This creates a different problem.
Terrorists will have a strong incentive to hack into these cars, sending many of them into people at once. Each hacked car will have to be stopped by security, possibly killing the innocent passengers inside.
Thinking about this a bit more, the way that driverless car makers can prevent hacks is the same way that RSA (a security company) used to verify that its algorithms were still secure.
The car companies can create contests with cash prizes for hackers that are able to break through the vehicle’s security system, and report it responsibly (i.e. privately to the auto manufacturer, allowing them to fix the security hole) before disclosing the issue.
hopefully car companies will act more responsibly than say Equifax
That type of thing works well for mobile phones, ATMs, home operating systems, etc., but I would argue it is too risky for something where many, many thousands of lives could be at risk if that one little bug had slipped by and ended up being exploited by hackers on a mass scale.
I think they should probably just keep all life critical systems (steering, braking, sensors, etc.) on a closed loop that is unable to connect to any air interface or the internet.
@Jugulator20 - we would have a choice with these cars so they couldn’t just sit back and do whatever they wanted, like Equifax did. However, you have to always worry about cheating and cutting corners as we saw with VW and Takata airbags.
At some point, it may come. I don’t think we are there yet or close to it.
I was picked up by a driverless Uber. Normally I don’t take Uber, as I take a just a few taxis per years, and like to support the immigrants that drive taxis (or underemployed PhDs in my town). However, my friend had the app, and we were on a little vacation. The option for driverless came across her phone screen and the car showed up as we stared at the phone in terror, unsure how to react.
The SUV was nicely staffed by two young men, who were beta testing the driverless option in that part of the Phoenix area. They sat taking notes while the car drove itself. It slowed at odd times, with each intersection, rather abruptly, and apparently when it sensed something of concern, usually something we couldn’t see. My main complaint is that it took us on a rather roundabout route to our destination, an area I used to live in and know well. So our charge was about $5.00 more than expected. They also said that it never goes above the speed limit, which is not necessarily a positive, in my book!
So a little disconcerting, but an interesting experience.
I’m surprised they didn’t give you a free ride for using you as a guinea pig.
Human drivers are reskless.
https://jalopnik.com/a-human-driver-crashed-into-las-vegass-driverless-shutt-1820269207
Human drivers are very imperfect but drivers can also use judgement in situations a driverless car hasn’t been programmed for. For instance, in the situation above (truck backs into a stopped driverless car) a human car driver would have been alerted to the truck by the beep-beep-beep of its backup signal and may have been able to either hit the horn to warn the truck driver of the car’s presence or taken some kind of evasive maneuver.
My guess is that driverless cars will eventually be very safe, particularly as they’re programmed for more and more situations as more of them end up on the road, making the movements of other car more predictable.
I would expect driverless cars would sometimes act in ways which human drivers may not always expect. And vice versa. Having more driverless cars on the road in different circumstances, situations, etc. will create more opportunity for “learning” and improvement. At some point driverless cars will presumably be communicating with each other in terms of terms of traffic, weather, etc. at different points on various roads, planned exits/entrances making driving more efficient.
I will definitely miss driving. I am rarely in a car if I am not driving. I enjoy it. Even in the winter.
It’ll be interesting, that’s for sure. I used to think I would never want such a thing–that I want to be in control. We bought a 2017 Rav4 a year ago. It has the lane departure, adaptive cruise, auto braking, etc.
99% of my miles are highway. I bought it for the adaptive cruise, as this route is busy enough I had to re-set the cruise quite a lot after slowing for the driver in front. I would never buy another vehicle without this feature–it is great!
I’m a big fan now of the other safety features, too. There have been 7-8+ times in this year that the car’s radar sensed collision probabilities several cars in front of us, and braked. These were things we couldn’t have seen nor reacted to nearly as quickly.
I’m not ready for a car where I can’t override the controls though.
I said this in another thread. After seeing how well FaceID works, it looks more doable to me. FaceID is basically an eye. And it “sees” well. Response is lightening fast. Sooner or later it will see better than human eyes. Once they can see, driving will be just a matter of information processing. You don’t need GPS to drive.
You may want to leave larger following distances if you encounter your car autobraking that frequently.
Seeing is not the hard part. Remember that a new driver with a freshly printed permit can see quite well, but they are awful drivers.
The real problem is interpretation and response.There is a reason why you can implement facial recognition on a phone but nobody has suggested that a phone has sufficient processing power to become the brains behind a self driving car.
I mentioned NVIDIA a while ago, but most people have no idea of the extraordinary power that exists in a high end graphics card.You can think of the Apple A11 chip powering the latest iPhones as being equivalent to a Yugo, whereas a high-end graphics card is a 911 Turbo. Continuing this analogy further, the NVIDIA Pegasus PX is a Formula 1 race car. They believe it has sufficient brainpower to get to Level 5 automation. We shall see.
The driverless car has more than two eyes. It does not have to turn the head to see other cars in the blind spots.
A family member of mine is looking into possibly moving into the autonomous driving division of his company. The bummer is that they have a very tight confidentiality rule, so they will never share what they are doing and what they are working on.
That’s where a machine does better than humans. They don’t need to experience to learn since experienced humans will program them to act like an experienced driver.
FaceID is tiny housed in a small phone. In a car, you expand its capability to a roomier instrument with a bigger processing power, You plant the eye all over the car. The only issue left is to program the information all eyes gather to driving instruction. That seems quite doable. It’s no longer just detecting what’s a head of you and relying on GPS to move the car.
Let me point out, again, that Google driverless cars have been driving around my area for years, and in that time driverless cars have had one accident. One. You can blather all you want about how driverless cars must be worse than human-driven cars, but they’re not.
Hard to make such far reaching conclusions yet because those cars were (a) driving around urban areas in small numbers and (b) had a human inside.
It will be much more interesting to see the results of Waymo’s AZ experiment. No humans inside.