California May Soon Allow Passengers In Driverless Cars (reuters.com)

According to Reuters, California's public utility regulator on Friday signaled it would allow passengers to ride in self-driving cars without a backup driver in the vehicle. It is a big step forward for autonomous car developers, especially as the industry faces heightened scrutiny over safety concerns. From the report: The California Public Utilities Commission, the body that regulates utilities including transportation companies such as ride-hailing apps, issued a proposal that could clear the way for companies such as Alphabet's Waymo and General Motors to give members of the public a ride in a self-driving car without any backup driver present, which has been the practice of most companies so far. The California Department of Motor Vehicles had already issued rules allowing for autonomous vehicle testing without drivers, which took effect this week. The commission said its proposed rules complement the existing DMV rules but provide additional protections for passengers. The proposal, which is set to be voted on at the commission's meeting next month, would clear the way for autonomous vehicle companies to do more testing and get the public more closely acquainted with driverless cars in a state that has closely regulated the industry. It also comes as regulators across the country are taking a harder look at self-driving cars in the aftermath of a crash in Arizona that killed a pedestrian.

Comments

Re:How come each and every one of these cars...

[fluffernutter]

Also, why doesn't it have to to a vision test? For a self driving car, it should be able to operate on two out of three sensing mechanisms in all conditions.

Re:How come each and every one of these cars...

[Anonymous Coward]

I don't think so Tim.

Self-driving is actually safer than a human with less than 5 years of driving experience. However there is a lot of missing context.

Teens who learned on a Standard transmission car, in BFN (eg, any rural area or city under 10,000) often know the roads like the back of their hands and know exactly what to be aware of, and most of their accidents are intoxication related. Not weather, animals, or speed. Most intersections are stop signs and the occasional red light, and little traffic flow. Kids in small towns know every place they can speed and get away with it.

Teens who learned on an Automatic transmission in a major city, have no spatial awareness. There are a dozen different things going on, and you often have to drive into places you've never been that may have been a half mile away from where you were comfortable. Speeding in the city is dangerous as hell.

A self-driving car will be smarter than the latter. As it has full spacial awareness, GPS and rarely has no reason to go faster than the speed limit. It will not be as smart as the rural driver, and is more likely to cause accidents because rural driving requires being aware of wild life and road conditions that city drivers have no experience with as well. Take any driver from east of the rocky mountains, and put them on a road in BC, California, Oregon, or Washington, they will slam the brakes on at every turn, they will skid off the road if there is frost or snow, they will slide into intersections.

The point here is that automated vehicles have more awareness in busy environments, but almost no awareness in rural environments, because rural driving requires skill, where city driving requires spacial awareness to get you from point A to point B. If you transplant the rural and city driver into each others environments, the rural driver will drive like a senior citizen in the city because they don't know where they are and they've never experienced multi-lane roads, freeways, advanced left-turn, etc. Where as the city driver in the rural area will likely hit the first vehicle that doesn't their turn signals.

Re:Are this motherfuckers...

[dgatwood]

Depending on whose numbers you believe, the National Safety Council says that the U.S. average is 1.25 deaths per 100 million miles. So that number is in the ballpark.

However, your estimate of the number of autopilot miles is probably about an order of magnitude low. There are news articles from late 2016 claiming over 300 million miles traveled with autopilot/autosteer active. If it's not at least half a billion by now, I'd be surprised, and I wouldn't be shocked if it hit a solid billion already.

So if you ignore Waymo (too small a sample size) and Uber (trying to deploy FSD before their tech was ready), and concentrate only on Tesla, that's a pretty sizable drop in fatalities — around a factor of 5–10.