Coming to a road near you? (Automated Cars)

[Jlabute]

As the article says, the sensors failed to distinguish, they were fooled... and the link makes Tesla partly responsible since their software lacked safeguards. There are a lot of factors that can cause sensors to fail or be fooled. Perhaps a whole new class of sensors are required for future reliable self-piloting. The driver relied on a vehicle feature that can’t be trusted. Right now auto-makers have a new technology they can sell for big bux even though it is immature and puts people at risk. They sell it for money. Auto-pilot is not very useful if you always have to keep your hands on the wheel. This is also a sign how immature Teslas technology is. Also, this model does not employ LiDAR. Even then, LiDAR is not all weather capable which is the pinnacle of self driving sensors.

https://www.wired.com/story/tesla-ntsb-autopilot-crash-death/

“Against a bright spring sky, the car’s sensors system failed to distinguish a large white 18-wheel truck and trailer crossing the highway, Tesla said. The car attempted to drive full speed under the trailer, “with the bottom of the trailer impacting the windshield of the Model S”, Tesla said in a blogpost.”

LiDAR can easily be fooled... one example, imagine someone with $60 who wants to wreak havoc on your self driving car
https://gizmodo.com/a-60-hack-can-fool-the-lidar-sensors-used-on-most-self-1729272292

So just saying, if you want foolproof sensors and cars coming off an assembly line before 2025, thoroughly tested to level 5, in all weather conditions, you’ll be waiting longer. Self-piloting cars will have to demonstrate robustness, rather than just rely on the absence of failures(MISRA).

[JagXKR]

Agreed. There are reasons why we don't have flying cars either. The simpletons of the world think that the movies are reality. Sorry but the snake oil salesmen selling this stupidity are incredibly myopic. Just a whole bunch of blah blah blah without the basic understanding of the real world complexities that have not been untangled. There is no way any of the lies that these snake oil salesmen have made will come true in any of the timelines stated.
But all the media jump on it as if it will happen tomorrow. Ridiculous.

[Dizzy1]

Why would they hit each other? The point of taking people out of the picture is to eliminate accidents.

Oh ok - they’ll never crash and be the science of perfect. Gotcha

[Dizzy1]

Just think of all the cool, extra maintenance all these systems require.

You guys think you’re getting hosed with an oil change? Wait until you go in for a recalibration

[Smurf]

If they can make the perfect self driving car I wish they would make the perfect computer and cell phone. If the systems controlling the self driving car are anything similar to computers and cell phones I want nothing to do with them. I believe it will be a long time before they develop something I want to trust with my life. what might be a minor glitch on a computer or cell phone could be instant death in a car.


EDIT TO ADD: When I see how far of the idea of driving many people can get when they fully understand they are in control of a killing machine I would hate to think where their mind will go if they think the car can even partially drive itself. Truthfully I'm not sure if it is the cars or the drivers that are far from ready for this leap. When you think it is probably both it is truly scary to think of the situations that could be created.

[Jflem1983]

If they can make the perfect self driving car I wish they would make the perfect computer and cell phone. If the systems controlling the self driving car are anything similar to computers and cell phones I want nothing to do with them. I believe it will be a long time before they develop something I want to trust with my life. what might be a minor glitch on a computer or cell phone could be instant death in a car.

My thoughts as well. My phone crashes about once a week.

[jimmy4321]

I don't like the dumb'n down of drivers of tasks that were second nature or at least doable 20yrs ago. I think if anything like this ever comes out and i have big doubts, It needs to be '100% hands off'.
As is now with Backing assist, lane change warnings, whatever you call that thing where the car overides braking to avoid rearenders etc. is making the driver LAZY and not fully engaged in driving.
Also i think they'll never work out the glitches in my life.

[Jlabute]

Just think of all the cool, extra maintenance all these systems require.

You guys think you’re getting hosed with an oil change? Wait until you go in for a recalibration

I've had my vehicle warn me during driving that "sensors are dirty", during a slush storm... and this could be typical poor driving conditions. In this instance, manual control would have to be given to the driver, or find a place to auto-park if it can, or all the cars on the road just have to stop moving. The vehicles have to work in areas other than sunny California.

Part of my regular oil change also involves software updates for the Navi, etc. The service station techs will need a new class of equipment and training for self-piloting cars.

I am sure it will all eventually come, but it won't be common place in 2025, it will still be somewhat more of an experiment. My understanding is it will be more of a commercial tool rather than something available to the general public to begin with.

"When Google originally started with autonomous vehicle research eight years ago, a LiDAR sensor made by Velodyne LiDAR, an industry leader, cost about $75,000. Although Velodyne declined to comment on the current price, John Krafcik, Waymo’s chief executive, mentioned recently that his company had reduced the cost of its LiDAR system by 90 percent. - $7,500." $7500 for LiDAR is still a hefty addition to a vehicle, and it is only one of the sensors required. A hefty cost will be associated with a proper self-piloting vehicle. Something better than LiDAR will be required, something yet to be invented that will allow all weather driving, or, use 4 LiDAR systems like Ford $$$$.

https://qz.com/637509/driverless-cars-have-a-new-way-to-navigate-in-rain-or-snow/

[CapitalB]

This article has a good unbiased review of the current state of the technology with lots of links to supporting articles for the curious.

Some of my favourite quotes from it;

For starters, we know that self-driving cars won't become commonplace overnight[...] But you won't likely find many in a dealer showroom for at least 10 years.

Another factor that will further delay the widespread use of autonomous cars is that many people hang on to their cars as long as possible. [...]

[...] Local governments are typically more nimble than state and federal agencies, which is one reason why you can find robotic taxis being tested right now on the streets of Pittsburgh, Phoenix and San Francisco.

One of the more interesting links in the article was this promising one about how they're trying to solve the snow / rain problem. Which doesn't solve the dirty sensors problem many people seem concerned about (though you could almost just use a wiper of some sort).

In regards to the slow transition of personal vehicles the expectation overall is that public transportation will be the earliest adopter of the technology. Despite the purchasing habits of most older vehicle owners automated public transportation in cities (where most people live) will entirely change how everyone else deals with transportation. Young people already seem less inclined to buy vehicles (for whatever reason, financial or other) and this fact combined with the pending wave of automated buses (testing in canada none the less) and automated ridesharing services (also testing in canada right now) are likely to change the face of things in Canada sooner rather than later.

Will there be an auto in every home in the next few years? No almost certainly not.
Will transportation services adopt them as soon as they possibly can? Absolutely they will, just like Walmart is looking to ditch its employees, dominoes is ditching its delivery drivers, and the washington post is using an AI to write articles.

All the links I've posted here are now things, testing, legislating, functioning technology that is as we speak making its way into the world around us. Even if you don't notice, even if you yourself won't upgrade, even if you think it'll take ten or more years I think you'll look back in five and be very surprised. I mean in the last ten years we've put super computers in every pocket, created functional prosthetics that can be used for ballroom dancing, have seen the overwhelming saturation of the internet (and ai and algorythms) into literally every facet of life, and a whole litany of things have been made faster smaller and more efficient.

Pinning a decade plus timeline on any technology that is already in the wild maturing is flat out delusional.

PS: $100 dollar Lidar for 2018

[36Drew]

As the article says, the sensors failed to distinguish, they were fooled... and the link makes Tesla partly responsible since their software lacked safeguards.

The link doesn't make Tesla at all responsible. Their position is that while this is horrible, you are still supposed to keep your feet on the pedals and hands on the wheel as their software is still experimental.

Also - it wasn't the tesla driver's "fault". The semi-truck driver shouldn't have turned. Just because both the tesla and the tesla driver failed to stop doesn't change this one little fact.

[36Drew]

the expectation overall is that public transportation will be the earliest adopter of the technology. Despite the purchasing habits of most older vehicle owners automated public transportation in cities (where most people live) will entirely change how everyone else deals with transportation.

I don't think so. At least, not quite yet, and not in a fully automated sense. Public transportation will still require a human presence for a while. Aside from monitoring fares and providing loading assistance to those who need it (wheelchairs...), it will be a while before we have the ability to decide if the people at a given bus stop - that might be shared by three different transit route - are waiting for a bus and this given bus in particular. A human presence will be required for a while - a button-pusher that is also busy training their AI replacement.

I think you'll find shipping/transport to be the first inroads. The entire process of loading up (or unloading) a shipping container is ripe for automation. The human driver of a shipping transport vehicle is just another fragile cog in the wheel that is ripe for replacement.

Another area that will be ripe for automation is carsharing. Right now, humans have to move car2go vehicles around to meet demand. There will come a time where you simply summon a car2go from your mobile phone and it will deliver itself to your location, as needed and when needed.