9 replies to this topic

[theone]

 

 

Hopefully one of these will drive us to the singularity

 

[Mind]

It is a shame that more countries and communities are not testing driverless cars on low-speed, well-marked, controlled roadways. The risks in these places are extremely low. The cars could immediately benefit the elderly, disabled and disadvantaged persons in such communities, while also proving the safety of the vehicles.

Edited by Mind, 01 December 2018 - 12:10 PM.

[Rocket]

I for one don't trust driverless cars. yet. You are risking loss of life and great physical injury to 1's and 0's flying around on a circuit board. My vehicle has some kind of stop-sense technology so that I can't back up and hit something. I was backing up in my driveway with nothing behind me and system sounds a loud alarm and hit the brakes very hard bringing me to a complete stop... Only there was absolutely nothing behind me. We're not there yet technologically.

[Mind]

Self-driving car crash rates are already LOWER than the human driver crash rate, (per million miles driven). 3.2 vs. 4.2.

 

Human psychology and statistics are very interesting. If you were gambling in Las Vegas and you knew there was a certain way to gamble that would decrease your odds of losing from 4.2 down to 3.2 percent, you would do it every time. Yet when our lives are literally on the line, on the roads, we would rather go with the higher odds of death, driving ourselves. Even if the technology "is not there yet" (not perfect), it is already better than human drivers, and will only get better as each day passes.

 

I wonder what crash rate would be "good enough" for most people? It will NEVER be zero. If the crash rate was half that of human drivers, wouldn't that be good enough? If it was less than 1 per million miles, that would be more than 4 times better than human drivers and pretty spectacular and would save thousands upon thousands of lives each year, yet I suspect there would be a lot of resistance to the cars.

 

It seems for a lot of people the autonomy/freedom and control of driving yourself is quite valuable and worth the much higher risk of death. I like driving myself, but I would have a self-driving car right now (for low-speed city routes) if my city would allow it (and I could afford it), because I know my risk of dying is higher if I drive myself (vs. public transportation or self-driving cars).

[Rocket]

Self drving cars are not conscious and a car on a road is 10000x totally different than an airplane on autopilot. For example if I see a large crater in the road covered by a thin sheet of plywood, I know not to drive over it. a self driving car with a LIDAR will continue to see the road as smooth and unbroken, drive over the plywood and into the crater, killing everyone in the vehicle.  If I see an oil slick on the road, on the freeway, I know that if I drive into it at 80mph that I will kill myself and my passengers. The LIDAR will not see it, not know it is oil, drive into and kill the occupants. If I am driving and see a heard of deer standing on the side of the road, I know to slow down because the deer can and do run in front of vehicles... The LIDAR and computer systems may not even see the deer through the foliage and don't have the intelligence to slow down in case the deer run in front of the vehicle. If I am driving on the Pacific coast hiway and I see a rock slide at high altitude and in front of me, I know to stop. The self driving car will not see it and drive into the rock slide.

 

I can come up with many, many scenarios that self driving cars are not capable of handling. Self driving cars cannot handle many, many conditions that drivers with consciousness can. 

 

I  am not even certain self driving cars are capable of driving on snow covered roads as they cannot see the lane markers, and the road is obscured by the snow... The car doesn't have consciousness to know from experience where the road begins and ends and where lanes begin and end. And then what about when there is no lanes in a heavy snow fall and drivers "make" their own lane for safety reasons? I am not certain self driving cars are capable of driving in heavy rain as the LIDAR cannot penetrate the rain.

 

If I am flying a Piper Warrior and I turn on the autopilot, the plane will hold an altitude and heading and the odds of hitting something in the sky is about 1 in a billion. Practically speaking the only thing I have to worry about is the plane flying into a weather system as the plane isn't conscious. There are no deer in the sky, nor are there rock slides, nor are there obstacles. The only thing to worry about is weather systems and the 1 in a hundred million chances of hitting another airplane.

 

We are a long way away from making self driving cars truly self driving and truly safe.

 

I just so happen to work in engineering on electronic modules. If you knew how loosey goosey automotive engineering gets, you would not trust self driving cars. There are a lot of lazy engineers and there are a lot of mistakes that are not caught and there are a lot of scenarios that cannot be tested and ALL scenarios CANNOT be tested for... You test a certain number of scenarios and extrapolate the data to mean that the system *should* operate Okay.

 

This isn't my opinion. It is a 100% fact that tall scenarios cannot be tested for and what is tested for is extrapolated to mean a system *should* operate Ok. That's one thing in an airplane with practically speaking nothing to hit. Its another thing on the ground with an infinite number of possibilities of things to hit and run into and an infinite number of road conditions with VERY VERY close tolerances between opposing traffic. Planes don't fly within 2 feet of each other in opposing directions the way cars do.

 

 

 

 

Edited by Rocket, 01 December 2018 - 05:26 PM.

[Mind]

Even with all of the scenarios that exist on the roads, self-driving cars are already safer than human drivers (albeit not tested much on snow-covered roads). Humans obviously do not handle "all of the scenarios" very well. In the U.S. alone 40,000 people per year die on the highway, well over 90% of the deaths due to human error.

 

Most human drivers are bad distracted drivers and kill thousands of people. If self-driving cars can cut that number down by even 1 person, it would be worth it, even if the self-driving cars are not perfect. I am unsure why we should resist saving lives. I am unsure why there can't be more controlled testing around the world.

[kurdishfella]

I think people just need more training this is too much, we have all heard the ''women are worse drivers than men'', yes but women are worse drivers but only because they have less training. It's good that things are becoming easier but I think certain control is necessary and not leave it all to technology. My idea would be to make cars that are 100% safe or ditch the whole thing and go for something else.

Edited by kurdishfella, 02 November 2020 - 11:39 AM.

[lawrancecarol]

I think people just need more training this is too much, we have all heard the ''women are worse drivers than men'', yes but women are worse drivers but only because they have less training. It's good that things are becoming easier but I think certain control is necessary and not leave it all to technology. My idea would be to make cars that are 100% safe or ditch the whole thing and go for something else.

 

+1

[Mind]

I still like the idea of self-driving cars, but I am not in favor of the current development model.

 

The original idea was that the you would own the car and it would be a self-contained independent device.

 

Instead, the development is now toward "connected cars" that you don't own and that you don't control - essentially public transportation. In this implementation, there is zero freedom, you basically go where the car company or government wants you to go.

[adamh]

Rocket is on the money. Those scenarios and many others are a danger to ai cars. I don't care if its a few percent safer than the average idiot because the average idiot drives poorly and I do not. Machine vision is simply not up to the task just yet. One person was killed because a semi pulled in front of them but it was white and the car thought it was seeing clear sky so it did not brake.

 

I do think ai and machine vision and processing will improve greatly in the future and selfie cars will be practical in the future. Right now, not so much. We do not have enough data to say they are safer than humans. When a selfie car is on the road, its monitored by someone, by the driver at a minimum. Drivers get lazy and stop paying attention but they could stop an accident. A fully self driving car always seems to get into trouble. They tried it in california and I saw a video of street corners jammed with self driving cars which could not figure a way out of the jam. A human could have sorted it out in no time

 

When they are perfected, collisions will go way down and no one will drive