Chip card neck, how will Chinese intelligent robot crack?

With the rapid development of artificial intelligence, the robotics industry ushered in the spring of development. According to statistics, the global service robot market will grow rapidly to 17 billion US dollars in 2020, of which China will occupy more than 40% of the global service robot market. But at present, the supporting facilities of the robot market are not perfect. Take chip supply as an example, the former mobile phone chips or other hardware chips are often difficult to adapt to the robot products.

Intelligent Core of Robot

Over 95% of smart phones use ARM processors. The chips authorized by ARM are mainly used in mobile computing, smart cars, security systems and the Internet of Things. Therefore, the British chip giant ARM has stopped cooperating with Huawei in accordance with the new US regulations and has hit Huawei's chips tremendously.

Unlike the mobile phone industry, robots use chip technology in physical positioning, in-depth learning algorithm, visual recognition, processing and transmission, planning and execution. Robot parts usually include five parts: control center, power supply, sensor, action and feedback, fuselage. In intelligent robots, chips can play a vital role. Generally speaking, several supporting chips will assemble the interfaces and then connect them to a microcontroller. Supporting chips can also pre-process signals, such as signals from sensors and buttons, so that the workload of the control center can be reduced. This is the basis of robot's quick response.

At the same time, many other components, such as circuits, connectors, capacitors, resistors, diodes and other electronic components, are also required to play an important role in the overall connection of the circuit. Some robots have control centers separate from robots, such as robots that require people to control through remote controls or game handles. In such robots, the simpler chip controls only a single component (such as a leg or an arm), and the chip does not know what the rest of the fuselage is doing. Therefore, for robots, a good chip does not play a single role in control, but the overall rate of increase.

For the control center of the robot, the best choice is the microcontroller chip. Microcontrollers are very similar to microprocessors, and they all appear in personal computers. The difference between microcontrollers and microprocessors is that the former is more like a complete microcomputer integrated into a chip. The memory and storage space of microcontroller are relatively small, both of them are directly embedded in the chip. Microprocessors in personal computers connect pins to high-speed memory, and microcontrollers have many different kinds of input and output pins. These pins can be directly connected to sensors, buttons and other strange parts.

Frequent action of robots with major chip faucets

Not only Qualcomm, but also the world's leading chips are locked in the robot market. For example, Intel, Invidia, etc.

In recent two years, Intel, a giant of computer and server chips, has shown a decadent trend. Its core business and some products such as mobile phone processors are not very prosperous in the market. Therefore, Intel has also begun to catch up with the trend, exploring, exploring and attacking in the fields of smart home, robots, automatic driving and so on.

As early as a few years ago, Intel Lab's internal project, Open Source Robots for the 21st Century, was launched by Intel to enable every customer to make their own robots at a low cost. Jimmy's 3D printing humanoid robot is Intel's achievement in this project. Later, Intel independently developed Edison, a general computing platform designed for aircraft, intelligent robots, wearable devices, and RealSense RDK, a tool kit for developing real-sense technology robots. Not only that, Intel has also acquired and invested in a number of companies and manufacturers of UAVs, chips and machine vision.

Another chip giant, Yingweida, recently launched Nvidia Isaac on a robotic platform to support the next generation of autonomous driving machines, which play a decisive role in manufacturing, logistics, agriculture, construction and other industries.

The core of the Nvidia Isaac Robot Platform is Jetson Xavier, which contains 9 billion transistors and is said to be capable of up to 30 TOPS (trillion times per second).

In addition, Jetson Xavier's processor includes Alta Tensor core graphics processing unit (GPU), eight-core ARM64 central processing unit (CPU), a dual NVDLA depth learning accelerator, an image processor, a visual processor and a video processor, which enables dozens of algorithms to be processed simultaneously in real time for sensor processing, measurement, positioning and mapping, vision and video processing. Perception, plus path planning, etc.

This kind of high-performance processor is particularly important for robots. Whether it is the ability to acquire information from sensors, locate, perceive the surrounding environment, recognize and predict the movement of nearby objects, and make self-judgment to acquire data, it is crucial.

At present, many robotic chip products are being developed. Whether LG plans to cooperate with Qualcomm in developing clean robots and robotic assistants for use in airports, or Orion Star and Jingdong plans to launch enterprise-class service-oriented robots, a large number of robots equipped with "core" will be on the market.

It is urgent to cultivate talents

With the withdrawal of supply from Huawei by the United States, the sense of crisis of technology among domestic enterprises is increasing day by day, and the sense of self-improvement of science and technology is also increasing. Chip supply failure is also a good opportunity for the rise of domestic chip brands. However, statistics show that the talent gap of domestic chips is over 300,000 in two years, with an average annual increase of over 100,000 talents, which doubles the demand of the existing talent stock. What causes the big gap of chip talents?

1. Chip growth is slow, iteration cycle is long, and talent training is slow.

Slow growth and long iteration cycle are one of the important reasons that restrict the cultivation of chip talents. After entering the job, the new students usually go through four or five chip project cycles, each of which lasts from half a year to two years, before they can "start to take the lead". It will take years to get to a higher level of technology.

2. Fast growth of IC industry market scale

It is reported that by the end of 2017, there are about 400,000 talents in China's integrated circuit industry, with a talent gap of 32,000 and an average annual demand of 100,000. Simply relying on Colleges and universities can not meet the demand for talent supply. In addition to continuing to strengthen the training of talents in Colleges and universities, it is necessary to promote the application and construction of microelectronics and integrated circuit related first-level disciplines, narrow the gap between the training of talents in Colleges and universities and the demand of employers in enterprises, and promote the "supply-side structure reform" of integrated circuit talents.

3. The United States is out of supply and domestic chips need to be self-sufficient.

In the US crackdown on Huawei, domestic entrepreneurs have seen the fact that core technologies need to be mastered by themselves, while in today's global shared economy, China has no allies to master chip manufacturing technology. However, the United States has suppressed and blocked China with advanced technology and many allies. After the outage, its influence continued to ferment. It also alerts Chinese entrepreneurs.

4. In recent years, the salary of chip talents is lower than that of software talents.

In fact, in addition to the lack of talent, the high brain drain rate in IC industry has become a common problem in the industry. Over the years, the low salary environment has led to low salary expectations for chip practitioners. Most graduates of integrated circuit majors prefer to go to the Internet, computer software, IT services, communications and real estate industries.

In 2017, less than 30,000 of the 200,000 graduates of integrated circuit majors in Colleges and universities entered the industry for employment, and relying solely on Colleges and universities can not meet the demand for talent supply.