General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsProposed Safety Law for Self-Driving Vehicles
This is my proposal, not one actually being considered. I think it should be considered.
All self-driving vehicles must be painted in a color that is unique to them and very visible to other drivers. I suggest Pepto Bismol Pink. That way, human-controlled vehicles could give those potentially dangerous self-driving ones a wide berth and avoid them whenever possible.
I selected that distinctive shade of Pink, because no self-respecting human would ever paint a car that color. The color, too, would limit the number of autonomous vehicles sold, I'm sure, further protecting motorists.
dalton99a
(81,073 posts)Nice big dots
MineralMan
(146,192 posts)An even better suggestion. Thanks.
TexasProgresive
(12,148 posts)As a cyclist I have big concerns about autonomous vehicles. At least with humans I feel fairly safe when I look the drivers in the eye.
Thor_MN
(11,843 posts)It was the slightest tap as I was just about to clear his front bumper. He didn't see me until he hit me, even though he was looking right at me. I wasn't hurt, didn't even go down, but he wrecked my back wheel.
I have doubts about every driver.
MineralMan
(146,192 posts)FLPanhandle
(7,107 posts)One day soon young people will be shocked we ever drove automobiles ourselves and probably think that was a suicidal moment in history.
trof
(54,255 posts)And they will be.
Soon.
Igel
(35,197 posts)I stared at it and almost crashed while driving.
This the color you had in mind?
https://www.autoevolution.com/news/pink-vw-beetle-a-joyful-cliche-photo-gallery-78825.html
MineralMan
(146,192 posts)The lime-green background in that photo is lovely, too.
rgbecker
(4,806 posts)the many signals given by drivers to each other as they encounter complex intersections, merging and 4 way stops. I can imagine the pink cars getting stuck, unable to proceed as they are unable to read the signals given by real human drivers.
That said, with the cars clearly marked pink, it will be easy to simply pull in front of them knowing they will automatically slam on their brakes to avoid the collision. It's gonna be great!
hunter
(38,264 posts)Computers are now winning at poker which indicates the most important "tells" are in the card play itself.
I'm guessing the same is true of cars. It's more important to watch what the other cars are actually doing than the body language of the person driving.
MyNameGoesHere
(7,638 posts)And I don't remember any traffic rule or law telling me I had to interpret a wink, wave, grimace or whatever. I have to follow the rules.
Sancho
(9,065 posts)And besides, I had a 1962 Mercury Comet that was Pepto Bismol pink...loved that car, but couldn't get a date for anything.
After I bought a yellow/black Fiat X-19 my social life improved.
MineralMan
(146,192 posts)I used to have a pink International Scout. One of those right hand drive ones the post office used to use. My father gave it to me, and I never bothered to repaint it.
Sancho
(9,065 posts)no AC of course....
My father tried to give me one of the very first Toyota Land Cruisers. I pawned it off on my little brother. It was impossible in the 60s to get foreign parts! Probably the same for International too. You'd wait a couple weeks for a door handle, and a month for an engine part.
rock
(13,218 posts)And those suggestions of others.
NightWatcher
(39,343 posts)Definitely mark them so the rest of us can avoid.
Adrahil
(13,340 posts)Self-driving vehicles will never be 100% safe. But if we look at the historical data, automated systems are typically more same than manually opreated systems. Often by an order of magnitude or more.
Let's not just feel pur way througb this... let's look at actual facts as they emerge.
MineralMan
(146,192 posts)vehicles, there are going to be many problems. Automation is great in a dedicated environment. Not so much when there are multiple variables involved, though. Although we're not usually aware of it, driving involves thousands of decisions on a constant basis, and often also involves unusual situations. The human brain, as an analog device, does fairly well, as long as the human involved is paying attention.
If all roads were uniform, ideally marked and isolated from the surrounding area with walls, then automated vehicles might work just fine. That is not the case, however on our roadways. Not even close.
For me, part of the job of driving is avoiding issues caused by other drivers who are not paying constant attention. It also involves looking far down the road for potential hazards, always knowing where I can go if there is an issue ahead and the ability to remember that there's a stop sign ahead that is blocked from view by untrimmed vegetation. I can also come to a four-way stop and assess the situation, sometimes allowing the driver who does NOT have the right of way to proceed instead of asserting my right of way.
An autonomous vehicle will never have the abilities of an alert human driver in anticipating potential issues. An autonomous vehicle will not be able to avoid, say, the wheel that came off a truck in the lane going the opposite direction by estimating its direction of travel and moving to avoid the collision while avoiding issues with drivers in the lanes going in my direction. It might be able to stop, but it won't be able to avoid that wheel and tire and continue on the journey without any problem.
Driving is an analog thing. Computers are not. Computers don't have any judgement; they have only what instructions are programmed into them. Instant assessment of a rare situation and adaptive driving to avoid problems has saved my sorry ass a number of times. I'm an excellent analog computer.
Adrahil
(13,340 posts)I am an engineer, and I've been working in the field of automation in aviation for MANY years. I am aware of all of the issues involved. And yes, mixed piloted/automated vehicles are the greatest challenge. But your post implies that it is the automated vehicles which introduce greater risk. To the contrary, it's the need, or desire, to maintain piloted vehicles that introduces risk.
And we have been, and are, working on algorithms that recognize corner cases, or cases where the the interaction oif humans and automated systems introduce ambiguities introduced by the complexities or informal communications.
We're still a long way, IMO, from systems that permit driveway-to-driveway automation. But we can, and will see the introduction of more and more systems that will permit automation for long-distance driving. Environments are well suited to these initial steps.
And careful with the word "never" I personally was involved with the development of a see and avoid system that outperformed human pilots by 10x (approx.). Would I trust my life to it? Not yet. But in ten years? Yeah, probably.
ETA: I don't oppose a visual marking to let human drivers know that they are dealing with an automated vehicle. But I oppose the 'shame color' approach. That's just silly IMO. It reminds me of laws requiring someone waving a flag to precede a car driven by a woman. Rather than trying to discourage the use of the technology, we need to focus on good regulations to ensure the these vehicles are properly developed and certified.
MineralMan
(146,192 posts)where we suddenly switch to automated vehicles, so the mix is a given. And, as I said, that is the problem, as well. you say that driveway to driveway automation is a long way off. Well, almost every time I get in my vehicle, I'm going from my driveway to some parking lot somewhere. 95% of my driving is of that type, and I don't see that changing.
If automated vehicles will be used for longer distance travel, then there will probably be a need for separate lanes or even roadways, because there will still be millions of cars with drivers in them for decades. The sheer economics of the thing doesn't make a lot of sense, frankly. Most driving today is exactly what you say won't be ready, driveway to driveway. Commuting is the most frequent use of automobiles, followed by other short trips for shopping, ect. Driving between population centers is way down the list, which is why our freeways and interstates are generally low-traffic between major population centers.
So, the problem is that people want their own sort of autonomy. They want to leave when it suits them and drive where it suits them. That pattern doesn't lend itself to automation, frankly. Today, for example, I'll be getting in my car and driving to one of several supermarkets in the area and then driving home again. It's about a 5-mile trip. Along the way, I will drive through residential streets, complete with those dreaded four-way stops, and then on arterial roads to the supermarket. The arterial streets are all two lanes in each direction. For efficiency and time considerations, I know what lane I want to be in to avoid being delays. There are 10 traffic signals on the route to my favorite supermarket. I can enter its parking lot at three different places. I choose the one to use, depending on whether I will go from there to another location or return to my home.
There, I will have to select a lane and a parking spot. I have personal preferences when it comes to parking in supermarket lots. I park only in pull-through spaces so I don't have to back out. I park far enough away from the store to make it unlikely that I'll be boxed in by two oversized SUVs. How will I indicate those preferences to my autonomous vehicle, I wonder? I also prefer to go into the supermarket through a particular door, most of the time, which also dictates where I park.
How is an autonomous vehicle going to assist me with that frequent journey? And how much is that car going to cost? What incentive will I have to choose such a vehicle? I'm seeing more inconvenience, higher cost, and more time consumed with the autonomous vehicle. Why don't I just keep right on driving to the supermarket for myself? Where is the advantage of this change?
Adrahil
(13,340 posts)And I do think we'll se so,e significant restrictions at first, as there should be. A gradual transition is a given. But rahter than discouraging that transition, we should be be looking at the best ways to make it happen safely.
As to advanatages? Well for people who cannot drive themselves duel to disability, the advantages are obvious. Businesses see an advantage, of course, though that will be seen as a negqtive by many. For me the main advantage would be long road trips... permitting me to relax and enjoy the ride. Another advantage would be the eleconic sensors which could help me avoid collisions by allowing automated systems to intervene.
In fact, in my view, the main dividends of this developments for the short term is augmenting the capabilities of human drivers, rather than completely eliminating the need for them.... we're a long way off from that, IMO.
MineralMan
(146,192 posts)because there will be few of them, compared to driven vehicles. I almost never take long road trips, for example, so that would not be an inducement for me to own an autonomous vehicle. Almost all of my driving is in the urban area where I live. It's a mix of freeway and surface street driving, and requires constant attention at all times. That's not going to change. Where I drive is not going to change.
I want to know instantly if an autonomous vehicle is near me in traffic. I want it to be completely obvious to me. I will adjust my driving to suit that condition, just as I adjust it to suit other issues in traffic. I do not want to discover it only when I am alongside such a vehicle.
Hence my tongue-in-cheek color scheme. I don't care so much whether such cars will be on the road. I just want to know which ones have drivers and which ones do not. I may actually feel safer around autonomous cars than cars with drivers, but I want to know which is which among the many vehicles I will encounter on a given drive.
It's information that I need.
Already, many cars include sensors that detect things like lane drifting, vehicles in blind spots and potential collisions when approaching slower moving or stationary vehicles. Those are great. My next car, of course, will have a rear-view video system. Very useful. I don't mind if my car brakes to avoid hitting a car in front of me. That's OK with me. I wouldn't even mind if it matched speed with cars ahead of me in the same lane and maintained distance between vehicles. All of those features are good ones, and I will welcome them.
However, I will continue to insist on being able to override any such feature at will. I will continue to drive my own vehicle. I will, as I have done for over 55 years, insist on avoiding traffic accidents on my own. My car can warn me of things. That's great. But I will maintain control and attention at all times. When I can no longer do that, I will no longer travel on the roads. Frankly, at my age, I'm sure I'll be dead by that time. This isn't going to be happening as quickly as some proponents believe it is.
So, keep adding those safety features and warning capabilities. I'm all for that, but my hands are still going to be on the wheel and my feet managing speed and braking for as long as I use a car.
USALiberal
(10,877 posts)MineralMan
(146,192 posts)USALiberal
(10,877 posts)studied footage of 100,000 hours of driving. Read more about it.
MineralMan
(146,192 posts)An application lock up?
USALiberal
(10,877 posts)how often does it freeze up?
You really need to read a lot more about this. You are living in the past commodore 64 days.
MineralMan
(146,192 posts)That's OK. Seeya!
USALiberal
(10,877 posts)the challenges of self driving cars. n/t
Adrahil
(13,340 posts)that software safety is taken very seriously in safety critical situations.
If Microsoft word locks up on you, it's not gonna kill anybody, generally speaking. The software that goes into a safety critical item, like a piece of surgical equipment, or an aircraft autopilot, gets a LOT more scrutiny and testing. Of cours,e it also makes it a LOT more expensive.
Check out MIL-STD-882 and DO-178.
In other words, we engineers understand that different types of software require different levels of safety assurance.
USALiberal
(10,877 posts)redgreenandblue
(2,088 posts)The faster human drivers disappear from the streets, the better.
Calculating
(2,954 posts)The self driving cars might work well enough alone on the roads, but when you mix them with human drivers it opens the door to a massive amount of problems. This leaves only a few options:
-Require self-driving cars to be properly marked, and restrict their usage to certain places such as freeways.
-Ban self-driving cars from the roads.
-Ban human-driven cars from the roads.
The top one seems to be the only viable option. I just don't see how self-driving cars will ever mix well with humans on chaotic city streets. "Autopilot" for the freeway? Sure. Having your car drive you to the movie theater? Not so much.
MineralMan
(146,192 posts)but where autonomous cars would not have avoided them.
I've been driving for 55 years now. Here are four situations I've found myself in that I avoided, and that I do not believe an automated vehicle would have. There are many others I could relate, as well.
1. A truck hauling bales of hay in front of me on a four-lane highway with no center divider. I noticed that the load had shifted, so I backed off the throttle and greatly increased the distance between my car and that truck. Within a mile, a dozen bales of hay fell off the back of the truck. An autonomous car would not have seen the potential danger.
2. A wheel and tire coming off a truck traveling the opposite direction on a two-lane highway. This has happened twice to me. I avoided a collision by estimating the path of the wheel and tire, knew the traffic situation around me and the state of the shoulder. I maneuvered to allow the wheel and tire to miss my vehicle, and then continued on my way.
3. A farm truck pulled out of a blind farm road onto the two-lane road I was traveling, making a right turn into my traffic lane. I was quite close to that intersection, and knew it existed, since I drove that road frequently, so I was alert to that possibility every time I drove past it. That knowledge made me aware of traffic conditions around me, both oncoming and traffic behind me. I was able to swerve into the oncoming lane safely and return to my lane after avoiding the truck.
4. Los Angeles, on I-405 in rush hour traffic. A station wagon ahead of me had a mattress tied to its top. The mattress was pushed up by the air. I was unable to change lanes immediately due to traffic density, although traffic was moving at about 60 MPH. So, I quickly, but smoothly, slowed and put about 20 car lengths between me and that station wagon. Sure enough, the mattress came off the station wagon, but I had plenty of time to stop and avoided a collision. The autonomous vehicle could not have predicted that, but I could.
USALiberal
(10,877 posts)hunter
(38,264 posts)That's not a very high bar.
The impact of AI gets real when the insurance companies begin to notice and refuse coverage to those with worse driving records than self-driving cars. It will also make it easier for courts to suspend drivers licenses because people with DUIs and bad driving records will still be able to get around in self driving cars, either their own, or autonomous taxis.
MineralMan
(146,192 posts)For me, they'll have to be near perfect. I do not want my life to end because of a driverless car. Nope.
Adrahil
(13,340 posts)Generally speaking, the average person wants an automated system to be 10-20 times better than a human before they will trust it as much as human.
That is not rational, of course, but it is a pretty universal phenomenon.
That's why augmentation technologies will come (and are coming) before full automation.
Buns_of_Fire
(17,119 posts)Not to give away any of our trade secrets, but the next generation will be programmed to drive ONLY on sidewalks, emergency lanes, and bicycle paths. Our calculations show that this should reduce Unwanted Car-to-Car Interaction by at least 78%.
We'll be good to go, as soon as we develop a reliable self-cleaning mechanism for the undercarriage.
MineralMan
(146,192 posts)Brilliant! Population control as a side benefit, too. What a plan!
NobodyHere
(2,810 posts)Cars driven by dangerous humans should be colored Trump Orange to make sure they stand out.
Personally I think that in a few years self driving cars will be much safer than human driven ones.
USALiberal
(10,877 posts)safer than a human driver.
MineralMan
(146,192 posts)Having reviewed many software products, I had the responsibility of looking for and mentioning bugs, flaws and missing features in that software. Autonomous vehicles run on software. How many programs do you use that are bug-free? I've never seen one, frankly, and I've also been a programmer and software developer.
USALiberal
(10,877 posts)CK_John
(10,005 posts)More of "They can't do that, as they do that".
tblue37
(64,982 posts)Will be one better, but human drivers don't get any better.
NCTraveler
(30,481 posts)I'm sure a lot of people would love to see them branded with a scarlet letter.