing power. In addition, the generation of heterogeneous chips such as RISC-V architecture chips also provides hardware support for the improvement of computing power of AI chips.

Algorithm:

AI deep learning algorithm is the future of robots. The AI deep learning algorithm gives the robot the ability to learn through input variables. Whether the future robot can have autonomous consciousness or not needs the continuous development of AI technology. The deep learning algorithm provides a possibility for the robot to acquire self-awareness. Through the training of neural network models, some algorithms have been able to surpass human beings in a single-point field. The success of Alpha Go shows us that human beings can achieve single-category self-learning ability in AI technology, and can compete with or even beat human beings in some fields, such as "go, knowledge contest" and so on.

AI deep learning algorithm, so that the robot has the ability of intelligent decision-making, get rid of the previous single input corresponding to a single output programming logic, but also make the robot more "intelligent". However, robots are still not comparable to human beings in the field of "multimodal". In particular, the signals that can not be quantified, such as smell, taste, touch, psychology and so on, still can not find a reasonable way to quantify.

7. AI+5G expands the activity boundary of the robot to provide more computing power and more storage space to form knowledge sharing.

Four pain points for mobile robots in the 4G era:

1) the scope of work is limited: the task can only be carried out within a fixed scope, and the constructed map is not easy to share and difficult to work in a large-scale environment.

2) the service coverage is limited: the operation is limited, the recognition performance still needs to be improved; the capacity is limited, only problems can be found, and it is difficult to deploy in batches quickly.

3) the provision of services is limited: the ability of complex services is poor, the ability of interaction needs to be improved, and the deployment efficiency of special services is low.

4) the cost of operation and maintenance is high: the deployment efficiency is low, and each scenario needs to build a map, plan the path, and be equipped with inspection tasks.

These four pain points restrict the penetration of mobile robots in the 4G era. On the whole, robots still need more storage space and stronger computing power. The low latency, high speed and wide connection of 5G will be able to solve the current pain points.

5G empowering mobile robots:

1) expand the working scope of the robot. The biggest capacity of 5G for the robot is to expand the physical boundary of the robot, and the support of 5G for TSN (time-sensitive network) makes the active boundary of the robot from the family to all aspects of the society. We can imagine the future of human beings and robots living together. In logistics, retail, patrol, security, fire fighting, traffic direction, medical treatment and other aspects, 5G and AI can empower robots to help human beings achieve smart cities.

2) provide more computing power and more storage space for the robot to form knowledge sharing. The promotion of 5G cloud robot provides more computing power and more storage space for the robot: flexible allocation of computing resources: to meet the synchronous positioning and mapping in complex environments. Access to a large number of databases: identifying and crawling objects; long-term location based on outsourced maps. Formation of knowledge sharing: knowledge sharing among multiple robots.

Brief introduction of main service robots 1. Floor-sweeping robots: AI technology solves the current pain points and helps the industry to flourish

It is estimated that by 2021, the global market will reach nearly 50 billion yuan, and the Chinese market will reach 15.1 billion yuan.

Calculation of the global market size of floor-sweeping robots in 2021 and other categories in 2018

2. Logistics robots: AI and 5G are giving birth to the development of AGV and unmanned distribution robots.

According to BIS Research statistics, global AGV sales increased by 38.8% to 26000 units in 2016 compared with the same period last year, mainly due to the use and promotion of e-commerce warehousing AGV; AGV sales in China increased significantly by 88.5% to 9950 units compared with the same period last year; RIC expects China's AGV sales to maintain an average annual growth rate of 43.0% from 2017 to 2021 and reach 61000 units in 2021. The automobile industry, home appliance manufacturing and other production logistics are still the main demand market for AGV in China, accounting for about 55% in 2016, and the demand is stable, but the demand for intelligent automation of products is getting higher and higher. The demand for AGV in warehousing and logistics such as e-commerce is growing rapidly, accounting for about 29% in 2016 and is expected to rise to 32% by 2021.

Automatic guided vehicle (AGV): the solution of unmanned warehousing: in the process of automatically guided transport vehicle (Automated Guided Vehicle,AGV), machine vision is used to judge the route, location, surrounding environment and other important information, which is the mainstream scheme to realize unmanned warehousing. Features of AGV: AGV is usually equipped with automatic guidance devices such as electromagnetic guidance, laser guidance or machine vision guidance, which can drive along the prescribed guidance path and is a transport vehicle with safety protection and various load transfer functions. AGV is characterized by wheeled movement, which has the advantages of fast action, high efficiency, simple structure, strong controllability and good security compared with other non-wheeled mobile robots. Compared with other equipment commonly used in material transportation, the active area of AGV does not need to lay tracks, supports and other fixed devices, and is not limited by site, road and space. Therefore, in the automatic logistics system, it can fully reflect its automaticity and flexibility, and realize efficient, economical and flexible unmanned warehousing.

Warehousing management system: in warehousing management, the application of AI algorithm to warehousing management system is relatively mature. Taking the warehouse management system based on RFID technology as an example, through the combination of RFID identification technology and computer database management query, manpower and material resources can be saved. Compared with the traditional (manual) warehouse management system, it has the following advantages:

1) fully automatic identification of incoming and outgoing items without personnel intervention, and can identify multiple items at the same time; 2) fast, accurate and automatic data collection to grasp the inventory situation in real time and accurately; 3) reduce the inventory cycle and improve the real-time performance of data to achieve visual management of inventory items; 4) reduce manpower and management costs. 3. Balanced vehicle robot: the rise of leisure sports market

Exploration period: locate transformative means of transportation. In 2001, Dean Dean Kamen launched a two-wheeled balancing car (with a pole in the middle) and founded Segway. In 2010, billionaire Jimmy Hossel bought Segway.

At first, Segway hopes to change the way the world travels, mainly to commercial customers. Due to the high price of products and few applicable scenarios, the revenue scale of Segway has been very small. Segway, which was acquired by Ninebot in 2015, has annual revenues of about $4000 million and annual sales of about 10, 000 units, according to news reports. Balance car is also known as somatosensory car, thinking car, camera car and so on. The main categories on the market include two-wheel balancing vehicles (with joystick), two-wheeled balancing vehicles (without joystick), one-wheeled balancing vehicles, balancing wheels (similar to electric roller skates). In 2010, Shane Chen (Chen Xing), a Chinese American, invented the SoloWheel, an one-wheeled balancing car. In 2012, Chen Xing invented the two-wheeled balanced car Hovertrax (without joystick).

Development period: Chinese enterprises start with copycat, and then acquire American companies. Since 2009, the Chinese market has quickly copied and sold back to the US market (the company's domestic operating entities were jointly established in 2012). In 2014, Segway sued five balanced car companies, including Le Hong and Ninebot, for infringing Segway patents in the United States. In April 2015, Ninebot wholly acquired Segway and obtained nearly 10 products and more than 400 patents of the three major product lines of Segway. Lehang acquired Solowheel in 2017 and obtained related patents.

Growth period: the rise of leisure sports market. Although the original intention of the balanced car is to change the way the world travels, what really makes the balanced car rise is the leisure sports market. After acquiring Segway and with the support of Xiaomi, Ninebot launched a low-cost product suitable for the consumer goods market (No. 9 balance car, priced at 1999 yuan, launched in October 2015). Balance cars are used by consumers as leisure sports products for children and young people, and sales have begun to increase significantly. Ninebot began to pay attention to the leisure sports market, introducing electric skateboards, kart kits and children's bicycles. In 2018, electric scooters and balancers accounted for 66% and 29% of revenue, respectively, with year-on-year changes of + 42ppt and-45ppt.

Balancing the right of way: balanced cars are allowed to be used on sidewalks in some states of the United States. In China, balancing cars are restricted to places such as closed residential roads and indoor venues. The control of a balanced car requires more skill, and there are some risks that are different from ordinary vehicles, such as Jimmy Jose, the former controller of the Segway, who died after falling off a cliff while walking his dog in a balanced car.

4. Companion robot: AI natural language processing solves human-computer interaction, and 5G promotes the development of cloud service robot.

There are two kinds of development paths for companion robots. One is that companies such as Sony and Sharp apply robot technology to make bionic robots with joints, mainly to accompany the elderly, children's education, entertainment and leisure. Another kind of companies such as Samsung, Amazon and other companies have launched display robots, mainly hoping to do robot OS platform, export software and AI services.

In terms of accompanying robots, we see two kinds of robots, one is a small robot. Such as Sony's Aibo robot dog, Sharp's robot cat, Takahashi's Robi and so on. The main functions of these robots include interaction with users, companionship, education, communication with others, etc., most of which have been commercialized.

Sony's Aibo Robot Dog

The other is large upright robots, the most famous of which are Honda's Asimo, Boston Dynamics and other excellent companies. This kind of large robot is generally equipped with biped, which can realize the functions of going up and down stairs, grabbing objects, interacting with humans and so on.

Honda Asimo

5. Medical robots: AI and 5G have a strong demand for medical diagnosis, surgery and limb rehabilitation.

Capsule robot: capsule robot + automatic diagnosis system: it is possible to automatically judge the possible problems of patients through the collected images, so as to provide reference for doctors' diagnosis. With the automatic diagnosis system in the cloud, remote diagnosis and multi-location diagnosis can be realized.

Capsule robot

Surgical robot: at present, the widely used surgical robots include object-holding robot, navigation robot and master-slave robot.

Surgical object-holding robot

Navigation robot: plan the surgical path for surgeons and give tips during the operation. At present, surgical navigation is divided into optical navigation and electromagnetic navigation. Optical navigation has high accuracy and is not subject to electromagnetic interference from other equipment, but it is easy to block the optical path. Electromagnetic navigation is flexible in operation and low in posture, but it will be subject to electromagnetic interference. Optical navigation is mainly used in neurosurgery, spinal surgery, joint surgery, maxillofacial surgery. Electromagnetic navigation is mainly used in intracranial biopsy, catheterization and bronchoscopy.

Master-slave robot: provide technical support for surgeons to operate remotely and off-stage. Leonardo da Vinci robotic surgery system is a master-slave robot system, which can help doctors to operate off Taiwan or even remotely in the future, so that doctors can sit for surgery and reduce medical accidents caused by fatigue and distraction. Limb movement rehabilitation robot. At present, the rehabilitation robots are mainly upper limb rehabilitation robot and lower limb exoskeleton robot.

6. Commercial retail robot: the scale and popularization of indoor unmanned distribution robot

In the field of commercial retail, in recent years, many robots have entered, including supermarket shopping guide robot, hotel water and food delivery robot and so on. these indoor distribution robots help commercial retail improve operational efficiency and reduce labor costs. and add interest to the service, welcomed by users. With the continuous rise of labor costs, the prospect of commercial robots is broad and worthy of attention.

Shopping guide robot interacts with human beings

7. Bionic robot: become a part of human being

Consciousness-controlled robot: in some cutting-edge robot research, consciousness-controlled robot arms have emerged. Users need to connect the robotic arm to the human brain system, and through some training, the robotic arm can be controlled to grab drinks.

Bionic robots: humans have developed some robots that mimic human organs, which can restore the normal function of the disabled by using the consciousness control mentioned above, which will be a big market.

Consciousness-controlled robot

5G+AI technology affects many industries, but none of them is as promising as the future of the robotics industry. AI will help traditional robots never cooperate with human beings, become able to serve human beings, and gradually advance to intelligent cognition, reasoning and decision-making, while the maturity of 5G technology will further expand the application boundary of robots. Through the characteristics of low delay, high speed and wide connection, 5G can expand the application boundary of the service robot, and provide more computing power and more storage space for the robot to form knowledge sharing.

In the future, as technology accumulates to the point of qualitative change, floor-sweeping robots and logistics robots will become "pediatrics". Robots like science fiction movies are likely to appear in the near future and fully infiltrate human society to help human beings live together.

Original link: https://developer.aliyun.com/article/753958

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.