This looks like the "Sensing" feature that Honda has implemented on some of their other vehicles. I just bought a 2016 Accord that does the same thing -- there's a camera mounted on the windshield, another camera in the front grille, and a radar sensor on the front bumper.
Calling it "self-driving" is kind of a misnomer and I think the article kind of blows it out of proportion.
It will track the car in front of you and keep a safe following distance, keeping either the maximum cruise control speed you've set, or whatever speed the vehicle ahead of you is driving, whichever is slower. It will accelerate or brake accordingly. It will also attempt to stay in the lane by using the onboard cameras for tracking the lane markings.
The lane keeping assist is not nearly as autonomous as the article makes it out to be. It does not like to work on sharper curves on the freeway, for one -- the system will disengage and tell you to steer manually. It still wants you to keep your hands on the wheel. It must be looking for very very subtle movements on the wheel, because the system will yell at you if you take your hands off the wheel for longer than 10-15 seconds.
All in all it's a pretty cool feature for longer road trips (keeping in your lane can just get kind of tiring, even with cruise control) but it's not the sort of autonomous driving that the article here paints it out to be.
Free-market ideology if you ask me. The first sentence suggests the hapless government wants to spent $4 billion on research for something you you can have today for $20K. A little later it mentions Washington's "time and billions" spent on fuel-efficiency instead of "saving lives". The government has also funded electric-car research and encouraged Detroit to build them. Contrasting self-driving with electric drive is just another way to suggest the dead hand of bureaucracy is impeding the invisible hand of the market.
By the way, the government has saved drivers' lives, too, by the thousands, more often than not over industry objection. Seat belts, unleaded fuel, crumple zones: it's a long list. The article's bias isn't just bias; it slants away from the facts.
It was stupid for this article to start talking about the $4 billion Obama has talked about for research.
This article is about features and not a real self driving car. If anything the features make this car very dangerous in the hands of people concerned with their phones and tablets and kind of proves the need for investments like the $4 billion.
To me it sounds like the free market is about to make the road damn dangerous for the rest of us. To quote the article, "For instance, some owners have posted videos of hands- and foot-free driving on YouTube and the car inevitably makes a mistake.".
In all fairness, the assistive driving features introduced to date are probably a net win. (After all, they've been available on high-end vehicles.) However, we're probably approaching the point where the automation is sufficient that people will routinely stop paying attention given that it "mostly works" for highway driving in good weather.
In cases like these, if a collision occurs, who is at fault? Presumably it is the "driver", especially in this case, right? But if/when this gets into the court system, when do people start going after the "self-driving" manufacturers, who have much, much deeper pockets than your average car owner.
Especially if people start arguing that the manufacturers are effectively going "pay full attention at all times. Nudge. Nudge. Wink. Wink." knowing that people routinely don't pay attention when automated systems like these are used.
There are places in the world(I saw it in Netherlands) where you have a "cruise control forbidden" signs. Yes, even the dumb cruise control that only holds a set speed. I saw those signs near construction zones and such, I guess the idea is that people should pay 100% attention in there.
This, unsurprisingly, has generally been the death of previous attempts at automated cars, especially when the efforts included modified roads and attempts to have cars form convoys to cut down on air resistance. [1]
[1] Granted, this is anecdotal, but it comes from a former faculty member who taught my control systems course in college. Last I heard he was working on systems for auto landing UAVs on aircraft carriers. He had worked on a number of grants related to autonomous driving and said that every time they had something technologically ready the lawyers basically couldn't find a way to make it happen.
To me (who, to be fair, has no experience in this area), it seems that the only long term solution to this liability issue is a government-funded and controlled fund accompanied by legal changes prohibiting these cases from going to "normal" court.
For something that is a demonstrable way of saving huge numbers of lives, but still carries some risk, using the model of the vaccination side effect fund seems like a good one to me.
For complete self-driving cars, it's seems obvious that the manufacturer should be liable. And a lot of people want it that way:
- Google wants to be liable.
- The passengers doesn't want to be liable.
- Insurance companies would perhaps like to go after the entity with deeper pockets, too?
Oh, and efficiency saves lives in and off itself. (One possible path, using less oil for cars, we can use more oil instead of coal in other applications. Thus mining less coal. Coal mines are crazy dangerous per Joule mined comparatively.)
In economics, the Jevons paradox occurs when technological progress increases the efficiency with which a resource is used but the rate of consumption of that resource rises because of increasing demand.
The only solution out of this paradox with respect to carbon fuels is taxation. Unfortunately, as much as we'd like, a global carbon tax regime is not practical, nor enforceable. So we're stuck.
Doesn't need to be global. European countries (and others) had high taxes on petrol for a while now. Extending that to a general carbon tax might be workable.
As ben1040 put it, "the article blows [self-driving] out of proportion." Jklowden hypothesizes that WSJ blew it out of proportion in order to paint a picture that the government is wasting money at best and not letting the free market innovate at worst. It wouldn't surprise me that a publication like Wall Street Journal would advocate in favor of a free market.
(They overlap quite a bit. But eg business don't like competition, especially foreign competition. But competition exactly one of the reasons the market works well for society.)
I imagine a free-market ideologue could make a point like: the government shouldn't spend money on fuel-efficiency, just put a tax on emissions and let the market figure it out.
Seat belts were invented by English engineer George Cayley in the mid-19th century,[3] though Edward J. Claghorn of New York, was granted the first patent.
The government did not invent the seatbelt. All it did was monopolize it through the patent system and later criminalize not wearing it. How heroic.
I didn't mean to suggest the government invented seat belts. It did mandate them, not only wearing them, but circa 1964 their very availability in cars. The industry, and a fair number of car buyers, didn't want buyers to be reminded that cars could be dangerous.
Heroic? No, but what government is heroic? That seat belts and seat belt laws save lives every year is indisputable. Perhaps to you that's an impermissible curtailment on your right to die in a crash. It might help to remember also benefit indirectly through lower insurance premiums.
(In answer down thread to why I posted my reply, apologies. I meant to reply to another question about why the article mentioned electric cars, but clicked the wrong link.)
I didn't mean to suggest the government invented seat belts. It did mandate them, not only wearing them, but circa 1964 their very availability in cars. The industry, and a fair number of car buyers, didn't want buyers to be reminded that cars could be dangerous.
I don't think that's true. Nash had made them standard years before, multiple other manufacturers had them as options, and Ford promoted cars with their Lifeguard system[1], including sealtbelts, long before those laws.
The reality is that people didn't care, as evidenced by the fact that in 1981, thirteen years after seatbelts became mandatory in all cars, the CDC estimated that only 11% of car occupants used them[2].
Yes, the WSJ is known for its free-market ideology :)
But I think they have a point. The gov't has largely focused on fuel economy for decades now, while tens of thousands of people continued to die or suffer horrible injuries in cars. Sure, seat belts, crumple zones -- a good start. Unleaded? Not sure how that saves you in a crash.
Unleaded fuel has saved many lives and probably prevents a lot of crime. There are plenty of studies that show a correlation between lead and crime rates.
> Yes, the WSJ is known for its free-market ideology :)
Yes, in its editorials. But it's otherwise known for quality, more-or-less objective journalism, which makes this article a bit of a disappointment.
Then again, I'm reminded of that Paul Graham article about working with newspapers―pretty much any topic except politics or war, according to him, is written about because someone with an interest pitched the article to the author. Perhaps Honda has more to do with this article's bent than the WSJ editors.
> But it's otherwise known for quality, more-or-less objective journalism
Sorry but no. Ever since Murdoch took over and replaced the editors the quality has noticeably dropped with a lot more inherent bias even in the non-editorial content.
And within the editorial section the change is pretty clear. It's shifted from being centrist to at times being on the hard right.
Even with politics (and by extension, war). The opposition will be routinely pitching stories to counter whoever is in power.
A huge percentage of stories are pitched by PR companies. I once shared office space with a PR company and saw how much it happened; it's eye-opening. The story in the careers lift-out of the paper - pitched as a profile for a new CEO of their client. The article about working from home, pitched to get a mention for the new consultancy client.
I would say that crime stories might be some of the only ones unaffected. But even then, the stories about crime statistics can be pitched - the giveaway will be a quote from security screen manufacturers or an industry rep.
I really wanted the automatic cruise control but the truth is it's a hassle in the city on the highway. I don't want to constantly adjust the distance between the car in front of me and my car as conditions change so I end up just managing it myself without cruise control or setting it 'classic' mode where it doesn't use the radar.
That said shortly after I bought the car I drove it across state and the automatic cruise control and lane keep were FANTASTIC. I know the lane keep seems kind of lazy but when you're driving across the plains with heavy cross winds it makes things SO much easier. Really a road trip feature.
The 'you must steer' alert is fun. I've discovered one or two places near my house where the street is straight enough (along with the alignment on my car) that it will briefly trigger some times. Note that it's ALWAYS active (unless you turn it off), it's not only in use when you're using the lane keep function.
EDIT:
Also, sadly, Honda doesn't have a full-stop automatic cruise control like some other brands. If the traffic gets too slow it will disengage and warn you that you're in control. It's never something you could use on a standard street, highway only.
That's sad that it doesn't handle heavy traffic well. That should be where it does best. Tesla's system, for example, is nice on the open road but absolutely wonderful in heavy traffic.
Too bad about not handling full stops, too. I wonder if that's a technical limitation, or if it's just something they didn't want to allow.
I thought Autopilot was nifty but it wasn't until I slogged through ~40 minutes of rush hour without touching the pedals or wheel when I realized how awesome it really was.
It's useful at-speed but the killer app really is rush hour traffic.
I thought Autopilot was nifty but it wasn't until I slogged through ~40 minutes of rush hour without touching the pedals or wheel when I realized how awesome it really was.
High-end Mercedes/BMW/Audi/Lexus have had these features for years now, although they still force you to keep your hands on the wheel, presumably for liability issues.
(And it's still awesome, it makes stop-and-go traffic bearable)
I think that's just a matter of risk tolerance. The traditional manufacturers are taking it slow because they have a huge potential downside, and limited upside. Having the best autonomous system on the market wins some sales. Being on the receiving end of Toyota Unintended Acceleration II: Autonomous Boogaloo would be a disaster. So they're taking it easy.
Tesla is in "go big or go home" mode. They intend to grow sales by a factor of ten in five years. They have no history and are just beginning to build brand loyalty, so they want to show everyone that they're the best. That makes them much more willing to take risks.
Restarting from a full stop requires good pedestrian awareness. In city traffic, there might be people crossing in front of the car when it's stopped. Restart also requires getting the driver back on line. They're probably on a phone. That's why some driver input to start motion is desirable.
The user interface problems for automatic driving are very tough.
I don't think the parent was talking about restarting from a full stop, it looks to me like he was saying that his car's radar-cruise system won't actually stop the car in heavy traffic, so if you're not paying attention and don't brake manually in time, you could rear-end someone.
Honda's system cuts out at some 'slow' speed like 35mph. If you don't do anything the car should fully stop but that's because you'll trigger the automatic emergency breaking.
The problem with the system is it's relatively simplistic. You set a following distance (there are 4 options) and the car will never get closer to the car in front of you than you've specified. It will slow down as necessary to maintain that distance.
This works fine on empty highways but in traffic it doesn't help. Since it won't go to (and resume) from a full stop you have to fiddle with it constantly. In tight traffic the distance has to be modified or people will continually jump in front of you (causing the car to slow down and back off).
I'll also say that it waits a surprisingly long time to start braking to the point it can be a little unnerving at times. I know it knows there is a car there (there is a symbol on the dash to show that it sees a car) but it doesn't act until you get much closer.
A system like Tesla or Volvo have that can bring the car to a stop and then speed back up or could decide to try to maintain a distance similar to that other drivers are maintaining would be much nicer.
Still for long trips the Honda system is great. And you can always put it in 'dumb' cruise control mode if necessary.
I'll also add that on my Mazda, there's a button on the steering wheel to adjust the following distance. There's only 4 choices like the Honda's. It appears to be selecting by car lengths, so this distance would be greater (at the same setting) at higher speeds than at lower speeds. I haven't used it enough yet to really verify this though, but it seems to be the case, and seems nonsensical to design it any other way: of course you're going to want to follow at a greater distance at higher speeds. In practice, I don't have to fiddle with mine much at all; I just set a speed and leave it there. But I also don't use it in tight traffic; I wouldn't trust it for that, and I'm sure I do a better job of driving smoothly (and fuel-efficiently) since I can see what other cars are doing around me and use my predictive ability, rather than constantly speeding up and slowing down as people change lanes in front of me as you say. (If traffic's not too tight, and is fairly smooth, I will use it though, but I do reduce the following distance to the second-lowest setting or sometimes lowest. But I wouldn't use it in stop-and-go bumper-to-bumper traffic.)
I don't believe there's a way to set mine in "dumb cruise control" mode. I'm not sure why you'd want to. It acts like normal CC as long as there's no one in front going slower than you, and if there is someone going slower, you're either going to want to slow down, or go around them. One thing that'd be nice, though, would be an additional indicator showing the speed of the vehicle ahead so I can see the speed differential and decide when/if to change lanes.
Wow, I'm glad I didn't get a Honda and pay a premium for such a system. I got a Mazda3 instead, and the radar cruise control in that is excellent. I cruise all the time at 30mph in it (I work at a facility where I have to drive a couple miles at that speed to get to my building), around turns even. It'll come to a complete stop if the person in front of me stops.
It seems weird that it would need to brake at all most of the time to maintain a following distance. Unless the car in front of you suddenly brakes or you're going down a steep hill or something, wind resistance usually provides all the braking you need at highway speeds.
I think that with cars that offer 'full stop' cruise control the driver must press a button to indicate that they're ready to roll again after the car has stopped it's self (or perhaps tap the gas).
Not positive, that what I remember hearing. That way it's still the driver's responsibility, but it's easy enough that it doesn't annoy the driver.
Oddly, it used to require manual intervention to resume after any stop longer than five seconds. Then they bumped it up to 30 seconds on highways. Then they (mostly) removed it.
The Model S now only requires human intervention if it has detected obstacles (e.g. pedestrians) while stopped, otherwise it should be able to resume automatically.
Restarting from a full stop requires good pedestrian awareness.
Cars that accelerate automatically from a stop already have the radar/ultrasonic sensors necessary for collision avoidance. So they should never accelerate into a pedestrian (and not allow you to, either!).
If anything, collision avoidance systems have better pedestrian/cyclist awareness than human drivers do, since they can "see" objects in blind spots, etc.
Tesla switched to electromechanical brakes with the introduction of it's Traffic-Aware Cruise Control (“autopilot”) system. https://www.teslamotors.com/blog/your-autopilot-has-arrived says “In October of last year we started equipping Model S with hardware to allow for the incremental introduction of self-driving technology: […] and a high-precision digitally-controlled electric assist braking system.”
On the one hand, this makes it sound like a technical limitation with hydraulic brakes.
On the other hand, learning the response of your particular car's brakes seems similar to a number of tasks that the car automation has to do anyway. It's a different program and running on different processors, but ABS is an existence proof of this and works fine with a variety of braking systems.
I'd be interested in hearing from someone who knows more about automative systems and automation.
>I really wanted the automatic cruise control but the truth is it's a hassle in the city on the highway.
Yeah, that's one thing I've noticed. Of the three options you get for following distance, even the "closest" one leaves a pretty long gap between you and the vehicle ahead of you.
So people will pass you, pull into that gap, and then your car will slow down to open up a long gap between them and you. Lather, rinse, repeat.
That sounds more like a problem with other drivers than with the cruise control system. Most drivers leave far too little distance between themselves and the next car in moderate to heavy traffic.
yeah, the automatic braking is worth it. Yesterday someone abruptly cut me off; the car hit the brakes to avoid a collision more quickly than I would have reacted myself.
I've only had it go off once when someone did something stupid (don't remember what). I was paying attention and slammed the brake on as fast as I could.
Didn't matter. I could feel how easy it was to push the brake in. The car had started to brake a few milliseconds before I could.
I would have been fine in that case since I was already paying full attention, but I was glad to know it was working. I live in an area with occasional deer on the highway and I've had VERY close run-ins before.
I blame Elon Musk for this shift in the meaning of "self-driving".
It used to be a term that was only applied to driverless cars, but last year Elon announced that they plan to make the Tesla a self-driving car at the same time as announcing similar driver assistance features.
While Tesla's driver assistance features are much more advanced than these, it's not much of a stretch after Elon already fired the first marketing shot by shifting the goalpost of what it means to be "self-driving" so much.
Language is organic and constantly changing. I used to insist on calling my drone, a remote controlled quadcopter, as drones were those autonomous military vehicles dropping bombs on weddings in the Middle East. But everyone now from the news to the FAA calls my little RC toy, a drone.
Sure, it bothers me to watch words like self-driving and hoverboard lose their meaning. If someone as influential as Obama can't protect my favorite Egyptian god, Isis, there's no point in blaming anyone or trying to fight it.
When you put it that way, it's really interesting to think of it as a language shift rather than just a question of marketing honesty. If the average person describes the car he already owns as "self-driving", perhaps that will accelerate acceptance of real driverless cars.
After reading that same article a half dozen times since starting my career, I still don't know what to call myself.
Engineer? I don't have an engineering degree.
Computer Scientist? The stuff I do feels a lot more engineering than science-- I have never completed a formal scientific methodology when writing code. Besides, my profs at university do a lot more talking/writing than coding.
Programmer? As others have said, this cheapens me to the simple action that I take, not the problems that I solve.
Developer? Along with being conflated with Real Estate Developers, this does very little to describe even the domain of my expertise. Is a chef a food developer?
People are indeed quite confused about what current driver assist features are capable of, but Tesla has been quite clear about exactly what they offer now, what they plan to offer in the near term, and what they hope to offer in the longer term.
The main problem I've seen people complain about is the use of the word "autopilot" to describe Tesla's features. And that's only because people don't understand what a real airplane autopilot does. I blame the general population for jumping to conclusions.
I agree with everything you say, and to me the goalposts shifted because he was purposely ambiguous in the way he announced autopilot at the same time as promising a fully self-driving car soon. So now a self-driving car isn't going to be a whole new product, but an accumulation of added features, one of which is autopilot. The implication is that cars with radars and lane detection are already self-driving cars that just need a few more upgrades to cover that other 20% of the time that you're not on the highway.
I just want to say, I have a similar vehicle system (Eyesight, rather than Honda Sensing) and I agree with everything you are saying. It isn't self-driving, it is just a bunch of really cool driver assist technology.
Unfortunately the reason why these systems fail on sharp turns is that the lane-line literally falls outside of the camera's vision. Aside from adding two more cameras (which I'd welcome) I cannot see them fixing it in software.
I totally would have purchased a 2016 Accord but no all-wheel-drive option. To me AWD, Blind Spot, Cross Traffic, and Eyesight made for one amazing overall safety package. The Accord has some nice unique features (e.g. Android Auto) but they need more than just Honda Sensing as OPTIONS.
Subaru has the best overall value with their Eyesight system when considered with the rest of the car IMHO. It's why I bought my Forester (not in the price range of the Civic). Definitely not self driving, but adaptive cruise control works in steady traffic- it has a tendency to "rubber-band" when traffic changes speeds- and has saved my bacon by preventing a collision when the car in front of me stopped. The only downside is that it is fully camera based and has no radar at all, and on a road trip heading west this winter the setting sun killed the Eyesight system and the cruise control completely disengaged and I had to manually control my speed for about an hour.
Agreed - my wife just bought a new Subaru with the Eyesight package. I was very dubious at first, but after a few long freeway trips, I'd much rather drive her car than my older vehicle. I believe that Subaru in Japan has published some impressive statistics about accident reduction with the Eyesight package. IMO, these technologies should be mandated on all new cars.
I'm impressed Honda is offering it on all models (LX, Sport, EX, EX-L, Touring) instead of saying "Sorry, Touring only".
Now that I've had Honda Sensing for a few months I really like (in a rental right now and miss parts of it), but I'd like real blind spot indicators and the cross traffic alert.
AWD is actually not that important in safe driving. AWD ensures you can always go forward. In a bad situation you don't care about going forward, you care about stopping and steering. So brakes and tires matter. We all have 4 wheel brakes, so set there. Get some awesome snow tires and 2WD and you will steer/stop fine in the snow. Plow up a hill through snowbanks? Maybe not.
AWD assures you always have traction, tracking means steering, therefore AWD means you retain steering in bad conditions. If you move to a snowy area the majority of vehicles have either AWD or 4WD, the few that don't NEED snow tires just to make it work.
AWD used to be questionable because of the price and fuel economy, but these days thanks to computer controls, it is close to 1 MpG, and the price has come down to within a thousand dollars of a non-AWD vehicle (in fact several manufacturers do 1-1.5K AWD upgrades).
> AWD assures you always have traction, tracking means steering, therefore AWD means you retain steering in bad conditions. If you move to a snowy area the majority of vehicles have either AWD or 4WD, the few that don't NEED snow tires just to make it work.
I have lived my entire life in areas where it snows. This is fundamentally not true.
Traction has to do with tires. Good snow tires - awesome traction. Bald summer tires - bad traction. If you need to steer around something, the tires is what matter - NOT AWD or 4WD.
The reason the majority of cars up north have AWD or 4WD is because people are sheep, and believe in marketing material over facts.
To me this sounds horribly unsafe. You are lulled into this state where you don't have to pay attention to steering UNLESS you have to pay attention to steering, and you will find this out suddenly. So if you always have to pay attention, even when the system is active, then what exactly is the point of the system in the first place? Wouldn't it be safer to always pay attention?
I guess this is an empirical question. Is it just as safe to (a) not pay attention until suddenly you have to pay attention, vs (b) pay attention all the time?
I suppose the sudden aspect of (a) is the dangerous part.
On the other hand the constant vigilance demand of (b) is also dangerous (we know people are bad at staying vigilant for long periods of time when a stimulus doesn't change much)
Does anyone know of any peer reviewed published work on this?
I was actually thinking that may be the case, but (anecdata warning!) after switching to a car with lane assist + adaptive cruise control (so similar to the described one I think), I can concentrate on the road ahead much more now. I didn't realise how tiring the "normal driving" is until now. The ability to stop glancing at the dashboard to check speed and not working that hard to turn (even comparing to assisted steering) is amazing.
You may think those things are in muscle memory, but actually they do tire you out. Even after a 1h drive, I'm much more relaxed than before - and that means I'm actually able to just look ahead and think about the road, not about what I'm doing.
unless heavy traffic on highway, this is already achievable for me with usual cruise control (I did few home trips that are 1500km far within 1 day - tiring but doable... 1 hour is a breeze in good car).
if in heavy traffic, I turn off cruise control anyway, you better focus properly.
I was once in an accident caused by the driver using cruise control. My driver had his feet relaxed and off the pedals when a car swerved in front of us.. he reacted by going to hit the brakes and because his feet were not in a position where he knew the relative position of the pedals, he hit the accelerator instead of the brake. We swerved off the highway down an embankment at 75-80mph. We were all very lucky to be relatively uninjured, but to this day I refuse to touch cruise control or any other automated driving features. I don't know if modern adaptive cruise control and collision avoidance would have helped in my case, but I really, really don't trust a driver who is relaxed and not being constantly vigilant.
When driving with cruise control, I have a habit to lightly touch the brake pedal once in a while (without actually braking) to "warm up" my muscle memory for emergency situation.
I just bought the 2016 Civic Touring last week and it came with this. It certainly seems to be getting ahead of things to say it's self-driving. Your description is accurate, it's more of a helper than truly autonomous. I will say I was concerned, because there was a Car and Driver review where they seemed to indicate it had issues thinking you were drifting and trying to correct when you were fine, but I haven't had a problem. It detects the lane very well, and the cruise control works like a champ.
I am glad to see features like this coming to less expensive vehicles.
I have the 2016 Accord with Sensing and occasionally the road departure mitigation system will freak out and warn you even where there is no drifting at all. At one point on my drive the pavement goes from asphalt to concrete and the lane markings lighten, and this fools it nearly every time.
Overall, the whole system was worth it for the forward collision mitigation system alone. It can be a bit of a nanny but it can save me from rear-ending someone.
I find the lane keeping assist system to be of dubious utility. It can't completely drive the car--it doesn't like relatively mild curves on the freeway--so if it can't completely drive the car what good is it? The adaptive cruise control is nice for those situations when highway-speed cruise control is actually useful--which is to say, it's not too useful in a congested city. What would be more useful is if the thing could automatically follow the car in front of me in a traffic jam.
I hope the government requires these forward-collision mitigation systems as standard equipment in a few years. They are that useful. The rest of it? Meh.
I have to say that with respect to regular cruise control, I got out of the habit of using it years ago and don't remember the last time I tried it. I don't drive a lot of wide-open roads and even when I'm on rural interstates that are relatively low traffic, it's the case that I just got out of the habit.
I'm all for collision avoidance systems but, for the rest, I think I'd need to genuinely (and relatively safely) be able to take my attention off the road and do other tasks. Maybe decrease the amount of mental energy in stop and go but the big win is being able to do other things than actively drive.
I'll have to test it out on a different bit of highway, perhaps one of our older more weathered areas where lines are less defined to see how it really holds up.
I'm fortunate in that I usually only drive off peak hours, so cruise control is usually an option. I typically only use it on long drives, but my wife is a cruise control junkie and she loves the features on the new Honda.
I'd agree, the most useful bit seems to be the collision mitigation, but I still find the other stuff cool :) Oh, and the lane change camera is pretty awesome (though not relevant to our self-driving topic).
I didn't find this to be an improvement over the right side mirror. The optics of the mirror are such that there is practically no blind spot on the right at all. Cars on the right vanish from the mirror only when they are right next to my car--at that point I will see the car through the window itself when I glance over to the mirror.
It's the left side that's more problematic. Oddly the left side mirror does not have a true blind-spot mirror. There is this left part of the mirror that has different optics but it doesn't cover the whole blind spot. So there is a point around the left passenger door when the car is completely invisible in the mirror. This could have been remedied either with a true blind spot mirror on the left (I think Ford Fusions have these) or with a camera. Instead there is neither. I might just glue a $5 blind spot mirror on there to remedy something that could have been standard equipment.
Sometimes they made strange choices with this car.
What this WSJ article described definitely seems more like a Adaptive Cruise Control (ACC)[1] than true autonomous driving [2].
Note that ACC has been around for LONG time - as early as 1992. And the new Honda Civic's tech package's ACC + Lane keeping is only considered to be Level 2 Autonomous driving, whereas Tesla has achieved Level 3 autonomy [2].
Mercedes has had the ability to keep a safe following distance for a decade or more. The lane assist is not something I've heard of, but I stopped staring at the latest and greatest features on cars around the time that Mercedes came out with the follow feature.
Lots of cars have had these features for a while, but it's really only been in the higher end models. What's interesting is that it's becoming available in cheaper cars now, and is a step towards it becoming a standard featured in all cars.
The Driver Assistance Package ($2500) on the CLA250 brings the total of the base model to ~$35K. It includes DISTRONIC PLUS (which regulates distance and speed), Blind Spot Assist, and Lane Keeping Assist. I'm not sure what that entails in the newer models. My 2013 MB seems to only alert me if I'm drifting onto the shoulder. I don't think it corrects it for me.
I have a 2014 E-class with DISTRONIC PLUS, and it turns the steering to stay in the lane. I'm assuming a current CLA has the same if it has DISTRONIC PLUS.
The shoulder drifting feature is a different feature, at least on my car, and it's not tied to whether cruise control is engaged or not.
In Germany, you can get an A-klasse with DISTRONIC PLUS (and absolutely nothing else) for $30k, but it's an almost $4k accessory, it's not standard, and if you're shopping in that segment it would be very weird to pick that option and nothing else.
So Honda's offering is actually pretty spectacular!
I would guess you are correct about how they make sure your hands are on the wheel. I've done some car hacking and one of the things that surprised me was how sensitive some cars steering wheels are to pressure. It was impossible to keep it in the same position according to the car's internal electronic systems with my hands on it. This is car is 4 years old, so it's probably only gotten more sensitive.
I have a Tesla, that also requires hands on the wheel from time to time. I'd assumed it was capacitive. It hadn't occurred to me that it might be subtle movements, because the wheel is unresponsive until you make a purposeful movement, at which point it starts moving with your hands; and you can just rest your hands against it lightly to let it know they're there. But of course it could be sensing subtle forces, even if it switching modes only on larger forces.
Is there a test for whether it's capacitive? Like, does the fact that it works with gloves on mean it's not capacitive? (That's diagnostic for capacitive touch screens, but they're smaller and trying to detect finer motion and from just the fingers.)
One interesting thing to try, still, would be to stuff a full water bottle into the gaps in the steering wheel to emulate a hand, like so: https://www.youtube.com/watch?v=qi2oIRMwmZY
This definitely works for many capacitive systems, but I think this might also work for some of the movement-based systems, since it will add extra mass/inertia to the wheel and the water sloshing around could emulate small hand movements.
I have the 2016 Honda Civic LX with Sensing. It doesn't require me to keep my hands on wheel. However, it keeps losing the lane, requiring you to steer manually. Maybe the lane markings are not clear enough here in Los Angeles.
Also, the adaptive cruise control often brakes abruptly, even making the front collision warning system go off. I also don't know if it activates the brake lights when it brakes (since traditional cruse controls don't). This means that I usually brake manually.
All those things make this feature less useful than it should be. Before I bought this car, I actually did some research on this, but I haven't found any test that compares how well these features do in practice.
For normal (non-adaptive) cruise control there are few situations where the brakes would make sense, but one of those is a steep hill. An easy way to find out what happens if you have a decent grade nearby: set a low speed and point yourself down the hill. I commuted over a large grade for some years and all the cars I drove would brake on it. Can't speak for every system ever made, obviously.
traditional cruise control is just about keeping constant speed, no adaption to situation, road inclination etc. my 12 year old BMW has one, wonderful feature and keeps me still engaged with what's on the road
Traditional cruise control will most definitely hit the brakes. Set a speed and then drive down a nice grade, you'll see it in action. The oldest vehicle I've seen this in was a 1989 Dodge Caravan, you could literally watch the brake pedal move.
I have a 2016 VW GTI that has the same feature set. It's cool and all but more of a safety feature if you happen to fall asleep while driving than anything else. Just hope that if you fall asleep that there is sufficient contrast between the pavement and the lane markers.
Americans could rather use a robot highway driving instructor.
"Consider moving over to the right line; you're driving at the speed limit, and a speeding car is approaching; you may confirm this in the rear-view mirror."
"I have detected that you came to a full stop at the end of a generous freeway entrance ramp in light traffic. Suggested future action: look over the shoulder as early as possible and match the speed of the traffic."
"Suddenly exiting out of the left lane is dangerous. Please know which exit you're supposed to take, watch for its approach early, and change lanes ahead of time. If you miss an exit, do not make a sudden, dangerous action. Look for an alternate route or U-turn starting at the next exit."
Yes, there are some cases where, well, stuff happens. A driver forgets until the last minute. Many times I see them hit the blinker a split second after they've changed lanes.
But far more frequently, I see the driver who is drifting between lanes at will, without thinking of the blinker, and seemingly without a care in the world. This is disturbing, especially since if I'm seeing the behavior, I'm behind the driver. That means it's my job to figure out where the winds will blow them next.
Too many times I see idiots posting on Facebook or texting as I pass. Many times this is when we're both doing 65mph or more. Insane stuff.
This isn't full auto driving, but every little step helps.
If you want to cut in, gaining an advantage over traffic, it behooves you to keep your intent undisclosed. Some drivers in the advantaged lane will make it difficult to merge if they see the intent, by closing the gaps. (Those who make it easy are enablers of cutting-in behavior, despised by everyone).
It's safer if you don't announce your intent; you don't want to be encouraging the driver in the target lane to be trying to close the gap that you're swinging into.
This is fucking ridiculous. Surely all drivers have to change lanes at some point to get where they're going; why do they all actively attempt to thwart other drivers from getting into the lanes they need?
That's the competitive reality of driving in congested, urban areas.
Drivers who cut in can save a lot of time over the course of a commute. They disrupt traffic flows and delay everyone else. So efficient traffic flows have to be defended.
> Surely all drivers have to change lanes at some point to get where they're going;
Yes, and some drivers like, for instance, to pass around 25 other cars who are making the same merge.
The problem is that traffic engineers disagree with your assessment of efficiency. You are effectively using aggressive driving (a crime) in order to circumvent actual laws (yield to mergers) and to self-righteously defend the worldview that you have constructed in order to provide a superficially logically consistent explanation for your emotional frustration about losing out.
> traffic engineers disagree with your assessment of efficiency.
Do they?
> You are effectively using aggressive driving
Not me, personally. (I ride a bicycle and use transit from time to time, and the occasional rental car on weekends and holidays: out of the vicious rush-hour traffic.)
Fact is, if you throw a blinker, there will be those drivers who will reduce gaps and not let you in. Usually not in merge situations (as in, you can't join this freeway, bugger off!) but situations in where the lane change is not necessary (you're not entitled to swerve into this lane just because it's moving faster!). I agree with them; they are doing the right thing. It can usually be done safely, within the "two second rule" for following another vehicle, depending on speed. Be it right or wrong, safe or unsafe, if you want to outwit those who are doing it, you have to do the wrong thing an not blink.
Not letting people join a freeway is despicable. (It happens too: "Hey, you can merge here ... just not in front of me, thank you very much!") The do-not-blink principle does not apply there, because it's obvious that you want to merge whether you blink or not, being in a merging lane and all! My comments don't apply to that situation at all.
I'm not sure I'm understanding your position very well. Are you saying that people exiting the slow lanes to enter the "moving lane" constitutes a hazard, and by punishing those actors by not letting them in, you are minimizing that hazard? If so, i'm inclined to agree philosophically.
However, I don't believe anyone should impose their philosophy with 2 tons of steel.
> in order to circumvent actual laws (yield to mergers)
Mergers must yield to drivers already in the roadway. The only obligation those on the roadway have is to not speed up or slow down to prevent the merge.
That is completely inapplicable to drivers who are already merged into the correct section of freeway and just want to zig-zag between lanes based on what they think is faster at any moment.
If you're not one of these drivers, it benefits your trip time if you prevent their actions by not creating easy opportunities for them to slip in.
If you are one of these drivers, it benefits your trip time if you successfully carry out the zig-zagging. To that end, it behooves you not to announce your intended moves. If you do, you reduce your opportunities, because you're relying on those who enable this behavior by letting zig-zagging drivers in. In other words, not-blinking is a rational game strategy.
Okay, maybe we are talking about different things. Skipping a clearly identifiable line is one thing - the driver sees that they could just get in line and decides not to.
I am more interested in why you would actively avoid making space when there is no clear way to get into the lane except by finding an open space.
Because, for one thing, it's not someone's god-given right to move from a lane that is nearly standing still from one that is moving fast.
"Gee whiz, I just joined this freeway, and damn it, I'm gonna edge my my way all the way to the fast moving left most lane at practically ninety degrees to the traffic flow."
If they are just changing lanes to find the one that's fastest at any moment, sure. If their destination is only reachable from that lane then yes, it is their right to use the public roadways regardless of how much it displeases you that you share them with other people.
> If their destination is only reachable from that lane then ...
If there are jerks who don't want to let anyone in front of them in a situation where you're just trying to merge in, it will not help you not to blink; my comments about not blinking as a strategy do not apply.
Still, those merge-preventing jerks are acting rationally in their self-interest. If a flow of traffic behaves in such a way that it impedes the ingress of a joining flow, then that flow does in fact reduce its trip time. (The effect on the average trip time for everyone is another matter, which they don't care about, obviously).
I don't think it's about time or efficiency at all. I think it has to do with perceived fairness and our self-righteous culture. The faster-moving disappearing lane gets an inherent advantage, which could have been renounced by merging too early. People feel cheated. The problem is that, objectively, people merging all over the place creates more traffic than a uniform zipper merge so we all suffer.
In reality, the laws are very explicit (at least in GA where I live), and the person merging has priority. Anyone in an existing lane when a lane disappears is legally required to yield.
Additionally, everyone is entitled to this same advantage so it's really not unfair at all.
> at least in GA where I live [...] the person merging has priority [...] anyone in an existing lane when a lane disappears is legally required to yield.
That sounds so crazy and contrary to everywhere I've been in the civilized world that I refuse to believe it (as you have written it) without some chapter and verse citation from the GA traffic code.
If it is literally true, it means that in GA, I can drive down a merge ramp and not even look at what is coming down the freeway; they have to hit their brakes or swerve to adjust to what I'm doing. I'm in a merge lane, which ends, and thus I have priority over anyone in an existing lane that isn't ending!
Any special rules about making it mandatory to let merging vehicles in have to be somehow crafted to they only apply to the obvious situations that they are intended for.
Or they have to apply in special places where signage clearly indicates that two flows are merging such that neither has the right of way. For instance the "Form 1 Lane" traffic sign or whatever. "Form 1 lane" effectively means both lanes are ending, and a new one is starting with equal priority access.
"The driver of a vehicle about to enter or cross a roadway from any place other than another roadway shall yield the right of way to all vehicles approaching on the roadway to be entered or crossed."
My search suggested highway on ramps are "other than another roadway."
"If you are about to enter or cross a highway from an alley, private road or highway, you must stop and yield the right-of-way to all other pedestrians and vehicles already traveling on the roadway or sidewalk you are entering or crossing;"
I've never found that to be an advantage. Usually if I indicate that I want to pull in, the driver behind me will immediately make room. On the rare occasions when they don't, it often turns out they had a good reason for keeping me out (like something they could see but I couldn't.) I'm comfortable assuming that's always the case.
Not to mention the idiots who refuse to turn on their lights, even in a rainstorm with overcast skies, unless it's after 9pm, because after all, they can still see the road just fine!
Not using the blinker, or using one falsely, should be a rare practice which is always strategic. When it's done out of stupid neglect, it's such a waste.
I agree completely, but the very first thing they'd do would be figure out how to turn it off or muffle it. Bad drivers are not the type to take feedback well or let things like the rules get in the way.
Yeah... the same people who resent seat belt laws, because they're afraid of being trapped in the car after an accident, and feel much safer being thrown through the windshield so they can get up and walk away.
Seriously, putting cars with brake-assist technology in the hands of drivers who are already overrepresented in accident statistics is a good thing. Life doesn't have to be hard-mode all the time.
But, on the other hand, teenage children getting $20K cars will always fall into the problem category of ... teenage children getting $20K cars.
Even if a trend breaks out whereby cars with self-driving abilities are chosen by parents who are gifting their spoiled offspring with new automobiles, that trend is temporary; eventually the capability will become commonplace, like ABS and air bags. Teenagers will then be getting their licenses using such cars, whether or not their parents are spoiling them with new cars.
Eventually, even relatively poor teenagers will be learning to drive and taking their road tests on 25-year-old third-hand beaters which have still-working self-driving functions.
Just for comparison, you can get a Subaru Legacy Premium with Eyesight for $25,835 (or a Crosstrek for thereabouts, and an Outback for $2-3K more), since at least summer 2015. So nothing Honda are doing here is revolutionary technologically, they're just bringing the same technology to a new market ($5K cheaper), which is still something.
I highly recommend that if you invest in THIS technology that you go all in and get the blind spot detection and rear cross traffic alerts. I have had a Subaru with Eyesight for over six months now, and I don't regret buying it and definitely like the blind spot/cross traffic alerts, they're legitimately useful day to day.
I will say one downside of this system is what I call "alert fatigue" particularly lane drift warnings, ice warnings, traffic pulled away warnings, etc. You can disable many of these, but it would be my only whine.
I have had automatic braking pre-warn me a handful of times, but not had it activate yet except when I pulled into the garage and the dangling tennis ball confused it and even then it only slowed me slightly. You get a yellow warning then red, then brake, and most of the warnings are legitimate I am just ahead of them.
Lane keep assist and distance based cruise control are like crack. It feels like you just get on the freeway, push a few buttons, and the car almost drives itself.
I also really enjoy my Subaru EyeSight. I've only run into one terrible experience, and that is when I went into a automated car wash. As soon as the first big scrubby thing approached, the EyeSight flipped out and slammed on the brake, which was jolting and scary. Kind of obvious looking back on it, of course, and now I remember to turn it off when I wash my car.
They now warn you in the manual expressly about that (specifically mention car washes too). Although I'm sure that was an addition, and I'm also sure many people won't read the manual (and Subaru's manuals aren't the nicest manuals around as an aside).
We have a 2016 Legacy with the eyesight system and I actually use it a lot in traffic.
It's really useful in "stop" and go traffic where you're barely moving. I just set the adaptive cruise control and it makes driving feel a lot more relaxed. I'm not sitting there alternating between the gas and brake every 15 seconds.
It's a great system, I can't imagine having another car without this kind of system in place.
My brother purchased a 2016 Subaru Outback with Eyesight and his comments to me have largely been the same. The only complaint I've heard is that if it's snowing or the roads have snow cover many of the features are unreliable. His praise when the roads are clear, however, has convinced me to look into it when my current lease expires at the end of this year.
The article really wants to compare with electric vehicles, for some reason. The subheading says "they are being snapped up faster than electrified vehicles." This is repeated in the article, which says the relevant options packages "are being snapped up at a far higher rate than electrified vehicles." Discussing the Q50's technology package and how many buyers opt for it, it says "That’s three times as many people who pay extra to buy a hybrid-electric version."
What's the deal? Is this just because Tesla happens to be the one with the best system at the moment? It doesn't make any sense to me, and really distracts from the article's main thrust.
The WSJ is a paper for stock market professionals, their traditional customers want over-simplified "predictions for one sector vs another sector" and you see it in a lot of their business reporting.
There's also the possibility that this story (including choice phrases) was handed to the author by someone in PR.
In the public mind EV = futuristic car. Self driving vehicle = futuristic car. The article author is either making the conflation to appeal to the audience or is similarly myopic.
I noticed the same thing, and my guess is that the WSJ is generally dismissive of climate change risk, and against government incentives to combat climate change, and wanted to frame this article as what the government should be doing instead of encouraging hybrid/electric vehicles.
> After a decade of spending much of its time and billions focused on boosting fuel-efficiency, Washington is increasing its focus on technology that could save lives.
This ignores that one of the justifications for fuel-efficiency regulation is the lives saved from reduced pollution.
It's just another one of the pile of vitriol against electric vehicles. Day after day, whenever someone mentions Tesla or Prius Plug-ins or the Nissan Leaf or whatever, it seems to kick in a auto-response of "what is the ROI of electric vehicles? You'll have to drive a million miles to get your investment back!" (as if buying a new non-exotic car is any kind of financial instrument). They also always talk about how many fossil fuels it takes to manufacture (or even charge) the electric vehicle, ignoring the fact that it not only takes the same amount to make the ICE car but it still has to run _directly_ on fossil fuels. When all of these arguments are refuted, the last argument is always about how cars sound better with gas engines and the rumble and other appeals to manliness or whatever rubbish.
There is a very large contingent that wants the status quo and either hires astroturfers outright to spread FUD, or influences opposition via media/ad buys. The rest opposed to electric are just old-school fuddy-duddies and the kids they've managed to influence. And, I've seen these arguments play out on everywhere from car specialist forums, YouTube comment sections on Teslas racing ICE cars, to right here on Hacker News when talking about electric cars.
I am not particularly opposed to their argument (crony capitalism is evil, after all) but their omission of decades of handouts to fossil-based businesses, as well as variety of present-day deductions crafted specifically for those industries makes the argument hypocritical or incomplete, to say the least.
Many of those "handouts" are legitimate tax breaks that all manufacturing companies receive, even Apple and Microsoft, only oil companies get capped at 6% on their section 199 [1] where everyone else gets 9%
And many of them are subtle but gigantic subsidies, like allowing them and their customers to poison the population without consequence, not even paying for the damages.
I'm not paid and I don't want the status quo, but until we shift electricity production away from coal, there is absolutely no reason to subsidize these cars.
The linked article never mentioned investment. It just highlights to utterly stupid one-size-fits-all legislation that cause automakers to produce cars they otherwise wouldn't produce. How about just raising the tax on gas if you want less gas produced???
In the US, coal power is under 40% of the total generation mix already, and dropping. My car emits less CO2 than a Prius, despite being way larger, heavier, and more powerful, and that will only get better as time goes on. It also emits zero local pollution, which is a pretty big deal.
How much longer do you want to wait? Electric is better now and just keeps improving. It takes a long time to change the makeup of the vehicle fleet (the average vehicle on US roads is over 11 years old) so if you wait until the grid is 100% squeaky clean before you start to make changes there, you'll have to wait much longer before you see benefits.
I do agree that it would be much better to just tax things to account for their externalities and let the market shake it out. But since that doesn't seem to happen much, we have to work with what we can get.
Well, yeah. You should tax the thing you're actually trying to reduce (carbon emissions), rather than subsidize a specific means of reducing one mode of producing that thing.
The former unleashes entrepreneurial ingenuity on all ways of economizing on the targeted resource; the latter plugs only one of many leaks, rewards the wrong thing, and is rightly called "crony capitalism".
The interest in self driving cars is much higher than in EVs for the general population. EVs even with two hundred mile range models coming out; I hope to get a lease on a Bolt by next Spring; will still have more limitations than people will accept. Even if the limitations aren't that burdensome it will take time for people to understand that. Still they will see 200 mile range and half hour or more charging versus the 300 to 600 mile range that petrol cars use.
Automation has so many more payoffs than electrification that I think its a better investment in the short term. The lives than can be saved, the freedom offered to those who cannot drive for varied reasons, and the fuel savings as well, all conspire to make automation the next big thing.
I agree with all that. But what's the link? I mean, interest in self driving cars is much greater than interest in plastic flowers, but it wouldn't make any sense to compare the two in an article.
The real problem with self-driving cars is the car to human handoff. Over the long term it's incredible unlikely that humans will be any good atall at remaining aware and 'up to speed' on the current situation in the event that the car needs to give control back to the driver due to road conditions, hardware failure, or sudden situation that it cannot contend with.
Normal desktop failure modes don't work well here.
"Hi, this is your car. I've discovered that the road is icy, and am currently in the middle of a 360. Would you like to (a) take over, (b) call an ambulance, or (c) retry?"
You have no faith in AI. The car will always be better at driving than the human. It would be much better at handling a 360 than a human ever could. Actually humans are very bad at handling 360s unless you are a pro.
>The car will always be better at driving than the human.
I don't think this is true yet. We can interpret the incoming data incredibly well in ways the machine vision can't yet match. This is absolutely vital for driving. The difference between an empty trash bag in the road and a toddler in the road is a matter of vital importance, and correctly identifying the difference matters a lot.
Humans area actually capable of being incredibly good at driving. The problem is we are also incredibly unreliable.
I wouldn't say the difference matters a lot. Instead humans and computers would just have different thresholds for when to stop and when to run over something. If a computer is less accurate at distinguishing them, then it'll simply err on the side of caution more often than a human. You would find there are more inconvenient stops, not that more more babies get killed.
I'm sure I've stopped for empty trash bags before because I wasn't sure if they contained something solid. That doesn't mean I failed as a driver. Everything went fine. Same if it's a computer but it might stop more often than a human.
This assumes that in every situation coming to a controlled stop is an option. That is almost certainly not the case. Many times the choice is between running (an object) over or making some kind of risky maneuver to avoid it.
Humans are very, very good at handling unknown complex situations.
In order for any AI to handle any situation, it must first be programmed to do so. If that situation was an unknown, you get undefined behavior. Undefined behavior in a car traveling down the road is not a good thing to place a bet on... especially your life.
This is why majority of sane driverless vehicle talk has concluded a human driver must always be in the driver seat, ready to take control. There's an impossible volume of failure modes a driverless car must contend with, everything from someone pointing an infrared laser at the distance sensors, to complete power failure, to a sudden rockslide, a bird darting in front of the car, lines missing from a windy and hilly road, etc.
In normal, daily mundane driving - driverless cars are expected to excel. Sitting in traffic is mind numbing, and mistakes happen when driver's minds wonder. In "active" driving situations, a human brain will almost always excel.
>This is why majority of sane driverless vehicle talk has concluded a human driver must always be in the driver seat, ready to take control.
And people who do research in this area have concluded that this doesn't work. People let their attention drift (even if they don't actively perform some other task) and, even if people are more or less paying attention, they haven't been driving and switching modes in that way is difficult.
We're definitely reaching the point with assistive driving systems where there probably needs to be more discussion about how much assistance makes sense if no one is willing to fully turn over control.
Humans are mostly terrible at handling unknown complex situations on the road.
For many human drivers on the roads right now, the response to a sudden weird situation is either going to be "brake heavily" or "swerve dangerously."
You don't have to get "undefined behavior" in response to an unknown situation. The AI can simply default to braking to a stop in the absence of any better programmed alternative. This will already put it on par with most drivers. Better, in fact, because the AI can begin braking instantly.
I personally find that the majority of driverless vehicle talk vastly overestimates the average human driver, and demands unreasonable perfection from the computers.
Not only can a computer begin braking more promptly, it can brake more accurately. Many human drivers hit the brakes 100% when they sense danger, then the full force of braking startles them (no training), then they let off again. This can increase stopping distances dramatically.
Next up is per-wheel braking. Not just stability control, but actual directed thrust. The computer can control each wheel independently. Humans just have one brake pedal. Four wheel steering might even make a comeback.
> For many human drivers on the roads right now, the response to a sudden weird situation is either going to be "brake heavily" or "swerve dangerously."
Or a combination of both, potentially causing the tires to slip and the car to leave the roadway!
I see this stated a lot, but I think it's exaggerated. My car's adaptive cruise control follows at about 2.5 seconds when on the maximum setting, which is where I almost always keep it, including in heavy traffic. People do occasionally jump in, but then the car just backs off a bit and starts following the new car at 2.5 seconds. It's not a constant churn, and it doesn't cause any significant delay. And this is in the DC area, with dense traffic and ridiculous drivers.
When people say that it doesn't work, I really think they're just saying that they can't stand having other people get ahead by breaking the rules, and they'd rather drive more dangerously than let that happen.
I tend to think the "A human driver must be ready to take over" line of thinking has a lot to do with liability as well. I just think it sets up an unfair situation where the human will often fail.
But you only have to program it once and then all the cars get the update. That's a lot better than hoping every individual human got properly trained and remembers what to do.
The list of situations a car has to react to is not infinite. You can program a reasonable response for a broader list of situations than you could for a human. And really, a human would fare very badly in your situations so a computer could hardly do worse.
> But you only have to program it once and then all the cars get the update
Not exactly. The situations we're discussing are more-or-less unique, having their own environment variables to contend with.
Not to mention, there are scenarios where a driverless car will never be able to operate (such as off roads, or back-country/non-mapped areas, or when road lines are covered by snow or just not there due to road construction, etc...).
Roads are anything but a constant. I cannot foresee any future where someone can get into a vehicle, in the back seat, tell the vehicle where to go, and then promptly take a nap.
Even with high-grade multi-million dollar vehicles such as military drones, they still require two humans to be at the controls at any time (even though in the air, collision opportunities are almost non-existent). The sensors and computers in a mere $20,000 vehicle aren't going to compete with this stuff.
A human will always have to pay attention to the road and be ready to take control. This aspect almost completely nullifies the desire for a driverless vehicle.
> In order for any AI to handle any situation, it must first be programmed to do so. If that situation was an unknown, you get undefined behavior.
No, that's not how AI nor computers work. It's quite trivial to have a default behavior like "brake", the notion that a new situation will confuse the computer is just you watching too many movies.
> This is why majority of sane driverless vehicle talk has concluded a human driver must always be in the driver seat, ready to take control.
Would add also something that requires an actual driver to take the wheel periodically to not get lazy and forget what it's like to drive (as apparently what has happened with some airline pilots?)
There's a related joke about engineers at an aviation software convention. This one might even be true.
At a talk to an audience of avionics programmers, the speaker at one point asks: "Tell me truthfully; how many of you would be comfortable flying in a plane where your team had done the software?"
Onlt a couple of people put their hands up. The speaker targets someone at the front. "You have that much confidence in your work?"
"I'd be fine," they reply. "If my team wrote the software, the plane would never even leave the ground."
That assumes that you have a car to human handoff, and that it would be useful. Honestly that's a pretty strong assumption because of the reasons that you mention and related issues.
The technological solution may be that there is no (or vanishingly small numbers) situation where emergency transfer is likely to improve outcomes. So you just don't do it. Non-emergency (i.e. I'm going to park now and hand off control to you) are a separate case.
This may plausibly be the best technical solution - how marketable it is, that's another story.
What you are describing is a very real issue, but I think you have to weigh it against just how bad humans are at driving in general (DUI, Texting, phone calls, distracted, tired).
I wonder if even at the really primitive state of SDC if they aren't better (in terms of safety) overall than human drivers.
Object avoidance and control in slippery conditions is a vastly easier problem than a 100% self driving car. So, if the default is a car that safely slows down to a stop while beeping loudly then hand-offs are a non issue.
Remember, the state of the art in human driving still kills 32,000+ people per year in the US and something like 1+ million worldwide. Beating that seems very doable.
EX: You show up at a concert and parking attendants are directing people to an open field. Car goes WTF, stops and hands off.
Yeah, that is why the failure response can't be to hand control back to a human. It needs to do other things to get out of the situation, and not rely on a human taking over.
Well, if you have an inference engine or something, and there's a situation that it hasn't encountered before, but there are sufficient rules to derive a solution, it may do pretty well. But if it's a completely novel situation, AIs of any flavor tend to fail pretty spectacularly.
Humans also can fail spectacularly. But they also can just try something, realize quickly that it isn't working, and switch to trying something else. An AI could do the same, of course. But humans are better (at the moment at least) at eyeballing the situation and realizing that the current approach isn't working. (Or so it seems to me.)
This is why Google canned the whole idea of requiring a human to be alert and ready to take over. They want to get rid of the steering wheel and pedals, because they found in their extensive road testing that their own employees just didn't pay as much attention as they wanted them to.
This is still facing regulatory hurdles, obviously.
This is anecdotal and subjective, and doesn't directly address the parent's claim (which was about “the long term”), but FWIW:
I've been driving a Tesla with “Autopilot” for five months now. Since then I've probably logged about 10,000 miles of use of “Autopilot” features – lots of two-hour trips between Eastern and Western MA; a MA<->NC road trip and some shorter road trips; and lots of 30-40 minute commutes.
Prior to getting the feature, I was concerned about handoff. I was concerned that it would be difficult to force myself to pay attention during those times that I wasn't directly responsible for the steering and acceleration.
Instead, I've found that it's easy to pay attention, but that I shift to a more strategic mode: scanning further ahead and around for hazards. It's actual easier to maintain situational awareness than when my attention is split between what's going on in the next 10s–1m, and more tactical tasks such as manually tuning my lane position, speed, and follow distance. Driving with automation feels (I told you this was subjective) safer than manual driving.
Just as in manual driving you keep your eye open for certain kinds of situations (passing a pedestrian or car on the shoulder, erratic driver ahead) and are poised to go into action, I'm especially primed to take the wheel when these risk factors add up (as well as other areas specific to this particular vehicle automation: its uncertainty about which way to steer around a small traffic island at the end of a turn; its failure to see stopped – as opposed to slow – cars just after a sharp turn or lane change; its penchant – mostly fixed now – for taking highway exits). And being ready to shift more cognitive resources over to driving is such a natural part of being a driver in any case, that I imagine this is a common experience.
When autopilot is turned off or I'm in a car that doesn't have it, it feels riskier not to have that second set of “eyes” on the road.
There still could be risk compensation: I could be driving faster because I feel safer, for example. I don't really trust anyone's attempts to figure out a priori whether I or the population of current or potential drivers of automated vehicles are subject to this; instead, I hope we see statistics soon.
And – I have a few decades practice driving non-automated vehicles, that I can pretty effortless fall back to. I've swerved around emerging accidents a few times, and handled drivers coming into my lane and other near misses; I'm decades from having to consciously think about where to put my foot to brake. I do wonder about the hand-off to people who are just now getting their drivers licenses and are going to have almost all their experience in automated vehicles is going to work.
We may be entering a (hopefully brief) pair of uncanny valleys of vehicle automation. (1) Where younger drivers don't have enough practice to deal with crises. (This has always been true of new drivers, but they'll no longer be learning at one year of experience per year; and the handoff requirement is new). And (2) the parent poster's concern, that drivers in general will zone out when the car is driving. My anecdotal evidence, above, is that this isn't yet a problem with the state of the art, but that's just a matter of whether this is the year that happens.
In the early 2000's, I was hanging out sometimes in western North Carolina, and there was this young woman who has in the habit of getting together with some friends and driving around the clock to get to Colorado and back on short trips, instead of flying. I'm wondering if this technology isn't going to be used for such purposes.
Actually, with buses today, it will take almost twice as long (14 hours) and require between 3-5 transfers [1]. In a car there are no transfers and it takes 8.5 hrs [2].
The big advantages of driving vs. flying are that the trip starts from your house, whenever you want it to start. There's no sharing space with passengers. Bathroom and food stops are at your convenience, no-one else's.
My initial thoughts about self-driving cars was that they would be less bad for the environment, thanks to more people being able to share one car - but the more likely scenario will be opposite: convoys of cars with one sleeping person in each.
As I mentioned in a previous thread some time ago, where does this notion of self-driving cars suddenly meaning shared cars come from? To some people these things seem to be joined at the hip.
Prompting a human that is also driving to make a 3 block detour is pretty hard to do safely. Prompting a passenger to see if they will take a 3 block detour to save cost is not so hard to do safely.
The self driving vehicles would also be reporting available ride shares to people looking to make trips.
I interpret "car sharing" as more about keeping utilization up. When you're at work, the car you used to get to work will be off doing something else during the day. When you want to go back home, it could be a completely different car.
You'd pay a subscription for a certain level of car service (time responsiveness, cargo capacity, comfort level). Maybe you'd get a discount for scheduling your commute a week in advance.
The # of cars on the road might be the same, but the amount of resources we would consume as a society would be lower. Or maybe it would be the same, but the flexible model would allow scenarios that weren't previously possible (e.g., your child could "drive" to school without you).
Once cars are autonomous, it makes it drastically simpler to carpool/rideshare or even pick up and drop off riders that may share the same or partial route as you.
Also if your car is able to drop you off, it can be a shared resource when you don't need it.
I want to reply to all three of you (turtlebits, maxerickson, teod) with basically the same points. I'll put it here because yours is currently the top-ranked reply, but I am going to address all three of you.
> Once cars are autonomous, it makes it drastically simpler to carpool/rideshare or even pick up and drop off riders that may share the same or partial route as you.
I don't see how it is drastically simpler. Yes the logistics would be easier. There still needs to be an infrastructure for identification of passengers and dispatch of vehicles that is completely independent of what is driving the vehicles themselves.
> Also if your car is able to drop you off, it can be a shared resource when you don't need it.
Assuming I want my car shuttling other people around, particularly when I'm not in it.
> Prompting a human that is also driving to make a 3 block detour is pretty hard to do safely. Prompting a passenger to see if they will take a 3 block detour to save cost is not so hard to do safely.
True, but the bigger issue is, will the current occupant of the vehicle ever say yes?
> The self driving vehicles would also be reporting available ride shares to people looking to make trips.
Somebody could already do this with human-driven cars.
> To me, it seems like the incentives to own a car may not be so great once self-driving taxis are available, especially in the city. No need to worry about parking, insurance, maintainence, or gas.
How is this different than human-driven cars? Do you really think cabbie salary is enough of a driver in the pricing of taxis now that shared car services will suddenly become attractive once they are eliminated?
I pretty much agree with y'all on the technical barriers being overcome, but I think y'all and others seriously underestimate the sociological barriers that would need to be overcome to make this a reality on a large scale. Maybe I don't want to pick up random strangers on my way to work. Maybe I don't want to add five minutes to my commute to pick someone else up. Maybe I don't want other people in my car.
The reason these shared car services don't really exist today has little to do with who or what is actually driving the vehicles and quite a bit to do with the nature of our society. You can't change that by just building a fleet of autonomous, on-demand cars.
> I don't see how it is drastically simpler. Yes the logistics would be easier. There still needs to be an infrastructure for identification of passengers and dispatch of vehicles that is completely independent of what is driving the vehicles themselves.
I don't think anyone expects it to be drastically simpler overnight, but when you look at the value proposition there is every reason to expect that someone will figure it out, package it, and sell it as a service.
Several major automakers are working on this, as well as Uber, Lyft, and a number of startups.
I do think it will be an iterative process where the first few attempts may backfire, and I do think that there will still be a large number of people who want a non-shared car; On the other hand there may also be a very large number of people who would love to have a "free" car that other people get to use when they are not using it.
Sorry, I think I may have misunderstood your original question. When you mentioned shared cars, I was imagining a fleet of autonomous, on-demand cars but not one that would pick up other people during your transit. Essentially I'm imagining Uber without the driver. Depending on how it's implemented, self-driving cars won't need sleep and they won't care where they go.
Maybe you're right and currently self-driving cars doesn't offer much more than the current human-driven cars; however, I don't think sociological barriers come into play as much in this scenario (beside the general concept of a self-driving car).
Cabbie pay is huge compared to the depreciation of the car. Medallion systems create deliberate taxi shortages to prop drivers' daily income up somewhere near a livable level. It might be viable to have a self-driving van on every block if they could idle with nearly zero cost.
from another post above a trip from SF to Vegas is 569 miles and 8 hours. The standard reimbursement (gas + wear and tear) is $0.50/mile. So that is $260 in gas + wear and tear. a quick search finds taxi drivers are $13/hr - $20/hr. that is an additional $104 - $160.
The slower your drive, the higher a percentage the taxi driver will account for.
>self-driving cars could lower costs by 30% - 60% just from labor savings.
Which is almost certainly not enough to drive the radical changes in car ownership behavior that many people seem to assume will happen. To be sure with a human driver you probably need to also account for utilization rates and returning to their starting point.
I've never been quite able to square the rates of things like Zipcar with taxi/Uber. Given cabbie salaries, it seems they should be closer--at least in places that don't have medallion systems that really skew things.
To me, it seems like the incentives to own a car may not be so great once self-driving taxis are available, especially in the city. No need to worry about parking, insurance, maintainence, or gas.
Another possibility is that if self-driving taxis take over, cars might get a lot smaller. Most of the space in your car is wasted most of the time and is only there for the occasional time that you need it.
Although, I'm not arguing whether self-driving cars will be a net good or bad from an environmental standpoint, I'm not sure.
Would you be more willing to live 2 hours outside of SF if you could hop in a self-driving car 3x/week and watch a movie/surf the net during a trip into the office? That's the most exciting part of it for me.
On one hand I get this; folks have been pulling crazy long commutes in the NYC area for decades. But the very idea of four hours per day of commuting being "exciting" in the age of completely practical remote work is baffling to me.
Sure, but many [crazy] people are already doing 4 hour commutes, so they can keep they job they want, in the office, and have an affordable house. If they can do that and make commute time entertainment time, that's actually a significant improvement in their life.
My commute is currently the longest it's been in 10 years at 7 minutes (as my family has grown, I've had to move further from downtown). Having commuted longer distances, both by bus and by car, I still say that nothing beats a short commute.
With self-driving cars it would probably be possible to design the seats so that with only one passenger it would be possible to lie down fully. This could make overnight drives in personal cars quite pleasant relative to buses.
To me this is the most exciting part of fully autonomous cars. Some of the new car designs we are going to see will drastically change automobiles forever when you no longer have the concept of a "driver" or steering wheel. Where should input/control surfaces go? How should the car be built? How can it be built safer?
Imagine being able to 'road trip' by going to sleep in your car at home, waking up in a different city, spending the day there, and being home against the next morning.
> With self-driving cars it would probably be possible to design the seats so that with only one passenger it would be possible to lie down fully.
Is there a way to properly and safely restrain someone who is lying down when a collision occurs? I can only imagine the number of broken necks or spinal compression injuries resulting from hitting another vehicle at 60mph+.
If you're lying down sideways and as you buckle in airbags inflate around you before you take off, in the event of an accident other than a rude awakening you should be in better shape than today. Accidents should be far less brutal anyway once we ban human driven vehicles.
Throw in 90-100 mph driving speed and we have a great way to travel. There"s no reason we can't have higher speeds on highways with only self-driving vehicles?
Emissions would be a problem (until Tesla takes over the world.) The energy required to sustain 90-100mph is mostly a factor of aerodynamics.
If the distance between the cars could be kept sufficiently short like perhaps a few inches only then maybe.
High-speed trains have flush coupling between the streamlines cars to deal with that. The incremental resistance from length is almost negligible and so they can maintain 200 mph speeds.
I wish north america would embrace high-speed rail like in Europe.
Actually a couple feet of separation is sufficient to get significant drafting aerodynamics going on. See, for example, the distance that bikers can draft behind a truck, or cyclists slipstreaming in the Tour de France, or drivers in Nascar or birds flying in a V.
I don't think there's a reason we couldn't, but as far as shouldn't, efficiency losses from air drag increase proportional to speed. Same reason we don't have any 100mph submarines.
Conversations about self-driving have focused on zero defect rate in-city self-driving vehicles, but a lot of these technologies are reaching Pareto-efficient levels of usefulness. I don't need my car to drive the first and last miles; I'd be perfectly happy if it just drove on the highway.
And why do humans drive long haul trucks for anything other than the first and last miles? Trucks should drive themselves between depots at the edges of metros and then humans can drive them into the city. https://www.mercedes-benz.com/en/mercedes-benz/innovation/th... "Self steering"... How long before that moniker is replaced by "Self driving".
It's going to shock our economy once industries begin constraining roles to the level at which robots can be "good enough". After figuring out how to manage them, we'll start to see robots deployed in force.
As parents, my wife and I are talking about this a lot as we think about how to raise our kids (and we are emphatically hands-off parents)...
Even fully automated, trucks have much longer response times in dangerous situations. The fatal accidents would be inevitable, and the legal response vicious. Never mind that automated trucks would probably be less likely overall to be in an accident.
When will highways be upgraded/updated for self driving? I don't mean V2I (vehicle to infrastructure) capabilities, but properly painted and maintained lane lines, reflectors, and signage. The infrastructure is just not there. You can't simply rely on a car's sensing abilities for self driving.
I think the current paint is OK, it just needs to be serviced. It might be a good idea to include the V2I sensors in the paint itself due to its thickness (I've measured the lane lines locally and they are about 10mm thick). But the basic maintenance is key. This is a bureaucratic problem due to how infrastructure maintenance is used as a political tool (politicians in power usually advertise how public works generate n jobs). The problem here is not the actual tech, but the politics. The tech will eventually get there. The politics? Good luck with that...
Am I missing something? Why is this top news? There are several cars out there with adaptive cruise control. In fact, I think most mainstream cars offer it as an option now. It's pretty impressive but calling it "self-driving" is hyperbole.
Subaru's eyesight is technically even more impressive considering it does image processing to detect vehicles, whereas most of these systems are based on radar. Although I don't see the point because radar is more reliable IMO. Unless you use some features which only camera can provide (stop at traffic lights?), which Subaru doesn't yet do. From whatever research I did before buying a car, Volvo's system (uses both radar and camera) seems to be one of the best overall, along with Mercedes, and Subaru being a close third.
I'm curious about whether this is going to be a net win or lose for safety.
If users treat these automated cruise systems as a "magic self-driving device" when it can potentially make mistakes or hand back control when it's confused, drivers are going to die.
If people "really want to look at their cell phones", and they take this as the tool that lets them pay no attention to the road, it better be up to that task.
The title is a bit misleading. I have a 2016 Civic Touring that has the Sensing Package. It's effectively a sensor package (cameras in the top-middle of the windshield and below passenger-side headlight) with integration with the steering and powertrain systems and very basic logic.
When you turn it on the car basically tries to stay in your lane by looking at the lines on the road. It actually tells you when it can and cannot see the lines. When it detects you going outside of the lane (without using your signal) it takes control of steering and corrects for you. You can also set it to stay within some distance of the car in front of you and it will control your speed. Supposedly, it will also auto-brake if you are in danger of collision but I haven't had a chance to test that yet (and hope not to have to).
The whole thing is more like a driver assistance system and if you take your hands off the wheel for more than 15 seconds or so a bunch of alerts start going off and the system disengages. After using it for a few months I think this is probably a good idea. There are quite a few places in the SF Bay area that have worn out and faded lines on the road and once the system loses sight of the lane markers it just stops working. Not a great moment to have your hands off the wheel or your eyes closed. ;-)
For the price its incredible that Honda offers something like this. Suburu offers something similar but the next best thing is buying a Tesla for much more. I treat it sort of as insurance on long trips. If I start dozing off or am distracted the system keeps me in check but it is not reliable enough to be truly autonomous. So yeah, it can sort of drive autonomously.
As a preview of the future it gives me hope and it's possible this may be the last car I actually buy (when cars drive themselves it could very well become a service industry).
This is a interesting proposition for Honda, but really not that new. Even the price is not /that/ new.
In fact, this is technology that has been sold at around the same price point in the industrial transportation sector -- think of logistics and lorries -- for quite a while. This is where the majority of the innovation is taking place.
For example, just this week it has been revealed that the UK Govt will likely to announcing tomorrow funding for driverless truck convoys in the North of England. What's the price differential between these intelligent trucks and regular trucks? $0.
In fact, so much innovation is taking place in the industrial sector that just last week Toyota announced that it has hired the FULL workforce of Jaybridge Robotics, a firm that specializes in autonomous industrial vehicles, mainly in the agricultural sector.
If you want to understand the tailwinds in the sector, follow the b2b and industrial segment of the market. Technologies and trends are already starting to filter down.
--
I also want to plug my email newsletter Driverless Weekly (http://driverlessweekly.com) while I'm at it. It's a once-weekly summary of the top news stories in the autonomous vehicles sector.
I've owned three cars with automatic cruise control for the past decade. This isn't exactly new technology, perhaps only at this price point.
The first car I had with this, an '06 Infiniti, was only able to slow to a crawl, not a full stop. So while it was useful on the highway it was useless in stop and start.
The second car, a '11 BMW, added "Stop & Go" to the formula. Great? Not quite. What would happen is that the car would come to a full stop, and then a timer would run, and if the car didn't start moving again within 10 seconds the cruise control would shut off, and to resume you'd have to push the pedal. This was especially maddening when stop & start traffic is inconsistent and the stops last 10.5 seconds. Basically the idea of being able to set & forget was completely undermined by this and driving with the feature on was more stressful than driving with it off and just doing everything manually. A complete bust. I don't know why it does this but I can see it being some combination of the product team needing to ship the feature in the state it was in, and legal requirements.
The problem with both of these implementations is that they promised to alleviate some of the issues with past "auto-drive" features (and you should consider Cruise Control to be the very first auto drive feature), but introduce their own. If you want the user experience of set & forget, you need very predictable conditions if you want any of these mass market systems to work for you, and unfortunately that's just not the way the roads are.
I think I have the feature in my latest car too, but I've given up and decided to enjoy manually driving, and just wait for fully autonomous vehicles.
How do we as drivers keep up with the rising levels of automation? It's challenging but doable for owners, but imagine jumping from a pain old 1990s car into a flashy new rental with all the bells and whistles... With a bit of exaggeration, one might make a case for individual type rating, like airliner pilots need to have.
Before we reach fully autonomous driving, we might see an age of widespread "car illiteracy", with more and more people who have a driver's license, but who have completely lost touch with the state of the art in car UI concepts. With not enough time on automatic transmission, people here in Europe occasionally even struggle with something as simple as park/neutral/reverse/drive (don't ask my where I got that)
Sounds like these cars are far from self driving. Just some added safety features that resemble self driving but, are dangerous to use without a foot near the break and hands on the wheel.
I think this sentence pretty much says it all, "For instance, some owners have posted videos of hands- and foot-free driving on YouTube and the car inevitably makes a mistake.".
It sounds like these features are going to end up being abused and probably causing serious fatalities. As we have seen people want to txt and even watch movies while driving. These new features will allow wreck less drivers to pay less attention to driving and more attention to how many Instagram followers they have.
I'll be sure to pay more attention to people driving Honda civics when I'm on the highway.
This looks like the same functionality the Tesla Model S has. One of the big draws for the new Model S cars. So much so that people are getting rid of their older Teslas because they did not have the autopilot.
I thought it's only a matter of time before autopilot becomes common place but it looks like its happening sooner than I thought. The good thing is this has nothing to do with the car being electric or not. But the main thing is, if Tesla thought its going to be a differentiator in terms of calling their cars "luxury" its going to be a problem for them. Given that the internals of the Model S itself are not particularly luxorious they need to think about it.
There was a startup that for $10,000 it was a third party option package that was self driving. I wonder how it is doing with all these car companies that create their own options for self-driving.
I think you might be referring to Mobileye. They're actually doing really well, because their technology is what drives most of these first-party systems.
I have yet to drive a "self-driving" or even "self-monitoring" car and the thought of doing so terrifies me. I know technology is good and it can do great things, but helping me drive is something I don't enjoy the idea of. What happens if, however unlikely, someone were to hack my car? They could potentially crash my car and leave without a trace. I think we really need to take a step back and ask if the benefits outweigh the potential costs/risks.
> The Obama administration has proposed spending $4 billion to accelerate autonomous-car technology during the next decade.
Hmm, what are they spending it on? There's a lot of money being spent on developing this technology already by multiple private companies. I assume it's for something else...
IMO government research money should go into stuff that private industry is unwilling to fund, like pure math, theoretical physics/CS, and other things that have very long-term yield/results timelines.
It might take a while before self driving is widely available, but self stopping is here today. It's now standard equipment in all new Mercedes -- including cars selling for about $30k. I have an entry-level Mercedes and it includes blind spot radar, lane tracking, and collision avoidance that will stop your car automatically if you're distracted or incapacitated. These features are available widely from most automakers now.
How does an autonomous car reduce the need for infrastructure investment? A bunch of commuters in self-driving cars will clog the road just as much. Less need for parking, maybe; less need for roads, no.
Maybe you could squeeze a bit more capacity out of freeways by running cars at high speed at short following distances. But a lot of congestion occurs at intersections and self-driving does little to solve that.
Not "maybe" less need for parking. Definitely, certainly less need for parking. And remember that a lot of "road" surface is actually reserved for parking and not actually available to drive on. Same thing for having to have massive parking lots EVERYWHERE that people with cars might want to go; that can potentially go away completely.
And if you need much much less parking, guess what you also need less of? Parking meters. And parking tickets. Imagine a society without that pernicious bullshit everywhere.
Or, alternatively, where hesitating to make enormous infrastructure investments for questionable return ended up to be the sensible approach after all.
the self driving car is a concept I have a hard time appreciating. It sounds cool, and offers enormous technological and legislative obstacles to overcome, yet I can't figure out the fundamental problem they will solve - compared to the attention they get. For any use case I can imagine (minus the cool factor) we either have more efficient solutions already, or there are better alternatives to investigate. Plus I figure, most people still like to drive.
I'm impressed, I'm much more relaxed than before - and that means I'm actually able to just look ahead and think about the road, not about what I'm doing.
My friend has a Tesla, and he said it's the same as cruise-control. Basically, it's an assistive feature, but you are still 100% totally responsible for anything your car does while you're at the helm. I don't know if you count that as driving itself or not, since you still have to do most of the things you do as a driver (i.e., pay attention, react to situations.)
From the article: "as long as lane markings remain visible and another vehicle is in front of the car." That's more like platooning, which was demoed two decades ago in Demo 97 [1], than self-driving. Are there more details about how this works, especially about autopilot disconnect and user takeover? Tesla's system is known to have trouble with offramps.[2] (Tesla customers are very forgiving. "It's a beta", one says in their forum.)
Honda's follow-the-leader system avoids some of Tesla's problems. Radar systems for not rear-ending the car ahead are already pretty good, and many are already on the road. Lane following by lane markings isn't that reliable, but restricting it to following the car ahead handles traffic jams nicely while locking out most of the hard cases.
The deployment of self-driving systems which are much dumber than Google's is worrisome. I've written before of the "deadly valley" of automated driving. This is another "deadly valley" system. The deadly valley begins where the driver can take their hands off the wheel and tune out. It ends where the automated system can correctly handle more situations than a human driver.
Google is trying hard to get to the far side of the deadly valley. That's good. Look at the problems they're having. Their only known fender-bender in autonomous mode was when the vehicle was trying to deal with a drain blocked with sandbags and very slowly maneuvered around it, to be hit by a bus, while in a wide lane at a right turn, because the AI misjudged the likely behavior of the bus driver. Google's dealing with the hard edge cases. Cruise, on the other hand, ran into a parked car in San Francisco when the autopilot lost lane tracking, veered left, overcorrected right, and the driver took back control too late. That's a more basic failure.
It also shows the problem with semi-autonomous systems. Expecting the driver to compensate for failures of the automation will not work.
Volvo Car Group President and CEO Håkan Samuelsson says that the company will accept full liability whenever one of its cars is in autonomous mode.[3] He has it right. This needs to be a requirement before the "move fast and break things" crowd gets on the road.
I agreed that there needs to be a system of accountability for both the driver and manufacturer as well as proper marketing of said features. This article and the misleading title could very well cause the same damage as directly marketing the technology as a "magic all-in-one solution." Much as the companies and supporters like to see automated driving pushed through development and testing quickly, the public and private corporations needs to realize that "breaking things" can involve fatal accidents.
Yes. Here's a really bad example of feature misunderstanding, a Volvo owner demoing the self-braking, where it plows into some pedestrians.[1] "Volvo spokesman Johan Larsson told Fusion the people in the video did not pay for the car’s “pedestrian detection functionality."
The levels of self-driving may need to be standardized, so that when a driver gets into a strange car, such as a rental, they know what to expect.
Calling it "self-driving" is kind of a misnomer and I think the article kind of blows it out of proportion.
It will track the car in front of you and keep a safe following distance, keeping either the maximum cruise control speed you've set, or whatever speed the vehicle ahead of you is driving, whichever is slower. It will accelerate or brake accordingly. It will also attempt to stay in the lane by using the onboard cameras for tracking the lane markings.
The lane keeping assist is not nearly as autonomous as the article makes it out to be. It does not like to work on sharper curves on the freeway, for one -- the system will disengage and tell you to steer manually. It still wants you to keep your hands on the wheel. It must be looking for very very subtle movements on the wheel, because the system will yell at you if you take your hands off the wheel for longer than 10-15 seconds.
All in all it's a pretty cool feature for longer road trips (keeping in your lane can just get kind of tiring, even with cruise control) but it's not the sort of autonomous driving that the article here paints it out to be.