bg line bg line

Seven 5G+AI Robotics Technologies That Have Revolutionized Modern Life

The interaction between AI and 5G with robotics has driven the emergence of new and unforeseen breakthroughs in consumer electronics. AI enables machines to comprehend their surroundings and perform human-machine interactions. 5G expands the scope of robotics and provides more computing power and storage space for robots (cloud robotics).

[ShangHai, China, March 14, 2020]According to the framework proposed by the International Federation of Robotics (IFR), robots can be classified as either industrial robots or service robots. Traditional industrial robots are mainly used in automobiles, parts and components, electronics manufacturing, metal and machinery, and food processing. Their main function is completing specific actions along a specified path, according to the preset procedures.

How Do AI and 5G Empower Robots?

      The emergence of AI has led to the creation of service robots. Service robots cover a wide range of fields, including healthcare, logistics, agriculture, and commerce, among others. By utilizing AI, service robots can learn human behavior, comprehend human intentions, and collaborate effectively with humans by harnessing data collection, analysis, and computing.

Major categories and typical products for service robots

According to the IFR, the global robot market was valued at US$23.2 billion in 2017, of which US$14.7 billion could be attributed to the industrial robot market. The number of robots owned per 10,000 people reached 85. The remaining US$8.5 billion falls under the service robot market.

By 2030, the robot market is expected to be worth a staggering $102.8 billion, about 20% the value of the smartphone market. The service robot market with reach US$56.1 billion, maintaining a compound annual growth rate of 16%, faster than the compound annual growth rate for industrial robots.

Robot market segmentation

1. Machine vision hardware can collect information related to the surrounding environment.
Vision sensors in common use include cameras, time-of-flight (ToF) lenses, and lidar technologies.

Machine vision cameras. The purpose of a machine vision camera is to transmit an image that is projected to a sensor through a lens to a machine device that can store, analyze, and (or) display the image. A single device, for example, a PC can be used to display, store, and analyze images.

Lidar technology. A laser radar is a scanning sensor that makes use of non-contact laser ranging technology. Its working principles are similar to those for common radar systems. It emits laser beams to detect targets, and collects reflected beams to form point clouds and obtain data. The data can then be converted into accurate three-dimensional (3D) stereo images after photoelectric processing. With this technology, high-precision physical space-related information in the environment can be obtained, accurate down to the centimeter level.

ToF technology. ToF is short for time of flight. It involves a sensor emitting modulated near-infrared light, and reflecting the light when it encounters an object. The sensor then calculates the time difference or phase difference between the light emission and reflection to determine the distance to the photographed object and generate depth information. With the help of traditional cameras, the 3D outlines of objects can be displayed in different colors to represent the different distances.

2. AI vision algorithms help robots identify surrounding environments.

Visual technology includes face technology, object detection, visual question answering (VQA), image description, visual embedded technology, and other related fields.

Face technology: Face detection quickly detects the presence of faces and returns the position of the face bounding box to accurately identify facial features. Face comparison extracts facial features and calculates the similarity percentage between two given faces. Face search looks for one or more faces in a specified face image library by comparing a specified facial image with the N faces in the facial image library. The system returns the user information and matching degree based on the face awaiting recognition, and the faces in the existing facial library.

Object detection: Based on deep learning and large-scale image training, object detection is capable of accurately identifying a comprehensive range of information such as the object type, location, and confidence in images. VQA: Provides natural language answers to natural language questions about the content of visual images. Image description: Captures semantic information for an image to generate human-readable sentences. Visual embedded technology: Includes human body detection and tracking, scene identification, etc.

3. SLAM enables robots to better plan mobility solutions.

SLAM stands for simultaneous localization and mapping. The SLAM theory consists of three important elements: localization, mapping, and path planning. Through the mapping of machine vision, robots can use complex algorithms to locate and draw a map for the location environment. SLAM technology can effectively resolve improper planning solutions and implement path planning to cover all areas.

When SLAM is not used, the robot will tend to turn back in a random direction every time that it encounters an obstacle, as there is no map or path planning solution in place. As a result, the robot is unable to cover every area. SLAM, by contrast, provides for full coverage. In addition, the robot can be equipped with cameras to identify shoes, socks, animal waste and other items, and implement intelligent avoidance mechanisms.

4. Ultra-wideband location technology, based on ToF machine vision

In robotics, high-precision ranging and positioning is mainly achieved via ToF technology. Ultra-wideband (UWB) positioning technology is commonly used.

UWB is a wireless communication technology that can be used for high-precision ranging and positioning. There are two types of simplified UWB sensors: labels and base stations. The most basic method involves using ToF to perform wireless ranging, then quickly and accurately calculating the location based on the ranging value.

5. AI-based natural language processing (NLP) is an important technology behind human-computer interactions.

Humans obtain 90% of information through vision, while relying on language for 90% of expression. Thus, language is the most natural way for humans and machines to interact. However, NLP can be extremely difficult, due to the complexity of grammar, semantics, and culture, as well as non-standard languages such as dialects. With the maturity of NLP, voice interactions between human beings and machines have become increasingly convenient, which will in term enhance the intelligence of robots.

With the rapid development of smart speakers and voice assistants in recent years, microphone arrays and micro speakers have been widely used. For example, in the Iron Man companion robot, voice interaction with users is dependent on microphone arrays and speakers.

Currently, dialog bots can be classified into general dialog bots and professional dialog bots. The development of NLP technologies will further enhance interactions between robots and humans, and make robots increasingly intelligent.

6. AI deep learning algorithms help robots evolve toward autonomy.

Hardware: The development of AI chip technologies enables robots to have higher computing power. According to Moore's Law, the number of transistors contained in a chip per unit area is expected to continue to increase, driving chip miniaturization and increased AI computing power. In addition, heterogeneous chips, such as RISC-V architecture chips, will also provide hardware support for enhanced computing in AI chips.

Algorithms: AI deep learning algorithms are the future of robotics AI deep learning algorithms provide robots with the ability to learn through the inputting of variables. For truly autonomous robots to come into being, AI technologies will need to continue to improve. Deep learning algorithms offer enticing potential, presenting the possibility that robots will gain autonomy. Through the training of neural network (NN) models, certain AI algorithms have already surpassed humans in such competitions as Go, Texas hold 'em, and trivia contests. A representative example is the success of AlphaGo.

AI deep learning algorithms provide robots with intelligent decision-making capabilities, freeing robots from the programming logic of single input and single output, and making them newly intelligent. However, robots are unable to rival humans in the "multimodal" field. In particular, we still lack methods for quantifying such signals as smells, tastes, sensations, and psychological states.

7. AI+5G expands the scope of robotics, providing more computing power and storage space to promote the boundless sharing of knowledge

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

(1) Limited work scope: Tasks can be executed only in a fixed scope. Maps that are created can't be shared, and don't work in extreme environments.
(2) Limited service coverage: Computing capabilities are limited, and identification performance still needs to be improved. Problems can only be detected, and are difficult to resolve in batches.
(3) Limited services: Capabilities for dealing with complex services remain poor, interaction capabilities need to be improved, and the deployment efficiency of special services remains low.
(4) High operations and maintenance (O&M) costs: Deployment efficiency is low. Maps need to be created, paths need to be planned for each scenario, and inspection tasks need to be prepared.

These four pain points have hindered the widespread deployment of mobile robots in the 4G era. Robots require abundant storage space and powerful computing capabilities that remain out of reach. However, 5G addresses these myriad of challenges, thanks to its low latency, high efficiency, and broad connectivity.

How 5G empowers mobile robots:

(1) Expands the work scope for robots. The biggest advantage of 5G for robots is in expanding their physical range. In supporting Time-Sensitive Networking (TSN), 5G enables robots to not only work in homes, but also in every other facet of daily life. It's easy to envision a future in which human beings and robots live together. 5G and AI can enable robots to assist humans with building smart cities, covering such areas as logistics, retail, inspection, security, firefighting, traffic, and healthcare.
(2) Provides more computing power and storage space for robots, promoting knowledge sharing. 5G-enabled cloud robots can provide more computing power and storage space for robots, and allocate computing resources on an elastic basis, to meet the requirements of synchronous positioning and drawing in complex environments. 5G-enabled cloud robots can also access a large number of databases to identify and capture objects, in addition to implementing long-term positioning based on outsourcing maps. Knowledge can be shared among robots as well.

Introduction to Major Service Robots

1. Robotic vacuum cleaners: AI technologies resolve pain points, spurring the development of the industry at large.

It is estimated that by 2021, the global market will approach CNY50 billion, with the size of the Chinese market at CNY15.1 billion.

Global and Chinese market estimates for robotic vacuum cleaners

Estimated global market scale of robotic vacuum cleaners in 2021, and other categories in 2018

2. Logistics robots: AI and 5G are driving the development of automated guided vehicles (AGVs) and unmanned delivery robots.

According to data from BIS Research, the global AGV sales volume increased by 38.8% in 2016 to 26,000 units, mainly due to the use and promotion of e-commerce warehousing AGVs. The sales volume of AGVs in China increased by 88.5% year-on-year to 9,950 units. RIC expects the sales volume of AGVs in China to maintain an average annual growth rate of 43.0% from 2017 to 2021, reaching 61,000 units. Automotive productive and logistics, and home appliance manufacturing, are still the main applications for AGVs in China. In 2016, the market share was estimated at 55%. Demand is stable, but the requirements for intelligent product automation are increasingly high. The demand for AGVs in warehousing and logistics services such as e-commerce have grown at a rapid pace. In 2016, demand accounted for about 29% of the market, but by 2021, this is expected to increase to 32%.

Global market forecast for AGVs

Chinese market forecast for AGVs

AGVs: When in motion, an AGV determines a range of important information, such as the driving route, placement position, and surrounding environment through machine vision. AGVs are often used for unmanned warehousing. AGV features: An AGV can travel along a specified guided path by using radio waves, vision cameras, magnets, or lasers for navigation. As transport vehicles, they come equipped with safeguards and wide-ranging transportation functions. AGVs are wheeled, which offers a number of advantages, including quick movement, high efficiency, simplified structures, easy controllability, and sufficient safety, when compared with non-wheeled mobile robots. Compared with other equipment commonly used to transport materials, AGVs do not move in areas with fixed devices, such as tracks and supports, and are not restricted by site, road and space-related limitations. In automatic logistics systems, AGVs have fully demonstrated their automated and flexible nature, and helped implement economical and versatile unmanned warehousing solutions.

Warehouse management systems: Applications for AI algorithms have matured. Warehouse management systems based on RFID technology serve as a representative example. The interaction between RFID identification technology and computer database management and querying can save manpower and materials. Compared with the traditional (manual) warehouse management systems, it offers the following advantages: (1) Automatically identifies inbound and outbound items without the need for human intervention, and identifies multiple items at the same time. (2) Quickly, accurately, and automatically collects data, and knows the inventory status in real time. (3) Shortens the counting period, improving the timeliness of data, and achieving visualized management of inventory items. (4) Reduces labor and management costs.

3. Self-balancing scooters: A new trend in the leisure sports market.

Exploratory phase: Positioned as groundbreaking transportation vehicles. In 2001, Dean Kamen launched a two-wheeled self-balancing scooter (with a handle in the middle) and founded Segway Inc. In 2010, Segway Inc. Was sold to Jimi Heselden.

In the beginning, Segway had hoped to bring about a revolution in transportation, mainly targeting commercial customers. However, Segway sales were low due to high product prices and limited application scenarios. In 2015, Segway was acquired by Ninebot. According to news reports, Segway's annual revenue was approximately US$40 million, with annual sales volume at 10,000 units. Self-balancing scooters are also known as hoverboards, self-balancing boards, and swegways. The main categories on the market include two-wheeled self-balancing scooters (with and without handles), one-wheeled self-balancing scooters, and self-balancing wheels (similar to electric skateboarding shoes). In 2010, A Chinese-American named Shane Chen invented a one-wheeled self-balancing scooter, which he named SoloWheel. In 2012, Chen also invented a two-wheeled self-balancing scooter (without a handle), which came to be called Hovertrax.

Development phase: Chinese enterprises started as imitators, before acquiring American enterprises. Since 2009, the number of Chinese copycat manufacturers selling self-balancing scooters to the US market, has increased substantially. In 2012, Ninebot was founded as a privately held company headquartered in Beijing, China. In 2014, Segway sued five self-balancing scooter companies in the U.S. for infringing upon Segway's patents, including InMotion and Ninebot. In April 2015, Ninebot acquired Segway and obtained more than 400 patents for Segway's nearly 10 products in three product series. In 2017, InMotion acquired SoloWheel and obtained its corresponding patents.

Growth phase: The rise of the leisure sports market. Although self-balancing scooters were originally intended to revolutionize the way people travel, it has instead fueled the emergence of a leisure sports market. After acquiring Segway and obtaining support from Xiaomi, Ninebot launched an inexpensive product for consumers. In October 2015, the No. 9 self-balancing scooter was released and priced at CNY1,999. Self-balancing scooters have come to be used as leisure sports products for children and young people, and sales have grown by leaps and bounds. Ninebot began to focus on the leisure sports market by introducing electric skateboards, kart kits, and children's bikes. In 2018, the portion of revenue from electric skateboards and self-balancing scooters was 66% and 29%, respectively, year-on-year changes of +42ppt and -45ppt, respectively.

Regulatory approval: Self-balancing scooters are permitted on sidewalks in certain U.S. states In China, self-balancing scooters are only permitted on enclosed roads and indoor stadiums. Self-balancing scooters require a high-level of coordination, and pose some unique risks. In 2010, the owner of Segway Jimi Heselden died after driving his own product off a cliff.

4. Companion robots: AI-assisted NLP facilitates easier human-machine interaction, and 5G promotes the development of cloud robotics.

There are two development paths for companion robots. The first involves such companies as Sony and Sharp, which harness cutting-edge technology to produce bionic robots with joints, mainly for the purposes of accompanying the elderly, teaching children, and entertainment and leisure. Other companies, such as Samsung and Amazon, have launched display robots, with the goal of building a robot OS platform to provide a wide range of software and AI services.

There are two different types of companion robots. One is small robots, a category which includes Sony's AIBO robot dog, Sharp's robot cat, and Tomotaka Takahashi's Robi. The main function of these robots is interacting with users, providing companionship and education, and communicating with others. Most of these robots have been commercialized.

Sony's AIBO robot dog

The other type of companion robots is large humanoid robots, the most famous of which include Honda's ASIMO and Boston-Power. This type of robot tends to come equipped with two feet, making it capable of climbing or descending stairs, grasping objects, interacting with human, and performing other tasks.

Honda's ASIMO

5. Medical robots: AI and 5G are in high demand for medical diagnosis, surgery, and physical rehabilitation

Capsule robot: Used in conjunction with automatic diagnosis systems, capsule robots can automatically determine potential health issues, based on the collected images, providing doctors with invaluable information. This cloud-based solution also enables remote and multi-site diagnosis.

Capsule robots

Surgical robots: Current surgical robots include robotic arms, navigation robots, and master-slave robots.

Surgical robotic arms

Navigation robots: Plan the operation path for the surgeon, and provides advice during the operation. Surgery navigation is currently divided into two types: optical navigation and electromagnetic navigation. Optical navigation has a high degree of precision and is not affected by electromagnetic interference from other devices, but its optical path is easily blocked. Electromagnetic navigation is flexible and has low requirements with regard to body position, but it is susceptible to electromagnetic interference.

Optical navigation is mainly used in neurosurgery, spinal surgery, joint surgery, and maxillofacial surgery. Electromagnetic navigation is mainly used in intracranial biopsy, catheterization, and bronchoscopy.

Master-slave robots: Provide technical support for surgeons during remote surgery and off-stage surgery.

The da Vinci robotic surgery system is a master-slave robotic system that helps doctors perform off-stage and potentially remote surgeries, reducing the presence of medical error caused by fatigue and loss of consciousness.
Body movement rehabilitation robot. Most current rehabilitation robots are rehabilitation robots for upper limbs and exoskeleton robots for lower limbs.

6. Commercial retail robots: Increased popularity of indoor unattended distribution robots

In recent years, many robots have entered the commercial retail field, including supermarket shopping robots and hotel water and meal delivery robots. These indoor distribution robots have helped improve the operations efficiency of commercial retail, reduce labor costs, and instill everyday services with fun, engaging attributes. Such robots have proven popular among users. As labor costs continue to rise, commercial robots have potentially broad applications, and are worthy of attention.

Interactions between shopping guide robots and humans

7. Bionic robots: Incorporated into the human body

Consciousness-controlled robots: Cutting-edge robotics research has demonstrated the feasibility of consciousness-controlled robotic arms. A robotic arm can be controlled after being connected to the human brain system to perform such movements as grabbing beverages, after trainings.
Bionic robots: Scientists have developed robots that mimic human organs and use awareness controls to assist disabled people, providing a huge potential market.

Consciousness-controlled robots

5G+AI technologies have impacted a wide range of industries, but it has the broadest and most promising implications for robotics. AI will help traditional robots evolve from machines unable to collaborate with human beings, to those able to serve human beings, and ultimately into intelligent decision-making units, capable of cognition and drawing inferences. The increasing maturity of 5G technologies will further expand the application scope for robots. Thanks to its low latency, efficient, and broad coverage connections, 5G is able to provide more computing power and storage space for robots, to facilitate the sharing of knowledge. In the future, with continued technological improvement, daily life will become more enriching and convenient, as evidenced by the use of robotic vacuum cleaners and logistics robots. High-level robots, as seen in science fiction movies, are just around the corner, and promise to benefit society in truly groundbreaking ways. (Source: Electronic Engineering World)

[About Us] HiSilicon (Shanghai) Technologies Co., Ltd. is a leading fabless semiconductor design company that offers innovative chipsets and smart city, home, and travel solutions, with 12 R&D centers, and a presence in more than 100 countries and regions.