Driverless vehicles park themselves in rapidly expanding market

Encouraging prospects attract heavy investment to sector

Sun Jiabei only started to drive in earnest in December, more than three years after she received her license.

"I used to hail a taxi if my father could not drive me to work and home," said the 30-year-old, who works as an office clerk in Zhengzhou, capital of Henan province.

"Driving was easy, but parking was so difficult for me, and like many other new drivers, I found it a different experience to what we were taught at the training school. I became anxious when there were other vehicles behind me," Sun said.

She added that this was why she quickly decided on an electric sport utility vehicle, or SUV, from Chinese carmaker WM Motor when her family was considering trading its 10-year-old Volkswagen Sagitar late last year.

"The salesperson told me the SUV could park itself. That was a big relief and was far more important to me than other things at the time," Sun said.

Sun's SUV can be easily parked, but only in clearly marked parking lots so that its cameras and radar can detect any obstacles before the vehicle steers, reverses and comes to a halt.

"Without these guidance lines, it doesn't know what to do and I have to do it myself, but fortunately there are now more delineated parking lots. I have also become better at parking," Sun said.

The WM Motor model is one of dozens of vehicles with similar functions available in the Chinese market from domestic and international carmakers.

Some of these vehicles allow drivers to get out before driving themselves to parking lots, before returning to collect owners when summoned with the touch of a phone.

According to Chinese carmaker GAC Aion, such tasks can be performed when a driver is within 150 meters of a vehicle, which has to be driven at least once in the parking lot so that it can remember the way and the pickup point.

WM Motor and GAC Aion use Chinese internet giant Baidu's smart driving solutions, while other carmakers, including Nio and Xpeng, have developed their own.

Such functions are classified at Level 2 in the Society of Automotive Engineers International's six-level system of driving automation, with Level 0 designating no automation and Level 6 fully automated.

Level 2 functions, including automatic parking and cruise control, are the highest level of automation that carmakers worldwide are currently allowed to offer. Known as the advanced driver-assistance system, or ADAS, these functions are fast becoming standard in new vehicle launches in China, especially in electric models.

Qin Lihong, president of Nio, a startup that is listed in New York, said that the survival of carmakers in an increasingly competitive market depends on their ability to produce smart vehicles.

Nio's ET7 sedan, which started deliveries late last month, boasts 33 sensors, including cameras and radar, and also features four Orin chips designed by United States company Nvidia for vehicles with a high level of automation.

The car can navigate itself on expressways and downtown roads and also self-park, according to Nio.

Mercedes-Benz said it is planning to introduce its Level 3 technology in China and the US from Germany, where the technology has been certified by local authorities.

Last month, the carmaker opened a research and development center in Shanghai, which will work in fields such as connectivity, automated driving and big data.

Consulting company IHS Markit said smart driving is gaining traction, especially in China, adding that new cars with functions at or above Level 2 will comprise at least 45 percent of the nation's market in 2025, and more than 80 percent in 2030.

These promising prospects are encouraging automakers to invest heavily in the sector. SAIC Motor, the Chinese partner of Volkswagen and General Motors, said it is allocating 300 billion yuan ($47 billion) to develop smart vehicle features, including functions at Level 2 and above.

Great Wall Motors, China's largest SUV maker, has spun off Haomo, an arm dedicated to autonomous driving.

Gu Weihao, CEO of the two-year-old startup, said its ADAS system will be available in 34 models this year, up from five Great Wall Motors models last year. He added that more than 1 million passenger vehicles will feature this system in the next three years.

Haomo also produces driverless Level 4 logistics vehicles, some of which are delivering groceries to residents in Beijing and also e-commerce parcels to university campuses in Wuhan, Hubei province, and Hangzhou, Zhejiang province.

Like Haomo, Baidu is betting big on Level 4 technology, which means that under most circumstances vehicles can navigate themselves.

The internet giant, which is based in Beijing, is targeting the robotaxi market. Apollo Go, Baidu's autonomous ride-hailing service, provided some 213,000 rides free of charge in the fourth quarter of last year, more than double the number in the third quarter.

The service is now available in eight cities, including Beijing and Shanghai, where users can hail a robotaxi via a smartphone app. Baidu CEO Robin Li told the media last month that the service will be available in 65 cities by 2025 and 100 cities by 2030.

Meanwhile, the authorities are rolling out policies to explore opportunities in the sector.

In November, China's first pilot area for commercial robotaxi fleets was launched in Beijing, with paid services authorized in an area of 60 square kilometers in the suburbs.

Baidu and the Toyota-backed startup Pony.ai were among the first to receive permits to offer such services in the pilot area.

Lou Tiancheng, Pony.ai co-founder and chief technology officer, said its virtual driver is "now equal or superior to a human driver" in most circumstances.

"We're confident about our autonomous vehicle technology readiness as we rapidly move toward robotaxi and robotruck commercialization and mass production," he said.

Zhang Xiang, a researcher at North China University of Technology in Beijing, said the pilot area is hugely significant in further accelerating the large-scale commercialization of autonomous driving technology nationwide.

He said other cities, such as Shanghai, along with Shenzhen and Guangzhou, Guangdong province, may soon follow suit in setting up similar commercial pilot areas for self-driving services.

IHS Markit said the value of the autonomous car service market will exceed 1.3 trillion yuan by 2030, accounting for 60 percent of the nation's ride-hailing market that year.

Li, the Baidu CEO, said China leads the world in self-driving technologies, but the absence of related policies is limiting the sector's development.

"The country that issues more innovative policies and takes the lead in realizing large-scale commercialization of autonomous driving will gain an upper hand in international competition," Li said.

He called for efforts to speed up the revision and implementation of traffic laws to bolster the large-scale commercial application of self-driving vehicles.

Li said that within a decade smart transportation can help end traffic jams in big cities. Smart transportation refers to a system that integrates elements such as autonomous vehicles, smart infrastructure, artificial intelligence, 5G and cloud computing.

In addition to congestion, Li said smart transportation can reduce road traffic accidents by 90 percent and cut carbon emissions.

By 2025, China is planning to realize scaled production of vehicles capable of conditional autonomous driving, along with the commercialization of highly autonomous vehicles in certain circumstances, according to a blueprint issued by the National Development and Reform Commission.

Yang Diange, a professor from the School of Vehicle and Mobility at Tsinghua University, said Level 4 self-driving technology will initially be deployed in taxis and trucks in designated areas. The large-scale application of such technology might be seen in private vehicles by 2030.

Chipmakers' chance

The growing popularity of autonomous driving technology worldwide is creating new opportunities for chipmakers.

Ondrej Burkacky, a partner at global management consulting company McKinsey, estimates that chips used for autonomous driving functions are expected to generate $29 billion in revenue by 2030, up from $11 billion in 2019.

Nvidia, a leading maker of graphics chips and a hardware supplier for vehicle infotainment systems, is looking increasingly to software and chips that underpin autonomous driving systems.

On March 22, the company said it had started delivering its autonomous vehicle chip Orin, which will be used by Chinese electric vehicle maker BYD and US startup Lucid Motors for their next-generation fleets.

To date, Nvidia has more than 25 carmakers and autonomous driving companies as clients. Jensen Huang, the company's CEO, said, "Auto is on its way to being our next multibillion-dollar business."

Qualcomm, which specializes in cellphone communication chips, is taking over Swedish automotive technology company Veoneer to help accelerate its auto business. Veoneer makes vision systems, radar and software for autonomous driving systems.

Chinese chipmakers are exploring the market, with Horizon Robotics, based in Beijing, becoming a rival to global giants. In the second half of last year, the company unveiled Journey 5, its latest auto-grade chip, which is designed for Level 4 functions, with up to 128 tera operations per second of artificial intelligence computing power.

Chinese carmakers, including Great Wall Motors, SAIC Motor, Changan and Li Auto, have shown an interest in the chip.

Yu Kai, founder of Horizon Robotics, said the Chinese market is the most competitive, adding that the first model to boast Nvidia's Orin chip is in China, as is the first one equipped with Qualcomm's Snapdragon Ride, a car-based computer system that can equip a vehicle with anything between Level 1 and Level 5 autonomy.

"We are nervous and excited at the same time. It is like a race, but it is not a local race in Beijing or Shanghai. It is a global race, and we would like to become No. 1," Yu said.

Carmakers are investing in chipmakers. In February last year, Great Wall Motors became an investor in Horizon Robotics, while in December it led a financing round for Synlight Crystal, which is based in the Baoding High Tech Development Zone, Hebei province.

Siengine, a joint venture between chipmaker Arm and EcarX, and which is owned by Geely Holding Group, will launch its first auto grade seven-nanometer system-on-chip products in the third quarter.

These products will be used for smart cabin functions. The company said it will roll out two five-nanometer high-performance chips in 2024-25. One of them, with a computing capability of 256 tera operations per second, will be used for Geely's autonomous vehicles.

Alert needed

Due to rapid developments in the sector, terms such as "autonomous driving", "driverless vehicles" and "driver-assistance functions" are often misused.

For marketing reasons, carmakers sometimes describe driver-assisting Level 2 functions as "autonomous", a word that should be reserved for driverless vehicles at Level 4 and above.

Carmakers have not mass-produced Level 4 or even Level 3 private vehicles, and there is no legislation anywhere that allows such vehicles to reach the market.

Tesla, the first to offer ADAS, refers to such systems as Autopilot and Full Self Driving, but the company said, "No Tesla cars are fully autonomous today and they require active driver supervision."

Tesla and Nio vehicles have been involved in accidents when their ADAS was operating.

Dong Yang, vice-president of leading automotive think tank China EV 100, said carmakers should be more careful with high-level smart driving technologies.

"They must be objective. They must not brag about such technologies to sell vehicles. More important, they need to educate users about the new technology," Dong said.

Li Xiang, founder of startup Li Auto, said carmakers should adopt a standard naming strategy for driver-assistance functions. For Level 2 functions, the word "auto" should be banned, Li said.

Shi Jianhua, deputy secretary-general of the China Association of Automobile Manufacturers, said there may be disparities between carmakers and drivers in their understanding of ADAS features.

Carmakers know that their Level 2 ADAS features require human drivers. They also know that autonomous driving is the future, but they may exaggerate the use of these features prematurely, Shi said.

He added that legislators should draft standards and laws to keep abreast of auto sector developments.

"Legislation will regulate the sector and facilitate its healthy progress," Shi said.