China needs new concepts for AI cars
Governments can stimulate markets in many ways. China has for some time concentrated on developing its transport infrastructure – with fast inter-urban trains, highways and regional airports, most cities are only a few hours apart. This permits human and material resources to flow more easily to nurture the nation’s economic growth prospects.
But rapid economic growth has come at the cost of increased atmospheric pollution, so the government joined with all other nations to combat climate change and has exceeded its promises. However, city pollution remains too high, so the government has promoted the manufacturing of batterypowered road transport.
The end-goal of this policy has massive ramifications and touches every person’s life. It begins with the electrification of road vehicles (trucks, buses and cars), together with the search for better computer systems guiding autonomous vehicle movements. Ultimately we will gain mobility as a service.
However, China’s market stimulation has resulted in almost 500 electric vehicle (EV) manufacturers being responsible for about 43 percent of global sales of battery augmented vehicles in 2016.
We must distinguish between PHEV models (plug-in hybrid EV) and BEV (pure battery-powered EV). China concentrates on BEVs while the rest of the globe sees growth in PHEVs.
Two EV licenses are usually required in China – one to build the car, the second to sell the car. However, some Chinese EV makers have developed their vehicles without licenses by simply making sample models. Others circumvent licensing rules by forming joint ventures with existing companies.
In an article published in 2017, Zinnov analyst noted that Asia had seen automotive research and development (R&D) increase 75 percent since 2008, faster than North America, Europe or Japan. The bulk of this R&D spending was in China, hitting $12 billion in 2015.
Most major vehicle manufacturers adopted EVs initially to reduce their total fleet emissions as demanded by regulations. Now, they are turning to merger and acquisition deals or joint ventures to increase their skills range and their supply-chain capabilities to construct a new EV more quickly.
Start-ups have arrived, perhaps stimulated by US-based Tesla, even though that company has been slow to increase its market share. China’s Nio states it will ramp up production in only three years to match Tesla’s production by 2020, and it has even held an IPO in the US.
Nio will vertically integrate its digital systems like Tesla. This is very important as the whole car’s performance is determined by the integrated performance of all its electronic components. A fully integrated digital system allows over-the air software upgrades (like a smartphone or PC), augmenting performance continuously.
Consolidation in the EV sector in China will occur. The government began this process in 2016 by closing underperforming producers and demanding that companies seek licenses.
China first issued licenses to testdrive autonomous vehicles on a public road in March 2018 and these only permitted driving along 5.6 kilometers of road in Shanghai by a few car manufacturers. In contrast Waymo (formerly a car project of Google) has logged more than 8 million miles (nearly 13 million kilomters) of road testing and thus information gathering for its AI computational database. These are tests in the US, so Waymo is considering an expansion of its testing into Europe as driving and pedestrian movements are different across national and cultural boundaries.
Unless Chinese R&D achieves a remarkable breakthrough, it would appear its reach for autonomous vehicles lags behind other nations.
At Level Five (the most automated), the vehicle needs many sensors to “see” the road ahead, all the other vehicles, and pedestrians and the road architecture to determine the best route to the desired destination without any driver intervention.
This is a heavy computational task suited to the abilities of complex AI and machine learning, sectors in which China aims to be a world leader. The complex integrated technology will raise costs – and to defray the manufacturer’s investment, the price of the vehicles with all their inbuilt computer systems will have to be raised.
Mass-market customers may reject a high-cost vehicle, so we need to look for a new business model. Many analysts have suggested a move to 24/7 on-call taxi systems.
Inevitably, we will move to redesign our towns to accommodate humans, rather than vehicles. Such strong social engineering is well within the scope of the Chinese government – recall its massive move to raise living standards for millions of its poor people within the Millennium Goals. Changing car ownership to public mobility as a service model ought to be relatively easy.