Pony.ai Q&A: Having a China background will be key to autonomous driving success

2050 aims to equip young people to take action and to become volunteers. Ahead of the event in May, we are taking a look at some the companies and people who are taking part in the massive unconference. 2050 is a volunteer-only, not-for-profit unconference. TechNode is organizing the Explore Expo, an exhibition area for young tech startups looking for exposure.

In January 2018, Pony.ai got $112 million in Series A funding only 1 year after its inception. This is enough funding to make it instantly a serious contender in the sector anywhere in the world. The two founders James Peng and Lou Tiancheng have backgrounds at both Baidu and Google’s autonomous driving projects.

Pony.ai (小马智行) is developing cars capable of Level 4 autonomous driving. That means fully self-driving, with no human input. Fully autonomous testing started on the Nansha island in Guangzhou this February, the first time on public roads in China. The company also signed a deal with Guangzhou Automotive Group, the country’s number two car maker.

Pony.ai is headquartered in California, but when we caught up with co-founder (and champion coder) Lou Tiancheng in Beijing, he explained how and why they’re actually a China-first company, the problems of guessing intention, and why he’s taking part in Yunqi 2050.

Do you consider yourselves a Chinese or an American company?

Neither, actually. We’re an international company, focused on the Chinese market. Our headquarters are in Fremont [California]…  But sometimes “headquarters” is just a word. So if you say, where was the company founded, I’d say the Cayman Islands. But I’d never say we’re a Cayman Islands company.

What makes you different to other companies involved in autonomous driving?

First of all, we’re China first. A large number of our employees are from China or have a Chinese background. So compared with other autonomous driving companies in the US, we have a much better understanding of the Chinese market, including the driving scenarios, environment and government. Another thing is that we’re Level 4 first—we focus on fully autonomous driving. We’re not trying to build cars with assisted driving, but a fully autonomous driving car. We’re also trying to build an overall system, not just the parts for autonomous driving.

What’s so different about China?

To design the product, we must first understand the requirements. Different requirements need completely different technologies. In China, some of the requirements can be very different to in the US.

So why did you set up in the US?

Talent, for sure. We’re China first, but not China only. We started our business in the US, but won’t put all or even the majority of our resources into the American market.

Much of the autonomous driving news pits US companies against Chinese. Is that how you see future competition?

My guess is that companies with a strong Chinese background–you don’t need to be a purely Chinese company–but should have a strong Chinese background [to be successful].

Is the data from China testing different to that from the US?

Oh, yes! That’s also one potential reason why companies that will be successful in China will have a strong Chinese background. The data, the patterns can be very different. So in Nansha [in Guangzhou] we have people running the traffic lights, but this seldom happens in the US. One of the tricky parts about autonomous driving is trying to understand the intention of other people. To understand the intention of someone else is much harder than image recognition. It’s even hard for human beings. The intention pattern in different countries can be completely different.

When will we see Level 4 driving go mainstream?

Let me put it this way, it’ll go from small areas to global. I will say that in one or two years we’ll be able to see Level 4 cars serving as a robotaxi [the name given to the autonomous test cars in Guangzhou], the taxi without driver so no Uber or Didi staff. We should see that happening in small areas, but for the whole country or majority of driving areas, that’ll take five to ten years. Like mobile phones in the early 1990s—restricted areas only.

From March, April this year Chinese cities have started policies that allow testing on the roads and I’d say that within a year most cities will have their own policies for autonomous driving. [We are sticking to our testing to Beijing and Guangzhou because] as a startup we have to focus, it’s too early to expand our business to other cities that quickly.

Why do you want to take part in Yunqi 2050?

It’s the perfect platform for me and for young people. It’s a great stage for young people to exchange ideas, learn, develop themselves so that in the future they can make an even bigger contribution to the country.

What advice would you give to a young person starting out?

Let’s take a step back. I’ve been in this industry for ten years and there’s always been ups and downs. We have to have a very good understanding of the ultimate goals of what we’re doing, so that in the long term we can encourage other people–and ourselves–when we’re in a downside. In recent years I’ve seen a lot of people get very disappointed and burn out in the donwnsides, and give up. So for today, with Yunqi we can set up the goals for the next 30 years and prove to ourselves that we’re doing something amazing and having a huge impact in the long term. That should help people keep motivated, and autonomous driving is just one area.

If China has a talent shortage, what can it do about it?

First I’d say that talent is international and we should accept and welcome talent from around the world and try to eliminate boundaries to this talent. We should have policies that support talent and help it deliver more impact to the industry or country, which could help it attract more people.

Five, ten years ago when I was an undergraduate I would complain that the tech courses I was taking at university were not the trending ones but that can be changed. Today I’m seeing lots of trending topics being taught and talks and even this 2050 meeting. There are many ways to help our undergraduates learn about trending technologies.