Not sure if they are planning to sell any in the US , maybe just Japan to start
Well that sounds cool.
I’d seriously look into buying one after the tech has been proven(read I can afford one)
Level 3 is too dangerous. What happens is that drivers assume the car will self-drive so don’t pay attention to what the car is doing. The car gets into a situation it can’t handle and the driver takes too long to figure out what is happening. So a crash results.
As opposed to now, when a driver thinks that they are great drivers and can text/talk/put on makeup, get into a situation they can’t handle, and a crash results.
As long as I can afford it, and statistically it is as safe or better than current tech, I’ll be first in line.
I can see using it on highways with no issues. But no way in city traffic now.
It certainly helps that I live in a place with no people and lots of highway…
The article in the OP says that the level 3 cars “will allow drivers to let their vehicles navigate congested expressway traffic.” So there’s no idea of self-driving in city traffic.
How much will you pay for insurance, assuming that insurers are willing to insure such a car?
And who will get sued the first time a level 3 car has an accident?
Tesla has an autopilot now so I assume some company is giving those people insurance.
Tesla’s autopilot is classified as level 2, and it already does a great job with congested expressway traffic. Mostly because the two most important things necessary for congested freeways is to stay in your lane, and don’t hit the car in front of you. It handles those functions extremely well.
Tesla’s level 2 can change lanes by itself, merge, and take exit ramps, but not nearly as well. What level 3 will need to do in congested traffic is learn how to time gaps for lane changes, and even more importantly know when a lane change is necessary based on traffic, not merely which lane is needed for exits and merges.
That is exactly my reaction. Wouldn’t touch it with a ten foot pole. Maybe level five. Maybe.
We in the airline business struggle with the equivalent of Level-2 automation. And that’s with two professionals sitting there being paid to pay attention. And when most of our problems develop over 5s and 10s of seconds, not 1s and 3s.
Joe or Jane careful driver will usually fail at this task and Bob or Sue typical = lousy driver doesn’t stand a chance.
It is a mystery to me why NHTSA is allowing these half-way house failures onto the road. Given the awful press that will accompany the mass of accidents as Level-2 cars becomes commonplace, I don’t really see why the manufacturers are so eager to blacken their own eyes delivering an unsafe product that amounts to a failing beta test for the good idea of self-driving.
TL; DR. Managing a car on level 2 is no different than actually driving a car, except that input is not required for every single steering, braking, and acceleration change. If somebody is not able to manage a car on level 2, then they shouldn’t be driving at all.
As I’ve posted before in other threads, I think level 2 will greatly decrease crashes. Tesla’s data seems to support this. Back when I commuted to work, every week I would see several rear end collisions pulled over on the shoulder of the freeway. These accidents would not happen with level 2. Yes, there have been a few news worthy incidents of Teslas in rear end collisions. So rare it makes the news, not a daily occurrence on a 10 mile stretch of freeway.
There is simply much less information to manage in a car on level 2, than in an airplane on autopilot. In the car on the freeway I manage the maximum speed, and the car manages the actual speed. In congested traffic there isn’t much for me to do. In medium traffic I need to pick a speed appropriate to the flow of traffic, so generally faster than what cars are actually traveling, and let the car adjust down. In light traffic, I pick the speed I want to cruise at. That is something that is set, and then ignored until it needs to be changed.
I also manage lane changes. In light enough traffic I tell the car to change lanes, and it does. In heavy traffic I’ll usually take over and manually change lanes.
I also make sure the car doesn’t do something wrong. Usually that is just phantom braking, or slowing down to let a car merge, when the car next to me isn’t actually going to merge (that heuristic needs some work). Sometimes when lanes split the car wants to take the wrong one, and needs to be corrected.
During all of that, I’m watching for other drivers behaving badly. This is something that needs to be done whether I’m driving or the car is. This can result in simple stuff, like slowing down a bit to let somebody in, or emergency braking because, “WTF? get your head out of your nethers and look before you change lanes!”
Most situations that develop very fast the car is going to handle. If the driver in front slams on their brakes, the car will notice and respond. If somebody next to you swerves into your lane, the car will respond. When the autopilot gets it wrong and sends me into a closed HOV lane, then I override it, and get it corrected. If I’m too engrossed in my phone to notice I’m heading down a closed lane, then what are the chances I would have noticed the car in front of me braking?
I think a much better analogy than a plane using autopilot, is riding with a student driver. The student driver is handling control inputs (and is good at it), but needs higher level guidance and occasional correction.
Anyway, I think freeways will be much safer with lots of cars on level 2. Nothing like that feeling of being tailgated, looking in the mirror and just seeing the top of the person’s head, as they’re looking down at their lap. Surface streets are different, and not nearly as well handled. I’m also extremely skeptical of anything higher than level 2. I do not think anybody is close to having it ready.
Good info. Sounds like I’m a bit confused about the levels. I just did some reading and it seems I was thinking more of level 3 than 2.
Nevertheless, once level 2 cars are commonplace, I predict many of their drivers will simply disengage their brain completely from the driving task once they’re in a level-2 compatible situation and HAL beeps its “I’ve got your back” signal. They’ll assume HAL is handling everything. Even those things that are far beyond HAL’s actual capability.
Hell, today we have legions of drivers who are trusting their car to stay in its lane and not hit the vehicle in front while they intermittently text and talk and eat and …, despite driving a vehicle with ZERO autonomous capability beyond basic straightline rolling stability.
On one show, I saw a demo of GM’s self-driving car. They showed it had sensors and cameras watching the driver. If the driver’s attention strayed from the road (say to talk to the passenger), the system beeped to alert the driver and then, if the driver continued not paying attention, it brought the car to a halt. So with that sort of system, you can’t goof off on your phone or read or fall asleep and expect the car to do all of the work.
I’m sure that will happen, but as you say it’s already happening now. I’d rather an automated system is paying attention than nothing is paying attention. It’s similar to the wrong argument that motorcycle helmets cause injuries, because they make people feel safe, so they ride more dangerously than non-helmeted riders. Or more recent, that people will gather in big crowds because they think their mask will keep them safe.
Counter to that, Consumer Reports recently gave GM’s Super Cruise the highest ranking in a test of level 2 systems, because it has the best driver engagement system, even though the car handling of Tesla’s autopilot was superior. I think Tesla really needs to pay attention to this. Tesla’s system can currently be satisfied by holding the steering wheel off center enough to apply some torque. As long as my hand is there, the car doesn’t care about what I’m paying attention to. It will be fine, right up until it isn’t.
For me, somebody who tries to be a conscientious driver, I like level 2. Because I’m not directing the car in every single action it takes, I have more time available to pay attention to the cars around me, and it has my back in the moments when I’m checking the navigation or adjusting something.
That is certainly the promise of automation when used right. Same idea in our world. By not being distracted with the minutiae you have brain bytes for the more strategic picture.
If one of the consequences of level 2 is that more people aren’t looking at their hood ornament and are instead seeing the road signs indicating their exit is 2 miles, 1 mile, 1/2 mile ahead maybe fewer of the [suddenly exit across 6 lanes of traffic to the sound of squealing tires] maneuvers will be happening. It is to be hoped for.
That’s probably about the level of nag system that will be required. It’ll be interesting to see how they harden the software to prevent installing a patch to kill the nag feature. It’ll also be interesting to see how smart the driver monitor is. As @echoreply just said, Tesla’s is trivially foolable.
It’ll also be interesting to see how GM, et al, react when the consumers complain about their cars bugging them too much. It, like the worst of the 1970s automatic seatbelts and seatbelt nag systems may prove to be politically and/or commercially untenable despite being very good for safety.