Those who thought self-driving vehicles with no backup human driver was decades away are about to find out the future starts in 2018 as Waymo Shifts to Robot-Only Chauffeurs with commercial service starting in 2018.
Waymo technicians are already hailing its Chrysler Pacifica Hybrid minivans in and around Phoenix via a mobile app and leaving it to the artificial intelligence operating the vehicles to figure out how to get to requested destinations. Within a few months, Waymo vans loaded with laser LiDAR, radar, cameras, computers, AI and no human safety drivers will pick up Arizonans registered in its “Early Riders” program.
In May it announced plans for an autonomous vehicle pilot program with ride-hailing service Lyft, and in June inked a service deal with rental car giant Avis to help maintain its Phoenix area test fleet. That month Waymo also disclosed that it’s testing self-driving technology on large commercial trucks.
Waymo minivans operate at SAE Level 4 autonomous capability which means they can drive without a human at the wheel in most circumstances. They are trying to reach Level 5 capability, in which vehicles can drive anywhere a human can under all conditions.
“Someday these cars will drive themselves. That day is today,” says a mom who watches young kids get into a Waymo van in a test service now operating in Phoenix.
I seem to recall readers telling me that parents would never let their kids get in such a vehicle.
Inquiring minds may wish to download Waymo’s 43 page PDF on who the technology work.s
Self-Driving Vehicle Sensors and Cybersecurity
Waymo security practices are built on the foundation of Google’s Security processes and are informed by publications like the NHTSA Cybersecurity Guidance and the Automotive Information Sharing and Analysis Center’s (Auto-ISAC) Automotive Cybersecurity Best Practices. Waymo has also joined the Auto-ISAC, an industry-operated initiative created to enhance cybersecurity awareness and collaboration across the global automotive industry.
Real World Experience
Over the last eight years, Waymo has tested our vehicles in four U.S. states and self-driven in more than 20 cities—from sunny Phoenix, AZ to rainy Kirkland, WA—accumulating more than 3.5 million autonomous miles in the process. Driving in Phoenix has tested sensors and software in desert conditions, including extreme temperatures and dust in the air. Waimo says it knows how to navigate around new types of vehicles, like watering trucks that move 3 mph on 45 mph roads while spraying plants in road medians. Austin provided horizontal traffic signals for the first time, and Kirkland gave Waymo wet weather practice.
Simulated Image
The simulated imagery shown demonstrates how Waymo software assigns predictions to each object surrounding their own vehicle — other vehicles, cyclists, pedestrians, etc.
Self-Driving Future Arrives in 2018
The self-driving future that allegedly was still decades away arrives in 2018, in actual paid-for commercial service.
I still expect naysayers to tell me about snow and ice and other situations that will make self-driving impossible. However, the worse the conditions, the safer self-driving will be.
As I have pointed out before, substrate mapping can see through six inches of snow. All it takes is mapping the roads. Highway transmitters might be coming as well. And of course, this is only 2018.
Ford claims it will offer self-driving vehicles by 2021. 2021 is still over three years away.
Widespread adoption by the average person may be slow, but mass adoption by trucks will happen within a year the department of transportation allows self-driving trucks on the highways.
I suggest that by 2022 a huge percentage if not an outright the majority of trucks on the highways will be self-driving.
Uber and Lyft plan fleets of self-driving cars. Those who drive for a living, seriously need to think about their next job.
Mike “Mish” Shedlock
@CautiousObserver
Like I said – I lean towards thinking it’s a bad idea. I don’t think *self-driving* cars are a bad idea however. But government will never let them exist without central control…
@WildBull
I agree.
@DBG8489 Cars don’t have to be networked, but they will be. I know it’s coming, but it is a huge safety/security problem. It is a cyberterrorists dream.
@DBG8489
I do not question your point about network redundancy, very high hardware reliability and ability of vehicles to operate on a standalone basis for periods of time.
It is central control and central system ownership that I really dislike. Notwithstanding that, inherently reliable design still does not address vulnerabilities such as extended system failures, malware, design errors, etc. Recently my area had an extended power outage. After the first 24 hours, nearly all services failed including internet, cell phone, and water because all used electricity and all had insufficient backups for an extended outage. As a separate example, I recently purchased some accounting software which demands internet access to fully function. Denying it internet access actually breaks part of the software and causes it to periodically throw errors.
If self-driving cars make a central connection mandatory I will count it as a strike against the design and that will make it less likely I will purchase one.
@Wildbull
Cars don’t need power from the grid to run – or share data among each other. And they can store enough mapping data locally to be without for long enough to either clear the roads, or get themselves to safety if necessary.
The biggest problem is vulnerability to hacking. To prevent this, they need to come up with a fairly foolproof method of data sharing and remote access. Something so secure it’s miles beyond what we have now. It can be done – I’m sure of it. However, the biggest issue I see with that is the unwillingness of government to allow the autonomy necessary for the networks to work without their interference. They will demand “back doors” and ways to shut down cars when they want and what-not. And these things can – and will – be exploited by external means.
@CautiousObserver
As a person who – before management life – built his career engineering ultra-high-speed, hyper-redundant networks for data center use – I can tell you that today’s networks are miles beyond what we had just ten years ago. Five nines of up-time (which allows for about 5.5 minutes *per year* of downtime) is no longer something we strive for – it’s expected. Using redundant fiber paths, hardware, and redundant UPS/generation systems any one of the data centers my employer owns can maintain itself forever – providing we can continue getting diesel. However, none of that matters to a computer-driven vehicle that doesn’t *need* that network to run safely. Maps don’t *have* to be updated in real time from a remote server. Maps will be maintained locally – and the “network” that the vehicles will use to communicate exists between *them* – without a server or router or anything in between. They will be “ad hoc” networks generated on-the-fly as vehicles become close enough in proximity to each other.
@Grumblenose
Ask yourself what makes these decisions so complex. What is it about a situation – say ten thousand vehicles navigating through a big city on freeways and surface streets – that makes it so difficult to predict? Human nature. Humans interpret situations differently for a variety of reasons – most of which have nothing to do with logic or reason – and they react to those situations in different ways. What if you and the drivers in the vehicles next to you could talk to each other in real time and tell each other what you were doing? Or what to do? Or what not to do? Now imagine those same ten thousand vehicles are driven by computers, interconnected with each other and able to communicate all the variables, the decisions begin to boil down to simple logic – nothing more. And if all the computers will make the same decision in the same situations, then there is no more “guessing” what the decisions will be.
It’s funny – you’re here waving your hands at it and claiming that “we won’t see it in our lifetimes” and meanwhile, the engineers are already doing it.
For what it’s worth, I don’t have an opinion on the intelligence of doing this. I guess if I had to lean one way or the other, I’d say it’s probably not the best idea. However, I don’t really care. I drive an early eighties model vehicle with a cable clutch and a carburetor. I do all of the work on it myself including machining new parts if necessary. I will continue to drive it no matter what. I’m just telling you that this is coming and it’s coming faster than you think – and all the hand waving and ignoring it isn’t going to make it go away. There are too many people who *do* want it and will do anything to make it happen.
Supposed “hackers” have already taken control of cars with WiFi. These were people with access to the security hole left for law enforcement or a password used by the manufacturer during testing. Many things today are not “hacked” but instead are subject to espionage, where passwords, obscure security flaws and other critical information are stolen and released by insiders. It is very difficult to keep this kind of information under control because so many people end up with access to it. A system run cooperatively among auto manufacturers, their vendors, government regulators, highway departments, their vendors, law enforcement and a slew of other players will be almost impossible to keep secure.
@WildBull
I also am not enthused by the idea of everything being networked together. Apparently the people pushing hard for this and claiming it is a positive feature have not read enough spy novels or watched enough dystopian sci-fi.
A single interconnected network potentially means: (1) one common point of control, (2) one common point of failure, (3) one common point of ownership. The negative scenarios based on these suppositions are too numerous to mention. Many cautionary tales have found their way to the big screen. If a pledge of fealty will be required in exchange for using a self-driving car, then I am going to keep a manually driven car as back-up. I might even drive my manual car most of the time.
And then what about hackers? These cars are supposed, in the future to communicate with each other and with the highway infrastructure. They will be on the F’ing internet!!! What would things look like if every vehicle on the road made a sharp left turn at 8:30AM EST or PST?? The NSA and CIA are worried about the power grid. I can do without electricity for a bit. A massive power outage would result in deaths, but nothing like 40,000,000 cars in the ditch.
@Grumblenose
Yes, the current generation of computer-driven vehicles like the Waymo above *can* recognize objects – including other vehicles, pedestrians, cyclists, and even animals. And yes – they all share this info in real time. Not only that, it can even make predictions about what *can* happen given a developing situation and then make decisions based on those predictions.
And if all the other vehicles are also computer-driven, then they are all talking and making decisions *together* about situations on the road.
What do you think the human brain is? It’s a computer – constantly processing and analyzing situations and making decisions based on predictions that are based on past experiences.
Imagine if your decisions weren’t just guided by *your* experiences and predictions, but on the experiences and predictions of twenty other drivers within a fifty yard radius accompanied by stored data from previous drivers in the area – in real time.
This is the system that’s being built.
You can keep waving your hands at it, but it’s not going away.
@Grumblenose
Yes – they are “better than poor human drivers” because they *never* forget. And they *constantly* share information.
Vehicle A crosses a bridge and hits a pothole that developed overnight. Vehicle A notes the exact location – and then within seconds, every other vehicle in the fleet within a 50 mile radius knows it as well – and they will all avoid it in the future. When it is repaired, that repair will be noted – by *all* vehicles in the fleet. And if another vehicle outside the radius enters the area, it will pick up the data and avoid it as well.
In a case described by @CautiousObserver – the logic would be developed the first time the incident occurred and in the future – it would be employed not just by the vehicle affected in that *one* case, but in *all* cases which means that same incident would never occur again.
When I was talking to my son – who is deathly afraid of driving – I tried to explain it like this: Road and highways – with all the vehicles on them – is a system like an engine. And as long as all the drivers try to follow the rules of that system, it is relatively safe. And in a perfect world, every person in that system would follow the exact same set of rules and use the exact same logic in every situation – and if they did that, there would be very few “human-caused” accidents – most would be chalked up to equipment failure.
The system these developers are building *is* that system.
The big question everyone points to is the moral one: If you could avoid hitting a busload of nuns, but doing so meant you killed a mother and her child on the side of the road, which do you choose? Some people say this is a problem that computers can’t solve. However, I’d wager that most of them couldn’t solve it in the moment either. This is because in any instance such as this, all humans will have one simple thing on their mind: Their own survival. It’s instinct and there’s nothing that can be done about it. In a moment where we have to make a split second risk assessment, we will nearly always make the decision that involves less risk for ourselves – regardless of the consequences to others.
The key to solving this is the computer’s ability to *see* the situation begin to develop and make a decision before it has to be life or death. In the case where all vehicles are computer-operated on a shared network, they are all constantly sharing info and following the exact same logic and rules which means the likelihood of an incident where a single vehicle has to make a split-second life-or-death morally-relevant decision is extremely low.
link to techcrunch.com
@Grumblenose:
I have seen humans make this kind of mistake before. Once when I turned left directly behind someone who was turning right onto the same entrance ramp in front of me, the driver in front of me slammed on her brakes because she was convinced I would hit her instead of pulling in behind her. She “stopped to avoid an accident.” What she actually did was stop directly in front of me so that I had to brake hard to avoid hitting her. She made a worse mistake than this Las Vegas robot.
Unlike the human driver I had my experience with, the self-driving car in this incident will have the software for this particular issue revised so that this does not happen again. Presto, problem solved – permanently.
2018 is not far away. We will not have to wait long to see how it turns out.
I’d like to see the effect of a simple bird’s mess on the radar. Vehicle stops and you are blocked and b’ocking forever ?
If we are provided proven solutions to standard unexpected wrongdoings, I will buy.
So far, I read that it is technically happening which does not surprise me like the Moore’s law.
If an obstacle on the road, how does it behave ?
I can vouch for our host that this is sooner than even his optimistic timeline. In August, 2016, Wired Magazine was predicting “a few” “Level 3” self-driving cars “within about five years.” Instead we get “Level 4” in eighteen months. Whoa.
link to wired.com
“We’re always one year away from self driving vehicles taking off. So right now, it will happen in 2018. Next year, it will be 2019. And so on.”
LOL I said 2022-2024
No one said 2018 – but here it is
Ah there they are
@grumblenose, i’m nothing short of amazed at the rate of progress. in contrast, you must be in complete denial.
is that the definition of denial?
…wow
“We’re always one year away from self driving vehicles taking off. So right now, it will happen in 2018. Next year, it will be 2019. And so on.”
@KidHorn – Mish has been consistent with his predictions. His only revisions have been in conjunction with stories which suggest he was too conservative, in which he says, “my predictions were probably too conservative.” So not sure who you’re talking about.
I would say we’re three years away from massive disruption to long-haul trucking, and five years from significant adoption for the personal auto side. I believe that by 2022 there will be ride hailing services available which many people will use for their daily commute, and will continue to maintain a regular drive car at home. Regular drive cars will probably not exist on interstates and major roadways passed the late 2020’s.
We’re always one year away from self driving vehicles taking off. So right now, it will happen in 2018. Next year, it will be 2019. And so on.
But Mish, what about all of the merry pranksters who will stand in front of the vehicles to trick them, steal from the semi-trucks, or hack the systems to crash them all into each other all day and all night? It will be mass hysteria!
Oh no, I just realized these pranksters and thieves will be joined by the millions of laid off out of work truckers. We all know that once an industry dies the unemployed will remain so permanently, so these millions will be in the streets blocking lanes and protesting endlessly.
But even if we get through all of that, this is just one big plot by the New World Order. They’re just waiting for us to get rid of our regular cars, and then they’ll drive us all off a cliff to reduce the population and control the survivor’s every move!
Has Waymo explained what ethics the software is following if/when faced with a choice like colliding with a bus queue or a baby carriage? (Asking for my lawyer.)
Surprised at all the positive comments so far. There’s no need to defend this technology as it is inevitable and manifest. Figured it would only be Luddites and Chicken Littles commenting but ok.
I wondered what these vans with the cherry on top were for. Every time I see one, I slow down thinking it’s a cop until I see the logo on the side.
Cue the wailing and gnashing of teeth from those being “put out of work” by this technological advancement.
Anyone with half a brain could look at the clues and realize where this is going. And they still have a couple of years at least to find a different field.
The question is: Will they?
link to cnbc.com Therefore, expect a lot more mass shootings.
The number of Americans not in the labor force is about to abruptly jump from 95 million to 100 million.