Having made it easier for passengers to hitch a ride, China’s biggest car hailing company has its sights set on the next challenge: easing road congestion so those same riders get to their destinations faster. One year after forming a smart traffic unit, Beijing based Didi Chuxing is in talks to export some of this knowhow to Brazil and Australia to help ease traffic jams. “The Brazil government wants us to help them optimize their traffic lights. Traffic jams there are even worse than in Beijing,” said Bob Zhang Bo, Didi’s co-founder and chief technology officer. Authorities in the city of Melbourne asked the same thing after Didi launched services in that city in June, but Zhang admits that Didi won’t be in a position to implement a full operational system until it builds up its fleet of cars in those locations. However, consulting with city authorities on traffic management is happening in mainland China where Didi has many more cars on the road – it doesn’t disclose how many – but last year there were 21 million paid Didi drivers. The company is a partner in a project that uses artificial intelligence to manage over 1,300 traffic lights in 20 mainland Chinese cities including Beijing, Jinan, Wuhan and Guiyang. The latest version of the system, launched in the second quarter, can optimize traffic conditions in a neighborhood instead of just a single intersection, Zhang said. VW in talks to manage Didi fleet, co-develop self-driving cars Data has become critical in the AI era, giving companies like Didi and e-commerce giant Alibaba Group – which operate in the world’s most populous country – a competitive advantage over rivals in other countries. Malaysia will be the first country outside China to adapt the smart city system developed by the cloud computing arm of Alibaba, which owns the South China Morning Post , while a similar project is being done in partnership with Macau authorities. Sidewalk Labs, a sister company of Google, last October formed a partnership with Canadian authorities to develop a US$50 million design to transform a derelict waterfront in Toronto into a sensor-enabled, highly wired smart city. These project represents a growing worldwide trend in which governments increasingly work with technology companies to deploy advanced technologies such as AI in transport, energy, governance, security and safety. “Over the past two decades, internet technologies and the mobile era have tackled the moving of information, while Didi hopes to solve the problems in the moving of physical things,” said Zhang. The company said it has more than 5,000 engineers and scientists among its total workforce of over 9,000 people. It has launched ride hailing services in Australia, Taiwan, Hong Kong and Mexico as part of a global expansion, and in January announced it would buy Brazilian ride-hailing company 99 Taxis. The company’s goal is to serve more than 2 billion customers, Didi’s chief executive Cheng Wei said in April. Zhang worked at Baidu before being part of the founding team at Didi, which merged with rival Kuaidi Dache in 2015. The new company then engaged in a brutal price war with Uber Technologies in China, which was settled a year later when Didi took over the US company’s China operations. 1 million vehicles in just 3 years: meet the world’s greatest used car salesman For Zhang, a computer science major, it all comes down to applying mathematical models to the problems of traffic management. Using the data from its more than 7 billion rides last year, it can apply AI analysis to help traffic lights automatically adjust their signal timing based on the real-time movement of cars on its platform. “Although the traffic lights have remote controls at the command centres, adjusting the signal timing has to be done manually with [human] operators using huge screens to monitor countless intersection,” he said. “Now the [AI-driven] system can adjust signal timing [at several intersections] within one second, and immediately analyze the result.” Another project Didi has been involved with is reversible lanes, where underused lanes on the opposite side are switched over to alleviate congestion. Data on the flow of vehicles and their speed is crunched to help determine when lanes can be reversed. Didi has been working with authorities in Jinan, Shenzhen and Wuhan on optimising traffic flow using reversible lanes. “It is like the tide flowing towards the office and receding in late afternoon,” said Zhang. Didi’s confidence in tackling an area that seems outside its core competency of ride hailing comes from the understanding it gained from analysing rider and driver behavior. “Didi has an immense amount of valuable data on how drivers react to various scenarios,” he said. While the data amassed by tech companies like Didi and Alibaba gives them an edge in developing smart traffic solutions, there are security concerns regarding privacy, including the use of video from cameras installed at lights, according to Gao Jian, deputy head of China Urban Construction Design and Research Institute. “It is a good thing to see their participation and technology support in urban planning, but the [traffic] authorities will still have to consider the different functions of special areas [like school zones]," he said. “It will be difficult to exchange data that would otherwise help optimize the solutions.” Li Yixin, vice-president of Freetech, a Hangzhou-based intelligent driving solutions provider, pointed out that smart traffic solutions are complex systems, involving not only cars, but pedestrians, bicycles and other non-motorized vehicles, requiring trust between conventional participants and the tech providers. “Building a smart city requires teamwork. Technology is a means to an end, so tech companies should work with urban planners and city administrators to complement their needs,” Li said. Didi’s data on passenger preferences is also being used in a different way. In April, the company formed an alliance with 31 car industry partners, including VW, BYD and electric car start-up CHJ, to develop vehicles tailored for ride-sharing. “Didi is likely to become the biggest consumer [of cars] in the future,” Zhang said, adding that it is also looking at investing more in auto distributors and car rental companies. Last year Didi injected US$200 million into Renrenche, an online platform for used car sales. Separately to traffic management, Didi has pledged to boost its fleet with electric vehicles that are supported by charging stations connected to a cloud-based “brain” that uses a math-based model to optimize use of the facilities. “With knowledge such as where the driver is heading or where the next ride order is likely to happen, we know the most suitable spots to install [them],” he said. “Drivers won’t need to worry about running out of batteries as we will optimize by planning ahead and assigning orders accordingly, as to when and where to charge.” Didi plans to have 1 million electric cars in service by 2020, according to CEO Cheng, who revealed the target to local media in December. There is economic sense behind the company’s push into EVs. The break even point for an electric car over a petrol-engine one is 200,000km on China’s roads, according to Zhang. While it would take an average private car owner 10 years to accumulate that mileage, Didi drivers do about 96,000 kilometers a year so can achieve ROI in only two years. “Didi drivers have more incentive to swap their gas-fueled cars with EVs,” Zhang said. Besides electric vehicles, Zhang said Didi is “all in” on autonomous driving. In the near term, he believes the advantage is with car hailing companies because it is unlikely the technology will be able to handle all geographical conditions under all types of weathers within the next decade. That means the business model for self-driving cars in the foreseeable future is a good match for mobility service providers instead of individual buyers. “As we know when and where cars are needed the moment riders place their orders, and also know the routes they are likely to use and real-time weather, Didi can send a robotaxi to pick up the passenger if we are confident about the trip, or send a human driver if we are not,” he said. “As there will be a combination of human drivers and self-driving cars, we will gradually expand the use of the latter.” However, Alphabet’s Waymo unit is seen maintaining its global leadership in autonomous driving, a market that could be worth up to US$2.8 trillion by 2030, according to a report from investment bank UBS. Baidu’s self-driving buses roll off production lines as AI push continues Separately, China’s Baidu said on Wednesday that the first 100 of its self driving buses, developed in partnership with King Long United Automotive Industry Co, have rolled off the production lines, and will be used in cities including including Beijing, Xiongan, Shenzhen and Tokyo. Didi has a fleet of over 40 autonomous cars being “relentlessly” tested in three cities in China and in the US. Zhang said the company is aiming to make it into the second-tier after Waymo by year end in terms of frequency of human driver intervention. Earlier this year, California's Department of Motor Vehicles reported that Waymo's self-driving cars had disengaged from autonomous control every 5,596 miles, compared with every 41 miles for Baidu. “I believe over the next decade, transportation will undergo an immense transformation, bigger than what we had in the past 50 years combined,” Zhang said. “Cars will evolve into a completely different species.”
