My job as an auto journalist offers me a tantalizing glimpse into highly automated cars, which navigates the unknown future of our relationship with vehicles with artificial intelligence.

Experiencing all the progress personally was an enlightening and intense experience for a newbie driver like me. Having barely developed the confidence of basic cruise control that sets my Mazda Axela at a certain pace — useful only on open roads free of traffic — I was put behind the wheels of a Tesla for a hands-and-foot-free test drive through Shanghai’s urban jungle.

Following a software update that wakes up sensors and cameras embedded in Tesla cars, the brand now features the most aggressive use of semi-autonomous technologies in the market. Enhanced autopilot and industry-first autosteer can be activated manually within a certain speed range, enough to support most of the driving scenarios on highways and city streets.

I was instructed to sit back and let the car follow the lane, maintain a safe distance with the car ahead, follow the traffic flow, obey roadside speed limit signs and even change lanes on my turn signal— all by itself.

Safe as it may sound, I was sweating in my hands, my foot hovering over the brake, my eyes locked on the road ahead.

“It’s normal to feel nervous the first time,” my Tesla driving coach said as he tried to put me at ease. “It also took me and my colleagues to learn and adapt. Now we feel comfortable enough to check our social media accounts as the car drives by itself.”

What he portrayed is a blithe sail that autonomous driving promises to replace our humdrum commute so as to justify its full commercialization. Deep down, I suspected it will still be a painful push of boundaries for multi-tasking drivers now that they are given an additional distracting task to dwell on the haunting questions on whether human-programmed machines can surpass man’s limits.

Highly automated cars are fully loaded with sensors, cameras, and data processing and integration units, making them stay focused and alert forever. But even with sharp eyes, ears, and brain, their ability to make sense of the world, and react accordingly has to be developed gradually through experience — our experience. Like they say, we simply mature with age!

Though I was impressed by Tesla’s performance, I couldn’t resist my instincts to take back the controls when I sensed it was a bit at a loss at a crossroad. I remember reading users’ posts saying that the vehicle tends to take every right exit on the highway just because there were no clear lane markings within the sight of their cameras.

As engineers keep searching for bugs to be fixed in programming, carmakers and lawmakers keep telling the world that the highly automated functions at this stage can only assist drivers — and not replace them or take responsibilities for them.

As an auto journalist, I would like to applaud every single little progress made in artificial intelligence, but as a consumer I cannot help wondering how much their presumably questionable judgment can be worth to car buyers.

Knowing their awkward place in the market, they are either testing the water as fancy add-ons in luxury vehicles like Tesla and also BMW 7 series, or trying to fit into the big picture of intelligent transportation drawn by ambitious officials.

Smart connected cars

Only days after the Tesla test drive, I went to the Shanghai International Automobile City to observe a government-led experiment about smart connected cars. They felt like ages apart.

Riding shotgun in a fully wired Roewe E50 in a simulated real-world urban environment at the Shanghai Auto Expo Park, I did not get what’s so special about its spot detection, front collusion and speed warning functions at first as I had seen them countless times elsewhere. But at the end of the demonstration, as a green wave belt appeared with traffic lights and vehicles coordinated to form a smooth traffic flow without unnecessary acceleration or braking, I sat up and took notice.

“Smart connected cars are based on a network of vehicle-to-vehicle and vehicle-to-infrastructure communications, enabling them to adapt to predictable road conditions,” explained my driver, an auto engineering student from Tongji University, which is a key partner in Shanghai’s smart connected vehicle program. “That shortens the time a car takes to react by 0.5 second compared to engagement triggered by passive detection, which is the point of us doing all those existing functions all over again.”

With such a fleeting moment’s difference, which is to be amplified by the technology’s wide application, this active approach is expected to take the idea of autonomous driving a further step forward, from an isolated self-running machine to a self-governing transportation network beyond the coordination ability of our own.

“It is very visionary. It would be such a shame if I did not live long enough to see it actually happen,” the student, in his mid 20s, said.

A systematic shake-up like this comes at an extremely high price, monetarily and also legally. Unless the government takes strong initiatives to push it through, the vision will unfold only in academic labs, a closed environment free of regulatory disputes or public security concerns.

I left the test field giving the demonstrator my best wishes to his research. “The world mocks idealists as much as it does to pragmatists. Just be the change you want to see.”

Cost concerns

In my interviews with companies, I could immediately sense a much stronger desire to bring out the real-world value of autonomous driving. Uncertainties over reliability aside, suppliers of technologies are doing their best to reconcile their vision with the reality of cost concerns. Even the most open-minded customers have to be attracted by price to explore the unknown world of robotics.

The recent awarding ceremony for the second innovation challenge contest run by Valeo, a leading developer of autonomous driving technologies, was equally an eye-opener for me.

Standing out from more than 1,300 teams from world-renowned academies, the winning team Falcon View from China’s Peking University replaced expensive radars with a cheap wheel-based system, using just basic cameras to help cars take in surroundings. It is called visual-based on-road vehicle detection through online road structure understanding, an innovative combination of analytical perception and human-like deduction drawing the power from algorithm.

“Usually to help cars better understand the environment, we add more sensors. And by doing that, we add more costs. It is our job to provide technologies that are affordable to customers,” said Edouard de Pirey, president of Valeo China, commenting on this out-of-the-box idea that attracts the company’s interest.

From parking sensor, a luxury item to have before 2000 and now a must-have, to park4U, a semi-automated parking pilot that has sold 7 million units since 2007, Valeo is known for its proactive industrialization of drivers’ assistance technologies to scale up the market.

Now to the end of economies of scale, it also sets out to leverage the power of computing to unlock the potential of existing hardware. Its latest 360 degree birds eye view of the vehicle’s immediate surroundings, which is to help drivers see through blind spots, is achieved through upgrades of softwares rather than cameras.

Before I could imagine an auto industry dominated by programmers to make cars ever smarter in this age of the Internet, a speech for the award ceremony made by Zha Hongbin, professor of machine intelligence from Peking University, dragged me back into reality.

“Robotics is a cross-disciplinary study that also involves psychology and sociology because you expect your self-driving car to understand your intention, take off your pressure of driving, bring your comfort, and even apply game theories and seek cooperation under China’s challenging driving conditions.

“It is a study about how we can truly become one,” he added.