China recently released some basic guidelines for self-driving car development. As an American, I don’t always fully comprehend the line in China between private companies and the government. How much guidance to self-driving car developers in China get from published laws and regulations, and how much comes from internal communication with the relevant government agencies?

Here are the public guidelines:

“The rules lay out requirements that vehicles must first be tested in non-public zones, that road tests can only be on designated streets and that a qualified person must always sit in the driver’s position, ready to take over control.”

As best I can tell from English translations, that is the extent of the rules. Presumably there must be more, but I don’t know if the rest is available in Chinese, or if you have to be in the industry there and know the right people to figure it out.

“Even when race car drivers leave the mayhem of the track, their skill doesn’t outweigh their appetite for risk: a study from the 1970s found that racing drivers from the Sports Car Club of America had a higher crash rate on public roads than other drivers from the same state of the same age and sex.”

Compare that to this:

“At the opposite end of the spectrum are those who use cautious driving styles to make up for their weak skills. Some elderly drivers who score poorly on a driving test nevertheless manage to drive crash-free by actively compensating for their deteriorating abilities, according to a Belgian study from 2000. They drive more slowly and avoid tailgating, leaving long safety gaps behind vehicles they’re following; they also plan their trips to avoid complex traffic or other challenging situations.”

Of course, this has implications for self-driving technology:

“Some of the more dramatic estimates have imagined quintupling the volume of traffic flowing down a road. But a short “headway” — the gap between one vehicle and the one just ahead — brings a higher crash risk than a long headway.”

From a thought-provoking essay by Antonio Loro.

Supposedly the reason the Google Self-Driving Car Project kept its vehicles tooling around Mountain View at 25mph for years is that a collision at 25mph results in something like a 20% chance of a human fatality. At 45mph, the likelihood of fatality flips, and becomes something like 80%. A 2015 report hints at this, although it is more vague about the statistics.

Chinese e-commerce giant Alibaba is building “self-driving technology”, according to the MIT Technology Review:

Alibaba says it has bigger ambitions than just robotic taxis. In June 2016, the company launched an AI-powered “city brain” system in Hangzhou, where it’s headquartered, to crunch data from mapping apps and increase traffic efficiency. Simon Hu, the president of Alibaba Cloud, says the firm’s ultimate goal is to produce the kind of autonomous driving that uses such data to help integrate transportation into urban infrastructure.

The common analogy is that Baidu is the Google of China, Tencent is the Facebook of China, Didi is the Uber of China, and Alibaba is the Amazon of China.

This announcement puts Baidu, Didi, and Alibaba all in the self-driving car race.

A police officer pulled over a Cruise self-driving car in San Francisco recently, and issued a citation for failing to yield to a pedestrian in a crosswalk. Cruise disputes the citation, but since nobody was hurt, the interesting thing to me is that the safety operator got the ticket.

Knowing nothing about the actual interaction, I imagine the police officer writing the citation, caring not a whit about who or what was actually operating the vehicle, and walking off.

Longer-term, how does this work? Specifically, what happens when safety drivers are no longer in the vehicle.

Maybe officers will ticket passengers. That will get the vehicle owners’ attention quickly.

What if nobody is in the vehicle?

Routine traffic stops are one of the most common interactions between citizens and police in the United States. For many people, this is their only interaction with the police. As self-driving cars progress, we’re going to need to rethink interactions between police and the public.

On top of its existing partnership with Fiat-Chrysler, and right on the heels of its recent announcement with Jaguar Land Rover, Waymo is supposedly on the cusp of announcing a partnership with Honda to develop driverless delivery vehicles.

This diversity of partnerships presumably helps Waymo on both offense and defense. Deploying across multiple partners on different continents will move Waymo further into a position as the industry-standard for autonomous vehicle software. And maintaining alternative vehicle suppliers protects Waymo against any one supplier leaving the fold.

On the other hand, managing multiple supplier relationships is a huge time investment. There are parallels between Waymo’s new position in the automotive supply chain and the current positions of the automotive manufacturers, who spend a lot of time managing their own supply chains.

NVIDIA (ahem, one of Udacity’s partners in developing the Udacity Self-Driving Car Engineer Nanodegree Program) announced their new DRIVE Constellation Autonomous Vehicle Simulator a few weeks ago at their annual GPU Technology Conference.

NVIDIA is taking the next step toward the holy grail of autonomous vehicle simulation: training autonomous driving software in the simulator and deploying it straight to the real world.

We’re not there yet. NVIDIA executives subtly make the point that DRIVE Sim is for “testing”:

“Deploying production self-driving cars requires a solution for testing and validating on billions of driving miles to achieve the safety and reliability needed for customers,” said Rob Csongor, vice president and general manager of Automotive at NVIDIA. “With DRIVE Constellation, we’ve accomplished that by combining our expertise in visual computing and datacenters. With virtual simulation, we can increase the robustness of our algorithms by testing on billions of miles of custom scenarios and rare corner cases, all in a fraction of the time and cost it would take to do so on physical roads.”

Testing and validation is hugely important to the autonomous vehicle industry. Every time a software developer tweaks a model or a the autonomous vehicle source code, the developer needs to verify that nothing else broke. The fastest way to do this is testing in simulation. NVIDIA is releasing what looks like the world’s best simulation platform for testing.

The biggest leap, though, is training in simulation. It’s one thing to write code, put it in a car, collect data about different scenarios, and then later test in simulation that the your self-driving software still handles those scenarios correctly.

It’s a whole different animal to be able to train self-driving software in simulation, without ever putting a car out on the road. Whoever cracks that nut will have a huge leg up in the autonomous vehicle race. Keep an eye on NVIDIA.

Back home on vacation this weekend, my brother was telling me about paying up to $30 one-way to drive Interstate 66 inside the DC Beltway. One the one hand, the pricing (which varies with traffic) is outrageous. On the other hand, people will pay it. My brother, coming off a 12- or 16- or 20-hour hospital shift as a medical resident, will pay whatever he has to pay to get home and sleep.

The latest round of self-driving car projections offer conflicting predictions about the effect self-driving cars will have on congestion. Self-driving cars should lower the psychological costs of vehicular transportation, which should in turn encourage demand and ultimately lead to more congestion.

On the other hand, self-driving cars may be more efficient drivers, and will encourage ride-sharing. Ride-sharing, in turn, has the potential to reduce parking needs and vehicle sizes. All of this should lead to less congestion.

Where all of this ultimately shakes out is to be seen, but congestion pricing has potential to reduce a lot of the zero-sum friction. With a few exceptions, most roads are free to use, paid for by tax dollars. Moving to a world in which we pay to use roads will lower our tax burdens, allocate space on the road more efficiently, and hopefully reduce congestion.