Baidu move sees 'robocars' turning a corner

The phrase "back-seat driver" may soon take on a new meaning as companies race to embrace the technology to produce fully autonomous vehicles for use on China's road network. Cheng Yingqi reports.

The news that Baidu will participate in a citywide trial of autonomous vehicles signals the latest move in the internet giant's plans to become a major player in the burgeoning market.

Under a five-year agreement between China's biggest search engine and the city of Wuhu in the eastern province of Anhui, the company's autonomous cars, buses and vans will gradually be introduced to selected roads and zones before eventually being allowed to use public highways throughout the city. The vehicles will not carry passengers during the first three years of the trial period.

The agreement has seen Baidu potentially steal a march on its rivals in the sector, which was kick-started by Google's self-driving-car program in 2009. Since then, automobile manufacturers such as BMW, Volvo and Toyota have all become leading lights in the race to commercialize so-called robocars.

Now, as basic self-driving technologies become more mature, companies in the world's largest auto market are anxious to deepen their involvement in the sector.

"Currently, the major bottlenecks in replacing human drivers with robots lie in cost reductions, the upgrading of the transportation infrastructure and legal issues. China, which has a large automobile market and higher acceptance of self-driving technology, will have a competitive advantage," said Jack Yu, general manager of ResearchInChina, an auto industry consultancy in Beijing.

The race is just starting. Ahead of the Beijing Auto Show, which ran from April 25 to May 4, Chang'an Automobile sent two driverless cars on a six-day, 2,000-kilometer test journey from Chongqing in the southwest to the Chinese capital in the north.

In early April, the Swedish carmaker Volvo, which is owned by Zhejiang Geely Holdings - headquartered in Hangzhou, in the province of Zhejiang in East China - announced a plan to launch a self-driving experiment involving as many as 100 cars.

Chinese tech giants, such as Baidu and LeAuto - a subsidiary of Leshi Internet Information & Technology - are also engaged in self-driving technologies. Baidu started a research program in July 2014, and undertook its first fully automatic road test in December, while LeAuto unveiled the LeSEE, its concept driverless vehicle, on April 20.

"Worldwide, self-driving technologies are becoming mature. I'm optimistic that it will only be five years or so before we will be able to buy self-driving cars," Yu said.

The timeline accords with the plans of the major players: Google, Toyota, General Motors, and Nissan have all disclosed plans to put self-driving cars in the market by 2020.

Cost challenges

However, there are still challenges, and right now the biggest obstacle to full commercialization is the high cost of the vehicles.

Based on advanced hardware - sonar devices, stereo cameras, lasers and radar, together with complex self-driving software - the cost of an autonomous car will be $7,000 to $10,000 higher than that of a traditional vehicle by 2025, according to a report by IHS, an auto industry consultancy, which said technological developments will see the price premium drop to around $5,000 in 2030 and $3,000 in 2035.

LiDAR, a laser radar system that Google installs on the roofs of vehicles to provide a 3-D picture of the surroundings, costs between $75,000 and $85,000, which means it is too expensive to be fitted in commercial vehicles.

Even after the supplier, Velodyne Acoustics, introduced a cut-price, minimal system at a cost of $30,000 to $40,000 in 2014, only automakers and suppliers could afford to buy the technology for testing purposes.

That offers an opportunity to Chinese entrepreneurs, whose forte is squeezing costs.

In December, Baidu tested two prototypes of autonomous cars - BMW 3 Series Gran Turismo models - priced at 400,000 yuan to 700,000 yuan ($61,520 to $107,660). Both cars were fitted with LiDAR systems that cost 700,000 yuan each.

"Such a high price will only be feasible at the testing stage, it will definitely need to be reduced substantially when being commercialized in the future," said Wang Jin, who runs Baidu's autonomous driving division, quoted in Shanghai's The Paper in December.

By using the word "substantially", Wang was making it clear that he expects the price of lower-resolution guidance models to fall by about 90 percent, not the 50 percent sugg-ested by suppliers such as Velodyne Acoustics.

In March, Wang told the media that the company is cooperating with suppliers and hopes to reduce the cost of the radar system to about 20,000 yuan - 2.85 percent of the cost of the LiDAR system used in December - in three to five years, by which time Baidu will begin mass production of driverless cars.

So far, Baidu has managed to reduce the cost to less than 200,000 yuan (less than one-third of the original cost), he added.

M&A drives development

According to Yu, from ResearchInChina, although Chinese companies have no advantages in terms of the core technologies, some of the larger companies have one major edge - they have plenty of money to invest in development.

"From a technological point of view, in general Chinese companies are lagging five years behind their international rivals, most of whom are still buying radars, cameras and control systems from suppliers in the US, Europe or Japan, which means it would be difficult for them to lower the cost to a level that's practical for commercialization.

"But with abundant funding, we have seen active cross-boundary mergers by Chinese companies in an attempt to atone for the technological shortfalls," he said.

In May last year, Zhejiang Asia-Pacific Mechanical & Electronic Co, a supplier of braking systems, initiated a plan to raise 1.2 billion yuan to finance the development of self-driving technologies. In April, after a series of mergers and acquisitions, the company announced breakthroughs in a number of core technologies, together with a new plan for 2.3 billion yuan of additional investment.

It also set itself the goal of successfully developing automatic driving technology by 2023.

In February, Ningbo Joyson Electronic Corp, a major auto-electronic products manufacturer in Zhejiang province, spent $1.1 billion to acquire two leading suppliers of automotive components - Key Safety Systems in the United States, and Germany's TechniSat Automotive.

On April 22, Baidu announced that it had established a self-driving car team in Silicon Valley, California. The team includes experts in machine learning and hardware, along with software engineers across a variety of technical domains, from robotics and computer vision to onboard computers and sensors. The company plans to expand the team to more than 100 researchers and engineers by the end of this year.

Potential profits

The driving force behind the moves to develop driverless cars, irrespective of cost, is the potential profit to be made from China's massive auto market.

A survey by Boston Consulting Group in April last year showed that China will soon corner more than 25 percent of the driverless car market, which is expected to hit 12 million units globally by 2035.

Meanwhile, a recent report by Roland Berger, a global strategy consultancy, showed that Chinese car owners are much more open to unmanned cars than Europeans and people in the US.

As many as 96 percent of Chinese drivers stating that they would consider an autonomous vehicle for almost all everyday driving, whereas in the US and Germany, the figure was 58 percent.

"Chinese people are probably the world's most adventurous consumers - they are always ready to try new things. Just look at how smartphones swept the country overnight and you can predict what will happen to the market for autonomous cars," said Wang Yong, founder of VVpinche, a cab-hailing company in Beijing based on the model popularized by Uber and Didi Chuxing.

For Wang, who is also a driver, sitting in an autonomous car is no more stressful than driving on a road full of new drivers, because "driverless cars are definitely safer than human drivers".

Cui Rui, co-founder of Jiewo, an app that allows people to share private buses, sees plenty of opportunities in self-driving buses. "In the initial stages, it is more likely that self-driving technologies will be used on vehicles that travel along fixed routes, which will reduce management challenges and offer improved safety," he said.

Wang and Cui both believe that the commercial vehicle sector will be the first testing ground for driverless cars because cab companies are more sensitive to costs - especially labor costs - than individual car buyers.

"Self-driving represents the future, although its commercialization could be difficult at the beginning. If there were some major accidents in the first few years, it would slow down the development of the whole industry," Wang said.

Legal liabilities

What concerns Wang is the universal problem with replacing human drivers with robots: a driverless car can avoid hitting other objects, but it cannot stop a manually driven car from crashing into it. If an accident occurs - and it happened a lot during Google's road tests, according to the company's own accounts - it will be difficult to determine who bears responsibility.

The debate about whether manufacturers should shoulder the liability raged for several years, hindering the commercialization of driverless cars. The problem was solved last year, when Volvo, Google and Mercedes-Benz announced that they will accept full liability if their autonomous vehicles cause a collision.

Although the announcement cleared the air with regard to liability, an overall lack of rules and regulations for autonomous cars is frustrating industry insiders.

"The US risks losing its leading position due to the lack of federal guidelines for the testing and certification of autonomous vehicles," said Hakan Samuelsson, the CEO of Volvo in an October report in Digitaltrends, a technology news and information website.

"Europe has suffered to some extent by having a patchwork of rules and regulations. It would be a shame if the US took a similar path," he said.

After years of lobbying by automakers, four US states - California, Nevada, Florida and Michigan - have decided allow autonomous vehicles to be tested on public roads.

Despite being a latecomer in the field, Chinese decision makers seem to be much more active in giving green lights.

In November, Shanghai opened a 3.6 km section of road for testing self-driving vehicles - the first such zone in China. A short time later, the Ministry of Industry and Information Technology approved similar test areas in the municipalities of Beijing and Chongqing and in the provinces of Hebei and Zhejiang.

At the 2016 Beijing Autonomous Driving Panel Debate in April, a senior official at the Ministry of Transport said the country is conducting preliminary studies into an intelligent transportation-control network and a pilot program will be undertaken in the near future.

"The government is positive, but cautious, about the development of the technologies. The country is establishing a specialized testing environment," said Wang Xiaojing, chief engineer at the ministry's Institute of Highway Research.

According to Yu, from ResearchInChina, official approval and encouragement could make all the difference for homegrown auto companies: "In China, the policy problem could be solved fairly easily if the government decided to develop the industry. That could give Chinese companies a real advantage."

Contact the writer at chengyingqi@chinadaily.com.cn