China's self-driving vehicles on track to take global leadership position, ahead of US

"Snow is unusual in Shanghai," says Li Xiao, an engineer at the National Intelligent Connected Vehicle Pilot Zone, "so we have to take the opportunity, to test the vehicle in the most adverse weather conditions."

Alibaba confirms self-driving car tests, joining Baidu and Tencent in China's autonomous car race

Li and colleague Chen Dong are about to oversee a trial of a driverless electric bus, produced by Xiamen Golden Dragon Bus, at the Enclosed Test Zone, the first testing area for autonomous vehicles in China. Covering 5 sq km and capable of replicating a range of road conditions, the zone has a section of motorway, a tunnel to simulate the loss of a Global Positioning System (GPS) and Long-Term Evolution (LTE) - used for high-speed wireless communication - signals, and huge metallic structures holding canvases printed with photographs of old Shanghai.

Canvases printed with photographs of old Shanghai line the test zone.

The overall effect is dystopian, enhanced by the test cars of a number of companies, which are running around covered in colourful fabrics; the brands don't want to disclose their secrets, so photography is forbidden.

Snow has settled on the minibus' sensors, but "the vehicle is equipped with systems to analyse the surroundings", says an unfazed Li.

China formulates new policies for autonomous cars in bid to catch up to US

"It has a radar that can detect objects but it doesn't recognise what they are," he continues, as his colleague sets up the on-board computers. "That's the job of the high-definition cameras. They capture all around and a powerful processor assesses each object in the images to decide every move in real time."

Programmers will analyse all the data and refine the ways the vehicle responds to various conditions.

"In the near future, roads will also send information to vehicles," Li says. "Self-driving cars, buses and trucks will communicate both with the infrastructure and with other vehicles, making driving much safer than it is now."

Engineer Li Xiao next to the computer that drives the bus.

The pilot zone is already equipped with these systems; Dedicated Short-Range Communications (DSRC), LTE-V (the V stands for "vehicles") and Wi-fi signals are beamed down from huge towers, alerting traffic to red lights, upcoming intersections and the like. This helps the vehicles maximise fuel and time efficiency.

"Traffic jams will be a thing of the past," Li promises, with a broad smile.

The vehicle has a steering wheel, but "it's just for safety, because this bus has the highest level of autonomy. It drives itself and you could theoretically fall asleep at the wheel", Li adds, although he's not suggesting anyone should do that just yet.

"We have tested it at 60km/h, but not in these conditions and with people aboard," the engineer says.

It is unnerving to see the bus drive with nobody at the wheel, but the ride is smooth. The vehicle reads traffic signals but there are no other obstacles on the track today; the snow would damage the dummies scientists use to simulate pedestrians, cyclists and other vehicles.

Dummies used in the pilot zone.

"When the weather is suitable, we test the bus' behaviour during unexpected events," Li says, with a grin. "Humans don't always drive or walk the way they should."

They know this well in Jiading. Home to the Chinese Grand Prix and a host of automotive industries, the huge northeastern district was, on March 1, chosen to be the site for the public testing of autonomous vehicles. Two domestic car companies, SAIC and NIO, were given the first licence plates that allow the running of autonomous automobiles in real conditions, along 5.6km of public roadway, where the pedestrians are made of flesh and blood.

China to green light more self-driving car tests amid US caution after Uber fatality

"Shanghai is going to further accelerate testing, applica­tion, research and development of intelligent and connected vehicles," Huang Ou, vice-chairman of the Shanghai Munici­pal Commission of Economy and Informatisation, said dur­ing the ceremony for the opening of the public section of road, which is just a few kilometres from the pilot zone. "The city will seize opportunities, take the initiative to meet challenges, boost innovation and speed up the industrial development of high-end, electric cars and intelligent vehicles."

High-definition cameras affixed to the front of the autonomous bus capture the surroundings in real time.

Huang was expanding on a theme raised by Xin Guobin, vice-minister of industry and information technology, at the 2017 World Autonomous Vehicle Ecosystem Conference, held in November in Shanghai. "The development of intelligent and connected vehicles can improve transport efficiency and safety while meeting energy-saving and emission-reduction targets. China has the foundation and advantages to succeed in this sector," Xin was quoted by state media as saying. "We will create an environment suitable for the development of intelli­gent and connected vehicles, and accelerate the integration of the automobile, intelligent transportation and information technology sectors."

"The licence provides a legal basis for us to test intelligent and connected vehicles on public roads, and will greatly promote our research and development of automated driving systems," NIO president Qin Lihong said at the ceremony. Two cars bearing paper copies of the new licence plates drove through the streets for the media. "We need to be able to test cars in real conditions to assess their viability," Qin added.

Chen Hailin, deputy director of pilot zone in Shanghai.

"These are the first steps towards making China a pioneer in the development of autonomous vehicles," Chen Hailin, deputy director of the pilot zone, tells Post Magazine.

His project was drafted in 2014 and received the green light a year later. "The test zone was built with an initial budget of 500 million yuan in 2016, in cooperation with numerous companies," Chen says. "Huawei, for example, provides communications infrastructure, Beidou - the Chinese GPS - is in charge of navigation systems, and we also test elements developed by local universities such as Tongji. Up until the end of 2017, we've conducted 40,000 tests and worked with 30 companies."

China's Huawei modifies Porsche Panamera to prove its smartphone chip's AI capability

As we tour the pilot zone, staffer Fan Xiaoxu says, "The success of autonomous vehicles depends not only on the cars. We can also do many things with the infrastructure to help them run smoothly. Here, for example, we use different kinds of paint on the road."

She points at thick white and yellow markings on the asphalt, a repetitive diamond pattern giving it a rugged texture. "Autonomous systems read it much better and it helps them to not veer off," Fan says, as we near a covered passage used to assess how self-driving cars manage when cut off from navigation systems.

Thick paint creates a rugged surface that helps autonomous vehicles to stick to the route.

Before long, all of Jiading's roads will be equipped to meet the new needs. "We are hoping to open 100km of roads for this kind of vehicle by next year, and extend the project to the Hongqiao Transport Hub in 2020, when we expect to have 10,000 autonomous vehicles in operation," Chen Hailin says. "Ninety per cent of those will be used for public transport."

He also foresees driverless buses like the one Li is testing being operative at tourist sites by the end of the decade. And the government believes that, by then, all new private cars will have features similar to the autopilot already in place in some Tesla vehicles. By 2030, China expects fully auto­nomous vehicles to account for 10 per cent of car sales.

"We already see assisted-driving vehicles like Tesla, and we will soon have cars that drive themselves in certain conditions where autonomous driving is easier and less dangerous: in a traffic jam or on expressways," Dong Yang, executive vice-chairman of the China Association of Automobile Manufacturers, said during an automotive forum organised by the China Europe International Business School (CEIBS) in 2016. "In the future, vehicles won't even have a steering wheel. It will be a disruptive tech­nology. And, combined with the zero-emission engines, it will provide a clean, safe - 90 per cent of accidents are caused by human error - produc­tive and relaxing mobility."

Self-driving car industry faces critical test after first death

But there will be hiccups, the fatal accident caused by a Tesla in the United States on March 23 being an example. The driver, 38-year-old Apple engineer Walter Huang, died in California when his Model X was on autopilot and crashed into a divider that separated the carpool lane. The car caught fire and Huang did not escape.

Although the accident is still under investigation, Tesla - which has "never seen this level of damage to a Model X in any other crash" - remains committed to the development of autonomous vehicles.

Testing the driverless bus.

On March 18, an autonomous Uber car hit and killed a woman in Tempe, Arizona, while she was pushing her bicycle. Elaine Herzberg, 49, became the first pedestrian killed by a self-driving car and the accident prompted Uber to stop testing in real conditions.

"The system won't work perfectly until all vehicles on the roads are driverless. Safety will be an issue as long as they have to share the space with traditional cars," Olaf Kastner, former chief executive of BMW in China, said during the CEIBS forum. "And we will have to solve the question of responsibility, because under the Vienna Convention, the driver is always accountable. If there is no driver, will manufacturers be liable?"

Uber shuts down self-driving car project after pedestrian is killed in world-first fatality

Chen Hailin is not keen to talk about regulations, saying, "We first set guidelines under the current laws, and this means that 'drivers' [those being driven] will be held accountable in the case of an accident. When we are closer to embracing fully automated vehicles, the government will have to set new regulations."

Despite the obstacles, China wants to lead the world in driverless technology. "I believe we are in a good position to do so, because the country is already a pioneer in many of the technologies shaping autonomous vehicles," Chen Hailin says. "Chinese companies are developing the new 5G networks, and Shanghai will have them installed in 2020. Their capability will greatly improve communications between vehicles, and the city will keep investing in infrastructure adaptation."

One of the autonomous buses at the pilot zone.

Beijing too is betting on autonomous vehicles. On March 23, authorities in the Chinese capital gave Baidu permission to test its driverless vehicles on 33 suburban roads, covering 105km. "With supportive policies, we believe that Beijing will become a rising hub for the autonomous driving industry," Baidu vice-president Zhao Cheng said in a statement.

Baidu completes 5G autonomous driving test in China's showcase tech city

Baidu's autonomous vehicle was tested during last June's CES Asia technology forum in Shanghai, which showcased many of indus­try developments.

"We believe highly autonomous vehicles will be a reality on highways by December 2020," Gu Weihao, head of Baidu's self-driving programme, said at CES Asia. "There are a few reasons why China is a good place to test them. First, the mess in many streets makes driving here a challenging issue," he laughed. "Second, because people support this technology more than in other countries."

In a Ford survey conducted last year, 83 per cent of Chinese respondents said they were "hopeful about the future of autonomous vehicles". That figure was 81 per cent for Indians and 75 per cent for Brazilians, with the Americans, British and Germans the least optimistic, 50 per cent, 45 per cent and 44 per cent, respectively, answering in the affirmative.

Big data is essential to the success of self-driving cars.

Research from German automotive inspection and certi­fication specialists TUV Rheinland, published in February, found that 63 per cent of Chinese drivers believe self-driving cars will increase road safety - compared with 34 per cent in the US - and that 71 per cent of Chinese drivers trust the cybersecurity competence of carmakers, whereas only 41 per cent in the US responded positively to that suggestion.

"Interestingly, however, people's doubts tend to increase and trust in the technology decreases as vehicles' level of automation goes up," reads the report. "Only 11 per cent of the respondents in Germany and 15 per cent in the USA state they fear 'a deterioration of road safety' due to partial automation, while nearly half of those same respondents believe that road safety will deteriorate with the advent of completely driver­less cars. In China, only 24 per cent expect road safety to decrease in the case of driverless cars."

"The Chinese are tech savvy and want to experience new things," says Xu Bin, associate dean at CEIBS. "They are very open-minded when it comes to embracing new technologies. And Chinese companies with growing knowledge and human capital will help China leapfrog to the forefront of numerous industries."

Baidu's autonomous-vehicle mapping software at CES Asia.

At CES Asia, Gu said, "But we have to cooperate among companies to accelerate the development of the autonomous vehicle." That's why Baidu made public its self-driving car platform - the Apollo project - in a way similar to how Google has with the Android operating system for mobile devices.

"Apollo provides an open, reliable and secure software platform for its partners to develop their own autonomous driving systems through on-vehicle and hardware platforms," explains Baidu, on its website. "As participation grows, more accumulated data becomes available. Compared to a closed ecosystem - such as that of Tesla - Apollo can evolve faster, bring greater benefits to members."

This Intel-backed China start-up is building 'brains' for self-driving cars, powering surveillance cams

Big data is crucial; Baidu creates the database and car­makers use it for their specific needs.

"We have hundreds of cars going around all kinds of roads equipped with dozens of sensors," Gu said. "They record not only the behaviour of the vehicle, but also that of our driver and other drivers on the road. All this data is fed to the auto­nomous system so it can better predict what it will have to do in different situations."

On a screen, Gu showed how the decisions of a profes­sional driver (represented by a red line on a road) and those of the autonomous vehicle (a green line) are almost identical.

"The car's system learns from the best drivers. And the more companies test the Apollo, the better it gets," Gu said. "We still have to improve algorithms and voice commands, and we will need to drive for about 200 million miles [320 million kilometres] without making any mistake to fine-tune all systems. But in the near future, autonomous cars will drive better than humans."

The autonomous vehicle tested by the writer at CES Asia in Shanghai last June.

At Baidu's temporary enclosed circuit at CES Asia, we tried an autonomous vehicle, made by Chinese carmaker Great Wall. The Haval is an SUV equipped with dozens of sensors and an on-board computer. A steering wheel and pedals are still present, as (unneeded, it transpired) safety precautions.

"Go!" The "driver" needed only give a voice command for the car to start moving around the circuit, the software correctly reading traffic signals along the way. Baidu employees ran ahead to set signs and convince amused passengers that the vehicle would behave differently at the same point if it encountered a different sign.

"No entry", "Turn right", speed limits, traffic lights. All were obeyed - and the response was fast.

"We all know it can be done, but it's a bit scary to experi­ence it," admitted a local journalist accompanying us for the short ride. "I believe there is no stopping this technology," the young woman added.

But the car did come to a halt, controlled in the same way that it was started: with a vocal command. "Stop!"

This article originally appeared on the South China Morning Post (SCMP).

Copyright (c) 2018. South China Morning Post Publishers Ltd. All rights reserved.