When 5G connects to autonomous driving, what will happen to future urban transportation?

A top-of-the-line BMW hybrid sports car parked at Intel's booth at CES 2017, with its doors wide open, showing Intel's latest achievements in the field of autonomous driving. Facing this hot concept in the technology field recently, Intel has a more practical understanding of it - autonomous driving should go hand in hand with 5G networks, and building an intelligent traffic environment can make driverless driving a reality, rather than living in a PPT.

In fact, this is not the first time Intel has brought a top sports car to the CES booth, but this BMW i8 is the epitome of autonomous driving. It is equipped with the latest Intel® GO™ autonomous driving solution and the latest 5G modem. It can be said that this is a self-driving car that has traveled through time and will be available in the next five years. In the next few years, there will be more self-driving cars on the road. The mature communication system relies on the high data speed built by the high-speed and stable network to help self-driving cars connect and think. Intel is using its powerful technical accumulation to continuously upgrade autonomous driving technology:

• Intel's status quo on autonomous driving and 5G development

At the CES 2017 press conference, Intel demonstrated the latest technological achievements of its cooperation with BMW and Mobileye for more than half a year. Executives of the three companies demonstrated a jointly built prototype car at the press conference. BMW Group is responsible for making the car run elegantly, as well as specific production and future sales; Mobileye contributed its proprietary EyeQ®5 high-performance computer vision processor, providing automotive-grade functional safety and low-power performance; the deconstructed 360-degree panoramic visual information transmits data to the Intel®GO™ computing platform, which is also the first complete solution launched by Intel for autonomous driving.

Intel® GO™ driverless solution provides world-class processor and FPGA technology, balancing performance and power consumption in the most efficient way. It can serve as the brain of driverless cars and has the ability to learn and upgrade itself. Through the thinking and learning of this car brain, the car can achieve sensor aggregation, driving strategy, environmental modeling, path planning and decision-making. In the remote data center, the artificial intelligence platform equipped with Intel® Xeon® processors, Intel® Arria® 10 FPGAs, Intel® Nervana™ and Intel® SSDs provides a powerful training and simulation infrastructure for the machine learning and deep learning required by the driverless industry.

In short, Intel® GO™ is like having a reliable driver for your car, and this driver can constantly learn the latest driving techniques. Intel has endowed this car with an artificial intelligence brain that simulates the human neural system, giving cars on the road human intelligence.

At the same time, Intel has considered the future of driverless driving and has prepared 5G modems in advance. The new Intel® 5G modem can help global companies develop and launch early 5G solutions. It will accelerate the development of 5G-enabled devices and help leaders in all walks of life use these early deployments to achieve innovation, including driverless cars, because a driverless car will generate 4TB of data traffic every day. Current communication systems cannot meet the huge bandwidth required to achieve this evolution, or support the ultra-low latency required for these devices and even vehicles to respond to instantaneous events in a timely manner.

The modem's baseband chip will work with a new 5G transceiver that supports sub-6 GHz bands and millimeter wave capabilities. This powerful combination uses key 3GPP5G NR technologies, including low-latency frame structure, advanced channel coding, and massive multiple-input and multiple-output, to provide faster connectivity and extreme responsiveness. Currently, this product mainly supports early trials and lays the foundation for accelerating the development of related products. These products will support 3GPPNR technical specifications and help promote the unified global adoption of 3GPP5G standards.

• What specifically can 5G do for autonomous driving?

Self-driving cars are like driving on data every day. This relies on a large number of sensors, including cameras, lidar and radar, which can judge and analyze the environment around the vehicle. For example, an unmanned car is driving on the road. Suddenly, the camera captures an object in front of it, and after contour recognition, it is determined that it may be a person; at the same time, the radar can detect the three-dimensional dimensions and speed of the object, and identify real people and human-shaped signs. Then it determines the next instruction to be executed, whether to bypass the obstacle or stop and wait for it to pass.

However, this is not the end. The driverless car will also record the entire process and upload it to the cloud data center for analysis and storage. It will receive all the data, use deep learning tools, and create communication instructions with other vehicles as a learning process. The next time a similar situation occurs, the driverless car will follow the same routine and continuously upload and summarize, thereby improving the reliability of autonomous driving. The upload and download here require a high-speed and reliable network as a support.

• The importance of 5G networks for the development of autonomous driving

Kevin Hattendorf, head of Intel's autonomous driving product marketing, believes that each vehicle is an individual in this complex ecosystem, and communication between vehicles and roadside infrastructure, including data centers and networks, is of paramount importance.

Unlocking the true potential of autonomous driving requires a stable, large, and widespread wireless network. 5G networks will be the foundation for solving this problem, and Intel expects it to be available starting in 2020.

As the next generation of wireless communication network, 5G can transmit data at high speed, with low latency and stability. It will become the cornerstone of the growth of the two hot fields of Internet of Things and artificial intelligence. However, autonomous driving is also an emerging application that spans these two fields. In addition to autonomous driving, smart buildings, smart cities, vertical industrial manufacturing, networked wineries, etc. will all be derivatives of the Internet of Things and artificial intelligence. From this description, the secret relationship between 5G and autonomous driving has surfaced.

On the other hand, autonomous driving will generate a large amount of data. Intel CEO Brian Krzanich showed a set of data: each autonomous driving car will generate 4,000GB of data every day. Yes, 4,000GB! Now most of our mobile phone monthly traffic is 1GB or 2GB a month, which is equivalent to at least 800 months of total traffic. However, the current downlink speed of 4G network is 100Mbps~150Mbps. The unit here is lowercase b, and the conversion relationship of 1B=8b. The conversion is 12.5MB/s-18.75MB/s. 4,000GB of data requires at least 200,000 seconds, or 59 hours. From the data, it can be seen that the current 4G network cannot meet the needs of autonomous driving cars in terms of traffic and network speed.

The above analysis is actually based on the case of only one unmanned car. However, the popularization of autonomous driving will definitely involve millions or tens of millions of cars running together. This also poses a greater challenge to the overall bandwidth and stress resistance of the network, so the 5G network is about to emerge.

• Conclusion:

It is no exaggeration to say that the bright prospect of a fully driverless car industry has already appeared in front of manufacturers in various fields such as technology and automobiles. In the next decade or decades, most people who read this article will be able to experience the travel experience brought by the intelligent transportation revolution. Intel is at the core of the driverless field, and the results it has achieved are exciting. Driverless cars will completely change personal transportation in ways we can't imagine, and Intel will continue to participate in this change.