Looking back at 2017, although the technology sector as a whole experienced a major boom, there is no doubt that the most watched was the rapidly advancing artificial intelligence. Driven by technological breakthroughs, market growth, and policy implementation, capital poured into the AI field. Major global tech companies continued to increase their investments, competing to strengthen their comprehensive layouts. In 2017, AI shone brilliantly under the spotlight.
Intelligent robots, as the ultimate embodiment of AI hardware, represent a sub-field of applied robotics that is advancing vigorously. From education to commercial use, with multi-scenario deployments, what is the business logic behind this booming industry?
Intelligent Robots: More a Test of Mechanical Capability Than a Battle of AI
Like every popular industry, when the waters are muddied, quality inevitably varies, and there are countless pretenders.
For startup companies, entering the AI robotics field makes it easier to gain investor favor than entering other fields.
For listed companies, currently at the convergence of the first and second stages of AI robotics, pure concepts and pure concept speculation are everywhere. The stock price increase of a company solidly engaged in robotics and a conveyor belt company labeled as "robotics" is actually quite similar — this shows the situation clearly.
Specifically, the main problem in the AI robotics industry is: severe assembly-line production and "borrowed" innovation.
AI robots are mainly divided into two parts. One is the software, i.e., the AI system, where the main interaction methods are voice and image interaction. For many companies jumping on the bandwagon, they lack the ability to independently develop voice recognition. In such cases, they typically use the technology of specialized voice recognition companies like iFlytek.
Unlike the smart speaker industry, where the technical requirements for speaker hardware are not high, the robotics industry demands extremely high hardware standards. Behind the wave of intelligent robots in China lies the constraint of significant "shortcomings" in areas such as upstream high-end material research and development and core component manufacturing. The manufacturing of intelligent robots is extremely complex — from making a small bearing or screw to developing high-end intelligent chips and AI algorithms — all require long-term, cumulative technological and industrial progress.
In the author's opinion: "For the future development of China's intelligent robot industry, the most critical thing is to continue promoting the development of the real economy and create a better environment for manufacturing. Only then can world-class manufacturing enterprises emerge."
The hardware part of AI robots is far from simple. For example, finger automation — the ability to grasp objects like a human — is extremely technically challenging. If AI robots cannot achieve automation of finger joints, it will be difficult to apply them to specific scenarios, and they will remain mere toys for self-satisfaction.
Achieving this is not easy for China in the short term. The root cause is still the underdeveloped industrialization level and a severe shortage of domestic hardware engineers. China has a vast number of software engineers, but in hardware, there is a considerable gap compared to advanced manufacturing countries like the United States, Japan, and Germany.
Therefore, for AI robots, the "borrowed" approach in software means that the competition among enterprises is not very meaningful. Many companies derive their technology from the same source; ultimately, what is being competed on is the manufacturing capability of the robots.
For the booming intelligent robot industry, it is more a test of a company's mechanical capabilities and intelligent manufacturing level than a battle of AI.
Undercurrents in Educational Robots: Policy Ceilings Guide the Industry Direction
Currently, the main implementation scenarios for intelligent robots are two: the education market and the commercial sector. In this part, we analyze AI robots in the education context.
According to data released by the China Robot Education Alliance in 2016, there were about 7,600 robot education institutions nationwide, a nearly 15-fold increase in the last five years. This year, this number will only increase. With the "National Medium and Long-Term Education Reform and Development Plan Outline (2010-2020)" highlighting innovation as a key focus, robot education has gradually entered primary and secondary schools and children's palaces. At the same time, the "General Competition Rules for China's Quality Sports Robot Movement" was officially promulgated, providing regulatory backing for various robot competitions, and the industry has entered a fast track.
Quality education has been advocated for many years. As early as when the author was in primary school, in the late 1990s, the country was already promoting quality education. Now, with the improvement of national comprehensive strength, in many first- and second-tier cities, extracurricular art training classes and advanced high-tech teaching materials have successively entered classrooms. The era of quality education has fully arrived, providing fertile soil for the development of the AI robot industry in educational contexts.
AI robots can take root and flourish in education because, on one hand, they cater to quality education by cultivating children's hands-on abilities. On the other hand, for robots to fully break through, they need to be completely linked to education — that is, performance in robot competitions must be linked to exam-oriented education. Recently, the Ministry of Education issued new regulations adjusting the registration requirements for independent enrollment, retaining only two categories: academic talent and innovation potential. The value of various robot competitions has further increased.
As a result, many parents are enrolling their children in robot classes for keywords like competition, special skills, and bonus points. This has led to parents' enthusiasm for K12 technology education being much higher than college students' enthusiasm for participating in technology competitions. Robot education is booming just like the former Olympic math competitions.
Domestically, companies prominently featuring programming robots include UBTECH's Jimu robots and Makeblock.
In the implementation of AI robot education in educational scenarios, intelligent robots also face an awkward situation. If robots are to be linked to exam-oriented education, the most likely assessment basis would be programming ability, rather than robot assembly or robot competitions. If it is about programming ability, its essence is not directly related to robots, bringing us back to the software part.
Furthermore, the implementation of intelligent robots in educational scenarios is overly dependent on policy. If not linked to entrance exams, it will be difficult to achieve significant progress, and they will be reduced to toys under the guise of teaching tools. At that point, they will be like shooting stars — brilliant for a moment, but soon falling, and the entire educational robotics industry will enter a quiet period.
Commercial Robots in Full Swing: Opportunities and Challenges in the Enterprise Service Market
Another main implementation scenario for AI robots is commercial settings. After recognizing the pseudo-prosperity of the 2C market driven by subsidies and conversion rates, in recent years, more and more capital has shifted its focus to the more valuable and less developed 2B market. With the prosperity of business models in the TOB market, enterprises' service scenarios are also showing diversified service extensions. From SaaS systems to the "cloud"布局 by giants, the software data direction has seen flourishing developments both domestically and internationally. With the upgrading of enterprise services, AI robots integrating hardware and software are gradually becoming new partners in the enterprise service market.
AI robots are a synthesis of various technologies. Various enterprise office service software is merely a basic application of big data. Robots, as the crystallization of software-hardware integration, should represent a higher dimension of enterprise service. Service robots have also received attention from the state and related organizations. For example, the China Robot Industry Alliance recently announced that it will successively release three alliance standards and 17 robot industry alliance standards. These signs suggest that the prospects for service robots are extremely promising in the foreseeable future.
With rising prices and increasing living costs, workers' wages are also rising, especially in first-tier cities. More importantly, once wage levels increase, they do not decrease, and labor costs will only get higher. Therefore, for enterprises looking to reduce operating costs, commercial service robots will undoubtedly play a key role.
The vast enterprise service market was previously mainly concentrated in the software field. Now, the advent of commercial robots has expanded the boundaries of this market. However, at the same time, commercial AI robots also face considerable challenges.
First is the high price threshold. Rising labor costs are an indisputable fact. On the surface, replacing humans with machines does save labor costs, but it also adds a new cost — the cost of the robot. Taking a small restaurant as an example: in first- and second-tier cities, the monthly salary of a food server is between 3,000 and 3,500 yuan. Many commercial robot companies can customize products for restaurant use, but the price is at least 50,000 yuan or more. The price of a single machine exceeds a server's annual salary. Considering the robot's usage time, maintenance costs for breakdowns, and other issues, robots have not reduced labor costs. Moreover, robots depreciate, and with Moore's Law, restaurants that are the first to introduce robots may not gain an advantage and might be better off waiting.
Another hurdle restricting the development of commercial robots is human-robot interaction.
The most critical technology for commercial robots is "smart voice." If we cannot give robots the ability to speak, they will be perceived as unintelligent in the eyes of consumers.
Currently, various commercial robots can be seen in banks, shopping malls, and homes. When people actually interact with these robots, they find that their voice capabilities have not yet reached the point of normal communication. Especially in noisy public places like shopping malls, the voice systems of robots can be very disappointing.
During laboratory testing, a robot's voice system may meet interaction requirements with voice recognition accuracy above 90%. However, once a customer signs off and it is applied in a public place, it becomes like "a chicken talking to a duck." This problem troubles many robot companies.
This technology is known in the industry as remote dynamic microphone array sound pickup technology. Its application in China's robotics industry is not yet mature.
Behind these technical pain points lie the enterprise's R&D strength and technological innovation capability.
Currently, the phenomenon of homogenization in appearance and functionality among commercial robots on the market is quite serious. One reason is the lack of core technology — products rely merely on a patchwork of concepts. Another reason is deviation from market orientation — design planning does not proceed from user experience or application scenarios.
Commercial robots are not just smart hardware; they are comprehensive technical solutions. They need to offer more customization freedom for different consumption scenarios of industry clients, carrying part of an enterprise's values. They should solve practical problems for traditional government and enterprise clients, create more value, and even bring revolutionary user experiences.
Intelligent robots at the forefront may seem glamorous, but undercurrents are surging. Where will the path to advancement lead in this period of uncertainty? Will 2018 see a major breakthrough? We will wait and see.
