How Didi is plotting to commercialise driverless cars

Ride-hailing firms are ideally placed in the race to put autonomous vehicles on the road, argues the Chinese unicorn’s CTO. Didi is making a big push to be first.

China’s homegrown ride-hailing unicorn Didi Chuxing is trying to solve one of the thorniest problems in motoring: how to profit from self-driving cars.

Car makers globally have invested heavily in creating autonomous vehicles but consumers are unlikely to buy them until the technology is proved safe in variable weather or road conditions, especially after an Uber self-driving SUV killed Elaine Herzberg on March 18 in Arizona.

“It doesn’t make sense for a consumer to buy a car to drive it only on a sunny day or during good road circulation,” said Bob Zhang, Didi’s chief technology officer, speaking at the RISE conference in Hong Kong.

However, Didi plans to gradually deploy fleets of self-driving cars on the road when its artificial intelligence systems tell it is safe to pick up a passenger in an autonomous vehicle. If not it will dispatch a human driver.

A ride-sharing network is able to collect vast amounts of data from its fleet of cars on the road 24/7 and use it to closely monitor road conditions and driver behaviour. Didi handles 30 million trips per day for 550 million users. This traffic generates over 100 terra bytes of vehicle trajectory data a day. Didi processes more than 4,800 terra bytes of data a day.

Crunching that data, Didi can already predict future demand for cars with 85% accuracy within the next 15 minutes. This allows Didi to divert vehicles to areas of future high demand.  

Another advantage Didi has in assessing likely driving conditions is that it already knows the route the customer requested. Didi handles over 40 billion route planning requests a day.  

“We can incrementally expand the area” of self-driving cars when conditions are right, Zhang said, believing it the only way to commercialise the technology on a large scale -- at least as far as the next 10 years is concerned.

The race to run a fleet of driverless cars is fuelled by the desire to cut costs and fear of a knockout competitive blow. For ride-hailing firms like Didi, employing drivers is around 90% of the cost, dwarfing fuel costs and the price of leasing cars. 

If ride-hailing companies can slash the overall costs, they will also be able to compete with the private car more effectively: bad news for car makers.

“Fewer people will choose to own a car [and] instead will just share one,” Zhang said.  

Didi’s peers are not standing idle either. Singapore-headquartered Grab is in the process of raising fresh capital and is working with Toyota to develop self-driving cars. Even car makers are pushing into ride sharing -- both Renault and Volkswagen have announced plans to roll out electric vehicle ride-hailing fleets in Paris and Germany, respectively.

However, San Francisco-headquartered Uber seems to be in reverse gear since the accident in Arizona, laying off hundreds of operators of autonomous vehicles.

LARGE INVESTMENT

 

Such breakthroughs in technology do not come cheaply.

The six-year-old company has already completed 15 funding rounds, raising a total of $20 billion from investors such as Tencent, Temasek and Foxconn, according to data provider Crunchbase. Didi announced a $4 billion fundraising from Japan’s Softbank and the Abu Dhabi government in late December, a deal that valued the high-flying startup at over $50 billion.

Chinese autonomous driving startup Pony.ai closed the $102 million second round of its series A fund raising on January 11, co-led by Fidelity's proprietary investment arm Eight Roads Ventures and private equity firm ClearVue Partners.

Pony.ai is working with Guangzhou-headquartered GAC Group to develop autonomous vehicles and together they are piloting ride-hailing cars.  

According to Chinese auto industry information platform Gasgoo, 24 autonomous driving technology startups have been established since August 2011, most of them in the last four years.

Eight Roads Ventures says that it screens prospective investments by selecting companies that have robust technology, an ability to commercialise their product and a strong management team. Not all the companies trying to raise funds have these attributes.

““There is an enormous amount of momentum money chasing the AI concept which creates a valuation bubble,” Jarlon Tsang, managing partner at Eight Roads Ventures, told FinanceAsia.

And that may not be good news for entrepreneurs either. As Tsang put it: ““Some fairly solid companies might have troubles raising money later on because of the unreasonably high early-stage valuations leading to unintended down-rounds or lack of investor interest.”

¬ Haymarket Media Limited. All rights reserved.