Eventually they won't need one but it's not legal for them to test the technology without a human present right now due to the potential for accidents. They need a person to override the AI if necessary.

It will take a long time to get to the point where they are safe to use unattended but it's a necessary process right now. I can't wait till they are ready!

Hopefully driverless cars will never come to pass, especially after today when a pedestrian was killed by a self-driving car.

https://www.theverge.com/2018/3/19/17139518/uber-self-driving-car-fatal-crash-tempe-arizona

ETA oops sorry, didn't realize there was already a thread about the accident;

I think self-driving cars will be common in less than 10 years. The advantages are significant.

I've read that one of the biggest challenges is figuring out how to operate in bad weather - rain and snow- that can obstruct the electronic visua/spatiall sensors. Whether or not that is easily resolved, fair weather use is here today. These cars have been successfully wandering the Bay Area for a number of years.

I have to admit, ONE casualty from a driverless car sure beats how many casualties we have from human driving. Computer accuracy will always surpass that of humans.

I agree, JCrigler. The death is sad but pales in comparison to the number of deaths each day from driver error or distraction. It sounds cold to look at it like that but the statistics are hard to ignore. Many are killed every day by cars through no fault of their own.

i Have a different perspective about driverless cars.

yes it’s 1 driverless car mortality

but think of how few driverless cars are out there. That’s 1 fatality out of say a very liberal estimate of 600 driverless cars in state of Arizona . that’s an accident rate of 0.15 %

the pedestrian mortality in Arizona is 153 out of 5 million vehicles on the road. thats 0.003% fatality rate - far fewer fatalities on the road given how many more vehicles are out there.

If we actually compare number of fatalities by miles driven , the number of current mortalities as a percent of miles driven is even lower.

The driverlesscar as a proof of concept is a winner. But As the means of mass transportation ? Not quite there yet.

yes the autonomous space has the potential to dramatically reduce mortalities attributable to driver distraction. That’s the promise

The reality is that driverless car requires a degree of unprecedented accuracy and fine tuning that capacity to make correct decisions is still work in progress. Imagine the millions of random permutations and combination of poor road conditions, bad weather, and human(pedestrian) error, etc that the AI has to correctly identify.

We need to analyze/recognize why the AI failed in this instance and devise ways to prevent them from re-occurring. feed that analytical Post Mortem back into the AI decision making process and improve its accuracy to avoid the reoccurrence.

Unfortunately it’s going to take many more deaths in the near future to achieve the reduction in fatalities required to make the driverlesscar as safe as with-driver cars.

In the original thread, I had said (with an edit for clarity):

"A TV station's video story I saw showed a partially scrunched bike on the sidewalk with garbage bags attached, homeless-person style. It's a leap of an assumption but it's possible there may have been mental health/cognitive issues involved with the poor victim. Or maybe not. But it's fair to say that few people are out at 10 at night crossing a non-residential street mid-block with a loaded bike, without paying attention to an oncoming car. A driver with good perception might have had a problem too, it may never be known."

The incident took place at night on a street in a commercial area. You're saying that the AI failed. How do you know that?

I agree Elmer. AI cars won’t completely end all traffic accidents & deaths on roadways. At least not until every car on the road is AI & we don’t have humans or animals on the roadways - as we can be quite unpredictable.

I don't believe that driverless cars will ever become common. My friend was almost run over by a driverless car just the other day.

I told my children (all in their 40s), that if self-driving cars get to be common, the only time that I will be on the road is going to the store at 4:00 a.m.

From what I remember about our ages, by the time self driving cars are common, we’ll all need one as we’ll be too old to drive.

Yes there could be tech issues, but you're already risking your life everytime you drive being around other drivers who are drunk, texting, speeding, doing makeup, running red lights, and all the other human activities that kill thousands each year. I still think the AI will be far safer in the end.

I think the real risk will be those who insist on driving their own cars when most are using the self-driving. Most people, including the one who engage in bad behavior, believe that they are better-than-average drivers. People tend to have blinders towards their own driving behavior, citing a lack of accidents as proof of good driving when that can be more luck than anything.

There has been some talk of having laws requiring that there be a human operator ready to take over when the computer system determines that it is out of its depth. At first this seems like a reasonable precaution, but it neglects to factor in that the human likely won't have been paying attention, isn't aware of the current situation and then must very quickly grab the controls, assess the current situation, figure out what to do to avoid the situation and then start to .... >>CRASH<<

The only benefit to such a system where the human is placed in charge in case of emergency, is to allow the autonomous car company to dodge blame saying "We notified the driver of the problem, he/she didn't respond correctly so they are to blame."

I'd expect it will be 5 years or so before they are anything but a rarity, and perhaps 10 years before the odds of seeing one in a normal day of driving would be > 50%. During that time the autonomous cars may be a higher risk to those around them then the average driver. Simply given that you can often expect the bad behavior a bad driver will engage in. Whereas an autonomous car might choose entirely novel, completely unexpected actions when its sensor data or control algorithms are confused.

If it is solely about safety for the driver and those in the vehicle, and those in proximity to the car, there are many technological measures that could be implemented using some of the same technologies that are being developed that would be vastly easier to bring to fruition then fully autonomous vehicles. Using similar sensors that might eventually allow autonomous control the vehicle cold determine when the driver is engaging in bad behavior: texting, running red lights, excessive speeding, lane departures, and then take some action to signal that behavior or correct it or punish it (For instance a system that detected every attempt to send or receive a text message while the phone and automatically added a fine to that person's phone bill)

Looks like double jeopardy here -

- first, on the part of the new-fangled machine, and

- second, on the part of the old-fangled human manager, who, it appears, missed the problem, as well.

To the major detriment to the one who wasn't protected by a surrounding metal box.

ole joyful

I think they are a bad idea. I mean maybe for very limited trip to the store or dr's offices for people who can't drive but can't get a ride. But a car should not have a mind of it's own. Think of how many times you've had your computer or phone mess up or freeze on you. What if a car did that while you were on the freeway? And weather isn't always predictable. A few years ago we got hail out of no where. No mention of it in the weather report. Imagine being in this car when random weather happens. They'd have a lot of tinkering to do, but I still don't think it's a good idea mainstream wise.

Once autonomous vehicles are out there the largely forgotten phrase "Blue Screen of Death" will come back, and it will involve actual death.

I don’t mean to scare everyone but all the jets in the air are 90% automated. The captain sets the perimeters & clicks on autopilot as soon as your airborne. It can even land the plane if necessary.

Just like the self driving cars, a human is there to take over if the computer poops out.

But there aren't any pedestrians, bicyclists, red lights, yellow lights, up in the sky and there is someone all seeing who prevents them from running into each other.

But then there is Die Hard 2.

Good point ;-)