[Steverino]

Starting later this month, Uber will allow customers in downtown Pittsburgh to summon self-driving cars from their phones, crossing an important milestone that no automotive or technology company has yet achieved. The cars will be the first of 100 custom Volvo XC90s, supervised by humans in the driver’s seat.

The Volvo deal isn’t exclusive; Uber plans to partner with other automakers as it races to recruit more engineers. In July the company reached an agreement to buy Otto, a 91-employee driverless truck startup that was founded earlier this year and includes engineers from a number of high-profile tech companies attempting to bring driverless cars to market, including Google, Apple, and Tesla.

http://www.bloomberg.com/news/featu...eet-arrives-in-pittsburgh-this-month-is06r7on

 

[ChiVolt]

Interesting. My first thought, before reading the linked article, was why does Uber want this technology so early? The technology is not close to being mature. Uber disclosed their rationale in the article:

The goal: to replace Uber’s more than 1 million human drivers with robot drivers—as quickly as possible.

Apparently Uber can't do away with humans behind the wheel quickly enough...

 

[DonC]

Super interesting. Autonomy is moving at a rapid pace. Note that GM has been testing autonomous Bolts in San Francisco and Phoenix (Tucson?, can't remember exactly) and that Ford has announced totally autonomous cars by 2020. Hope we'll actually see this type of progress. Then again, GM delayed its Supercruise, which is essentially Tesla's Autopilot, to 2017 after initially saying it would be out in 2016. In truth it's either difficult or impossible to predict timelines for things you've never done before. It might go smoothly or you might hit a showstopper. No way to tell until you do it.

 

[2VoltFamily]

I must be missing the real point. They will still have a driver.

 

[ChiVolt]

I must be missing the real point. They will still have a driver.

During the trials. That' not to say they'll have an in-car driver or attendant once the trial is over.

 

[sparks]

I must be missing the real point. They will still have a driver.

And if they don't, I'm not getting in!

 

[Dutch]

In order to understand this you have to understand the concept of "machine learning".

When human drivers train, each one has to train from scratch. When machines train, what one of them knows ALL of them know.

In the relatively short time Tesla's "auto-drive" program has been running the cars, collectively, have gained more experience than any 40 year human driver and now that "hive mind" is expanding it's knowledge to the less common and outright freakishly rare incidents and circumstances. Once those have been learned then ALL of the cars will immediately know how to handle those situations for the rest of time, even cars fresh off the production line.

That means, for example, that you might live in death valley Arizona for most of your life and have zero experience with snow and ice. but if you go visit your relatives in Michigan in the winter, the car will already know perfectly well how to handle itself because it's already learned from all the OTHER cars that drive in that area.

and it that alone doesn't stun you, the updates in learning will get ever more precise and be updated constantly to the extent that your car (coming from Arizona) will know about the pothole in the road two blocks from your relatives house and will avoid it automatically when you get to that area.

The transition from human drivers to autonomous is going to happen very VERY suddenly and sooner than most think and once it does anyone who wants to "human drive" is going to wind up paying insurance out the wazoo for the privilege.

In a way I'm a luddite about this. I -enjoy- driving (most of the time) and will be sad to see that skill go away. But it's going to happen and I'll probably be pretty glad when I'm 82 and blind or whatever and need to get to my doctors appointment.

 

[vltwatcher]

I wouldn't worry about this happening anytime soon except perhaps in special traffic lanes or roads. In fact , these "self driving" cars have two humans monitoring them during the trials with the primary human touching the steering wheel at all times. The legislation and litigation will limit this severely for another decade , if not longer, even if and when the technology is there for driving everywhere.

Sent from my XT1060 Moto X DE

 

[Steverino]

GM's Mary Barra said that progress is being made much faster than most people realize. She's not alone in that assessment. But the general public has not yet caught up with it.

My mother's knowledge is one data point: the Tesla driver that was killed. Even though her eyesight is failing, can no longer drive at night and worries how she'll get back and forth to stores, doctor appointments, etc., she want's nothing to do with autonomous cars. It's going to take the public some time to accept the idea. But like the introduction of ATM machines, drive-through windows and microwave ovens, the convenience factor will win many over to doing things a new way.

 

[Dutch]

I wouldn't worry about this happening anytime soon except perhaps in special traffic lanes or roads. In fact , these "self driving" cars have two humans monitoring them during the trials with the primary human touching the steering wheel at all times. The legislation and litigation will limit this severely for another decade , if not longer, even if and when the technology is there for driving everywhere.

Sent from my XT1060 Moto X DE

I firmly believe based on research done and working in a related field, that you are wrong. I guess we're just going to have to agree to disagree.

 

[edk-austin]

Like Dutch, I also believe it will become illegal to "drive human" on public roads. The statistics will show that humans kill other humans at a higher rate than the autonomous machine. Insurance costs already enjoy discounts to drivers willing to let their insurance company "monitor" their manual driving. Just a matter of time. If it's any consolation, this will also spur the development of more private tracks, where you can pay to "drive human". (I think we're already seeing that) And of course, if that's too much for you, you can always take the driving experience in virtual reality, where simulators continue to get closer and closer to the real thing.

 

[vltwatcher]

I firmly believe based on research done and working in a related field, that you are wrong. I guess we're just going to have to agree to disagree.

I can understand that the technical solutions will be there eventually (not convinced we're there yet) to be able to drive in all conditions that a human is capable, but that's not all that is required. We live and commute in a time dominated by insurance and litigation, not just physics and technology. Truly driverless cars everywhere, sharing the roads with reckless drivers, bicyclists, and pedestrians are a ways away IMO. I can see it happening sooner on isolated roads or lanes (which requires infrastructure support), but not everywhere, which is what Uber really needs for it to be useful.

Sent from my XT1060 Moto X DE

 

[Dutch]

insurance and litigation are what is going to kill "human-driving" faster than you will believe.

When a collision occurs, especially if the auto-drive vehicle was in the right and the other (manual) car made a mistake, a couple of things are going to happen at the same time.

1) due to it's sensors, the auto-drive vehicle will have recorded every aspect of what happened. If the human driven vehicle was at fault there will be no, zero, nada, "wiggle room" or excuses. Their insurance company isn't going to tolerate absolute proof of error many times before insurance rates start climbing... fast.

2) due to it's sensors, the auto-drive vehicle will have recorded every aspect of what happened. AND once it's analyzed, now ALL other auto-drive vehicles will know how to "watch" for that kind of situation and avoid it.

Human error at an unsafe intersection continues until the intersection is changed or until all the drivers in the area know that the intersection is dangerous. Machine learning means only ONE car has to find out "hey this spot is dangerous" and it will pass that along to all of it's "brethren".

And a tangentially related topic... there are already cities and counties discussing the topic of "If self diving cars obey the law 100% of the time, what are we going to do about the lost revenue from speeding and traffic fines?" (Hrmmm... I could swear everyone always says speeding tickets are about safety, not revenue... I guess we've ended that little charade). Also parking companies are having fits with the concept because you might not need to park your self driving car, just send it home and have it come back when you need it.

 

[Tim S FL]

First they get rid of a lot of taxis now they're going to get rid of a lot of guys trying to make a buck
At least Uber lets them sit in the car for a while

 

[rmay635703]

If we enter into straight machine driving only we are actually entering into a much more dangerous place.

A place of greater tyranny , less choice, a place of vested interests requiring specific "problems" addressed for a fee by specific companies.

Have a flat tire? It'll be addressed by someone automatically at five times the cost.

Don't like it, don't own a car