So much possibility for good.
So much possibility for bad.
It’s like Facebook, if you think about it.
1 Like
That’s why there are so many overlaps in the coverage; chances are that not all the cameras and sonar are going the be covered at the same time. But, realistically, if the car doesn’t see the proper movement in all the cameras during the first few feet it starts going, it will stop and tell you to fix the problem.
I would have assumed they are using parallel processing with redundancy, as failure is potentially lethal, much like aircraft. Don’t they run like three independently designed systems in parallel, and constantly compare results in aircraft systems?
1 Like
Of course if there’s no shoulder because there’s a huge snowbank piled up in it, there’s no place to pull over. My fear is what happens when an increasingly inexperienced driver gets told to “take control” in bad conditions.
3 Likes
So, my car can drive itself to the job that I no longer have, once it has been outsourced or replaced by AI.
This is not the future that The Jetsons promised.
2 Likes
Tim Harford’s name & latest book was mentioned here recently. Here’s a quote from the Guardian article.
This problem has a name: the paradox of automation. It applies in a wide variety of contexts, from the operators of nuclear power stations to the crew of cruise ships, from the simple fact that we can no longer remember phone numbers because we have them all stored in our mobile phones, to the way we now struggle with mental arithmetic because we are surrounded by electronic calculators. The better the automatic systems, the more out-of-practice human operators will be, and the more extreme the situations they will have to face. The psychologist James Reason, author of Human Error, wrote: “Manual control is a highly skilled activity, and skills need to be practised continuously in order to maintain them. Yet an automatic control system that fails only rarely denies operators the opportunity for practising these basic control skills … when manual takeover is necessary something has usually gone wrong; this means that operators need to be more rather than less skilled in order to cope with these atypical conditions.”
(I was thinking about just this sort of thing recently. I used to ride (and hope to again in the near future) one of these on public roads:
Cycling never felt so safe: not because of the twin drum brakes, or the fact that you can’t fall off it, but because motorists always consciously noticed me. If I was on a bike, I’d trust a well-designed system over a human motorist to see me. I’m not so sure about outliers like recumbents [unintended pun noted for future use], and I’d hate to end my life as a Tesla bug report.)
3 Likes
I dunno.
I ride bikes a fair bit, and it used to be pretty rare for someone to outright not see me. Still, they’d badly underestimate my speed or where I was going or whatever (and cause themselves and me all sorts of problems).
More recently, I don’t get seen by people “driving” who are actually dealing with cell phones in one way or another. I’m getting to the point where I’d almost feel better about a computer that’s not prone to distractions like that- know what I mean?
2 Likes
Fundamentally a big part of the problem with driving is that much of it is too simple to engage us so our minds wander off… Computers ARE much better at this sort of thing. But human visual processing is actually VERY advanced compared to computers. When people ARE paying attention their vision IS superior. In early radar scopes, they discovered that operators were able to see targets that were below the noise level…
4 Likes
Best I can say is that from my experience I think I’d have to discriminate between seeing and consciously being aware. I don’t know how many times [actually, not that many times] I’ve been on a bike and made definite eye contact with a motorist waiting at a side junction, seen them look directly at me, focus on me and lock eyes, only for them to pull out right in front of me. Two of these occasions, I ran into them at a lowish speed, and they stopped: both times they said they just didn’t see me. (I stopped looking at drivers after that: I look at their front wheels instead.)
On the trike, I see motorists actually react; they don’t quite know what I am, so I suspect the conscious mind kicks in and tells them to pay attention. I don’t know how much of this is actually true; but it seems to work for me.
More recently, I don’t get seen by people “driving” who are actually dealing with cell phones in one way or another. II’m getting to the point where I’d almost feel better about a computer that’s not prone to distractions like that- know what I mean?
Ohh, yes.
3 Likes
Seems to me there will be a lot of congestion caused by this sort of behaviour. Automated cars pulling off the road to “handle” a situation they can’t deal with. Then trailing cars noting cars with 4-way hazards flashing on the side of the road and subsequently slowing down to avoid.
I recall a story from Google’s self-driving car team a few years ago; they found they had to introduce some aggressiveness into to car to deal with 4-way stops. Their car would end up stuck at a 4-way because it needed to see absolutely perfect conditions before it would proceed. Wouldn’t even make small creeping motions to indicate it’s expecting to be next to move. Too timid for lack of a better word. Dealing with how humans really drive is difficult for the always-follow-the-rules robot.
I think it’s going to be like watching robot soccer games. Watching them mishandle the ball, falling over, the agonizing pace is amusing. All translated to highway speeds with multi-ton vehicles.
No disrespect to the robot soccer developers, or the self-driving car developers. I am generally tech-positive. I just think we’re a long way from general purpose self-driving cars. Might work very limited areas, like physically separated lanes.
And I still can’t understand why a car company would open itself to the massive liability when they provide the driver.
And I’m guessing that weather that is too bad for self driving cars will be more common than weather that is too bad for human drivers. Or, rather that instead of having a few fender benders, the roads will shut down because the self driving cars refuse to go out in it. And weather problems will be made worse because much of the research on these is going on in sunny California. The fact is, many people live in areas where people deal with bad weather on a very regular basis and that is foreign to these guys. Instead of being a 1% problem that they can ignore by not going anywhere, for many people bad weather is a 20% problem that the HAVE to be able to deal with.
3 Likes
I’m sure this is a novel idea to the people working on it and is completely unconsidered…
This thread is full of a lot of “armchair engineering” that assumes that everyone here is full of original ideas that no research team would consider and that the US government isn’t going to wrap this in 20 layers of testing and regulation before letting it go live. Have some trust that smart people are actually, you know, smart, and working on the problems.
You’re right of course. I shouldn’t have suggested that these cars will be on the road before they are ready or that the developers aren’t fully considering all the challenges.
I guess what I’m really getting at is that what look like huge strides in autonomous vehicles today are really just baby steps and that the general public should be prepared for a significant wait. They’re working on the “easy” stuff right now.
1 Like
And that’s why I think that these thing will brick themselves in bad weather or road conditions.
That’s better than the alternatives.
1 Like
Undoubtedly, but in parts of the country with weather, the utility of a vehicle that bricks itself 20-30 days a year is vastly reduced.
If that winds up being an actual issue, then people won’t buy them. I’m not sure it will be an issue because, surely, the engineers are aware of this “snow” thing.
1 Like
Automotive recalls are pretty common with parts designed by experienced automotive engineers. And automobile companies have become pretty good at designing vehicles that can be used in a wide variety of environments and legal regimes. Notice how the trunks of cars are designed so that by swapping out a small number of minor parts either North American or European license plates can be accommodated. But the software here is being done by software engineers who are not, I suspect as experienced at that sort of thing. And what seems like crazy esoteric extreme weather in coastal California that can be designed around, is common enough in much of the country that it has to be designed FOR.
I’m guessing that they wouldn’t eliminate manual controls for exactly this reason. Hey, look, it’s snowing, I guess I won’t be updating the TPS report on my way to work today.
1 Like
3 Likes
